From b0651021e5c042e0893929e1b80cf367c6611708 Mon Sep 17 00:00:00 2001 From: Yanan Cao Date: Wed, 4 Mar 2026 21:25:59 -0800 Subject: [PATCH] [Kernel] [Helion] [11/N] Retune configs for silu_mul_fp8 (#36062) --- vllm/kernels/helion/configs/silu_mul_fp8.json | 35764 ++++++++-------- 1 file changed, 17886 insertions(+), 17878 deletions(-) diff --git a/vllm/kernels/helion/configs/silu_mul_fp8.json b/vllm/kernels/helion/configs/silu_mul_fp8.json index b8f091d66..bdef5e0fc 100644 --- a/vllm/kernels/helion/configs/silu_mul_fp8.json +++ b/vllm/kernels/helion/configs/silu_mul_fp8.json @@ -47,8 +47,8 @@ }, "intermediate_4096_numtokens_256": { "block_sizes": [ - 32, - 512 + 256, + 32 ], "loop_orders": [ [ @@ -60,7 +60,7 @@ true ], "l2_groupings": [ - 2 + 1 ], "range_unroll_factors": [ 0 @@ -80,11 +80,11 @@ "", "" ], - "num_warps": 2, - "num_stages": 2, + "num_warps": 8, + "num_stages": 1, "indexing": [ "pointer", - "tensor_descriptor", + "pointer", "pointer", "pointer" ], @@ -137,8 +137,8 @@ }, "intermediate_8192_numtokens_256": { "block_sizes": [ - 32, - 8 + 256, + 64 ], "loop_orders": [ [ @@ -167,23 +167,23 @@ ], "load_eviction_policies": [ "", - "first", + "", "" ], - "num_warps": 4, + "num_warps": 8, "num_stages": 1, "indexing": [ - "tensor_descriptor", "pointer", "pointer", - "tensor_descriptor" + "pointer", + "pointer" ], "pid_type": "flat" }, "intermediate_14336_numtokens_256": { "block_sizes": [ - 16, - 32 + 8, + 4096 ], "loop_orders": [ [ @@ -215,13 +215,13 @@ "", "" ], - "num_warps": 1, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", "pointer", "pointer", - "tensor_descriptor" + "pointer" ], "pid_type": "flat" }, @@ -272,8 +272,8 @@ }, "intermediate_7688_numtokens_256": { "block_sizes": [ - 8, - 16 + 32, + 512 ], "loop_orders": [ [ @@ -302,10 +302,10 @@ ], "load_eviction_policies": [ "", - "last", + "", "" ], - "num_warps": 1, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -363,187 +363,7 @@ "intermediate_2048_numtokens_1": { "block_sizes": [ 1, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_1": { - "block_sizes": [ - 1, - 1 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_1": { - "block_sizes": [ - 1, - 32 - ], - "loop_orders": [ - [ - 1, - 0 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_1": { - "block_sizes": [ - 1, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_1": { - "block_sizes": [ - 1, - 32 + 256 ], "loop_orders": [ [ @@ -570,96 +390,6 @@ "range_flattens": [ null ], - "load_eviction_policies": [ - "first", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_1": { - "block_sizes": [ - 1, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "first" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_2": { - "block_sizes": [ - 2, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], "load_eviction_policies": [ "", "", @@ -670,15 +400,15 @@ "indexing": [ "pointer", "pointer", - "tensor_descriptor", + "pointer", "pointer" ], "pid_type": "flat" }, - "intermediate_2880_numtokens_2": { + "intermediate_2880_numtokens_1": { "block_sizes": [ 1, - 4 + 128 ], "loop_orders": [ [ @@ -687,7 +417,7 @@ ] ], "flatten_loops": [ - false + true ], "l2_groupings": [ 1 @@ -706,11 +436,281 @@ null ], "load_eviction_policies": [ - "first", + "", "", "" ], - "num_warps": 32, + "num_warps": 4, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_1": { + "block_sizes": [ + 1, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 2, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_1": { + "block_sizes": [ + 1, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 2, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_1": { + "block_sizes": [ + 1, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 2, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_1": { + "block_sizes": [ + 1, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 2, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_2": { + "block_sizes": [ + 2, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 2, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_2": { + "block_sizes": [ + 2, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 4, "num_stages": 1, "indexing": [ "pointer", @@ -723,7 +723,7 @@ "intermediate_4096_numtokens_2": { "block_sizes": [ 2, - 32 + 128 ], "loop_orders": [ [ @@ -755,8 +755,8 @@ "", "" ], - "num_warps": 8, - "num_stages": 2, + "num_warps": 2, + "num_stages": 1, "indexing": [ "pointer", "pointer", @@ -767,8 +767,8 @@ }, "intermediate_8192_numtokens_2": { "block_sizes": [ - 1, - 32 + 2, + 128 ], "loop_orders": [ [ @@ -777,7 +777,7 @@ ] ], "flatten_loops": [ - false + true ], "l2_groupings": [ 1 @@ -796,14 +796,14 @@ null ], "load_eviction_policies": [ - "first", + "", "", "" ], - "num_warps": 8, - "num_stages": 2, + "num_warps": 2, + "num_stages": 1, "indexing": [ - "tensor_descriptor", + "pointer", "pointer", "pointer", "pointer" @@ -813,200 +813,20 @@ "intermediate_11008_numtokens_2": { "block_sizes": [ 1, - 256 + 16384 ], "loop_orders": [ [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "first", - "" - ], - "num_warps": 4, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_2": { - "block_sizes": [ - 1, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_4": { - "block_sizes": [ - 1, - 256 - ], - "loop_orders": [ - [ - 0, - 1 + 1, + 0 ] ], "flatten_loops": [ true ], "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_4": { - "block_sizes": [ - 1, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_4": { - "block_sizes": [ - 4, 16 ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], "range_unroll_factors": [ 0 ], @@ -1022,10 +842,190 @@ ], "load_eviction_policies": [ "last", + "first", + "last" + ], + "num_warps": 32, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "tensor_descriptor" + ], + "pid_type": "xyz" + }, + "intermediate_14336_numtokens_2": { + "block_sizes": [ + 2, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", "", "" ], - "num_warps": 4, + "num_warps": 2, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_4": { + "block_sizes": [ + 4, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 2, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_4": { + "block_sizes": [ + 4, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "first" + ], + "num_warps": 8, + "num_stages": 7, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "xyz" + }, + "intermediate_4096_numtokens_4": { + "block_sizes": [ + 4, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 2, "num_stages": 1, "indexing": [ "pointer", @@ -1038,52 +1038,7 @@ "intermediate_8192_numtokens_4": { "block_sizes": [ 1, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_4": { - "block_sizes": [ - 1, - 32 + 2048 ], "loop_orders": [ [ @@ -1112,23 +1067,23 @@ ], "load_eviction_policies": [ "", - "first", + "", "" ], - "num_warps": 16, - "num_stages": 1, + "num_warps": 8, + "num_stages": 8, "indexing": [ - "pointer", "pointer", "tensor_descriptor", + "pointer", "pointer" ], "pid_type": "flat" }, - "intermediate_14336_numtokens_4": { + "intermediate_11008_numtokens_4": { "block_sizes": [ 4, - 16 + 2048 ], "loop_orders": [ [ @@ -1139,6 +1094,96 @@ "flatten_loops": [ true ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 8, + "num_stages": 6, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "xyz" + }, + "intermediate_14336_numtokens_4": { + "block_sizes": [ + 4, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 2, + "num_stages": 2, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "xyz" + }, + "intermediate_2048_numtokens_8": { + "block_sizes": [ + 8, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], "l2_groupings": [ 1 ], @@ -1160,7 +1205,7 @@ "", "" ], - "num_warps": 16, + "num_warps": 2, "num_stages": 1, "indexing": [ "pointer", @@ -1170,22 +1215,22 @@ ], "pid_type": "flat" }, - "intermediate_2048_numtokens_8": { + "intermediate_2880_numtokens_8": { "block_sizes": [ - 8, - 256 + 4, + 2048 ], "loop_orders": [ [ - 1, - 0 + 0, + 1 ] ], "flatten_loops": [ true ], "l2_groupings": [ - 1 + 4 ], "range_unroll_factors": [ 0 @@ -1202,20 +1247,20 @@ ], "load_eviction_policies": [ "last", - "", + "last", "" ], - "num_warps": 32, - "num_stages": 1, + "num_warps": 16, + "num_stages": 5, "indexing": [ + "pointer", + "tensor_descriptor", "tensor_descriptor", - "pointer", - "pointer", "pointer" ], - "pid_type": "flat" + "pid_type": "xyz" }, - "intermediate_2880_numtokens_8": { + "intermediate_4096_numtokens_8": { "block_sizes": [ 8, 32 @@ -1250,52 +1295,7 @@ "", "" ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_8": { - "block_sizes": [ - 2, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, + "num_warps": 2, "num_stages": 1, "indexing": [ "pointer", @@ -1307,8 +1307,8 @@ }, "intermediate_8192_numtokens_8": { "block_sizes": [ - 4, - 64 + 2, + 1024 ], "loop_orders": [ [ @@ -1317,10 +1317,10 @@ ] ], "flatten_loops": [ - false + true ], "l2_groupings": [ - 1 + 16 ], "range_unroll_factors": [ 0 @@ -1336,36 +1336,36 @@ null ], "load_eviction_policies": [ - "", - "", + "first", + "last", "" ], "num_warps": 1, - "num_stages": 1, + "num_stages": 8, "indexing": [ + "tensor_descriptor", "pointer", - "pointer", - "pointer", + "tensor_descriptor", "pointer" ], "pid_type": "flat" }, "intermediate_11008_numtokens_8": { "block_sizes": [ - 8, - 128 + 1, + 1024 ], "loop_orders": [ [ - 0, - 1 + 1, + 0 ] ], "flatten_loops": [ - false + true ], "l2_groupings": [ - 1 + 4 ], "range_unroll_factors": [ 0 @@ -1383,22 +1383,22 @@ "load_eviction_policies": [ "last", "", - "" + "first" ], - "num_warps": 4, - "num_stages": 2, + "num_warps": 2, + "num_stages": 5, "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" ], "pid_type": "flat" }, "intermediate_14336_numtokens_8": { "block_sizes": [ 8, - 32 + 128 ], "loop_orders": [ [ @@ -1407,7 +1407,7 @@ ] ], "flatten_loops": [ - false + true ], "l2_groupings": [ 1 @@ -1427,7 +1427,7 @@ ], "load_eviction_policies": [ "", - "first", + "", "" ], "num_warps": 4, @@ -1442,8 +1442,8 @@ }, "intermediate_2048_numtokens_16": { "block_sizes": [ - 16, - 64 + 8, + 512 ], "loop_orders": [ [ @@ -1452,10 +1452,55 @@ ] ], "flatten_loops": [ - false + true ], "l2_groupings": [ - 2 + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "first" + ], + "num_warps": 16, + "num_stages": 2, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "xyz" + }, + "intermediate_2880_numtokens_16": { + "block_sizes": [ + 2, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 ], "range_unroll_factors": [ 0 @@ -1476,51 +1521,6 @@ "" ], "num_warps": 4, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_16": { - "block_sizes": [ - 16, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -1533,7 +1533,97 @@ "intermediate_4096_numtokens_16": { "block_sizes": [ 16, - 32 + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "", + "last" + ], + "num_warps": 32, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_16": { + "block_sizes": [ + 16, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 4, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_16": { + "block_sizes": [ + 1, + 2048 ], "loop_orders": [ [ @@ -1566,7 +1656,7 @@ "" ], "num_warps": 8, - "num_stages": 2, + "num_stages": 1, "indexing": [ "pointer", "pointer", @@ -1575,22 +1665,22 @@ ], "pid_type": "flat" }, - "intermediate_8192_numtokens_16": { + "intermediate_14336_numtokens_16": { "block_sizes": [ - 4, - 32 + 2, + 256 ], "loop_orders": [ [ - 0, - 1 + 1, + 0 ] ], "flatten_loops": [ - false + true ], "l2_groupings": [ - 1 + 64 ], "range_unroll_factors": [ 0 @@ -1606,24 +1696,24 @@ null ], "load_eviction_policies": [ - "", - "", - "" + "last", + "last", + "last" ], - "num_warps": 1, - "num_stages": 1, + "num_warps": 16, + "num_stages": 7, "indexing": [ - "pointer", "tensor_descriptor", "pointer", - "pointer" + "tensor_descriptor", + "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_11008_numtokens_16": { + "intermediate_2048_numtokens_24": { "block_sizes": [ - 8, - 32 + 1, + 1024 ], "loop_orders": [ [ @@ -1635,7 +1725,7 @@ true ], "l2_groupings": [ - 1 + 64 ], "range_unroll_factors": [ 0 @@ -1655,115 +1745,70 @@ "", "first" ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_16": { - "block_sizes": [ - 16, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, + "num_warps": 4, + "num_stages": 8, "indexing": [ "tensor_descriptor", "pointer", "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_24": { - "block_sizes": [ - 16, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "last" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" + "tensor_descriptor" ], "pid_type": "flat" }, "intermediate_2880_numtokens_24": { "block_sizes": [ - 32, + 4, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "first" + ], + "num_warps": 16, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_24": { + "block_sizes": [ + 16, 64 ], "loop_orders": [ [ - 1, - 0 + 0, + 1 ] ], "flatten_loops": [ @@ -1793,17 +1838,17 @@ "num_warps": 4, "num_stages": 1, "indexing": [ - "tensor_descriptor", "pointer", "pointer", - "tensor_descriptor" + "pointer", + "pointer" ], "pid_type": "flat" }, - "intermediate_4096_numtokens_24": { + "intermediate_8192_numtokens_24": { "block_sizes": [ - 32, - 32 + 1, + 2048 ], "loop_orders": [ [ @@ -1815,7 +1860,7 @@ true ], "l2_groupings": [ - 2 + 32 ], "range_unroll_factors": [ 0 @@ -1832,11 +1877,11 @@ ], "load_eviction_policies": [ "", - "first", - "" + "last", + "last" ], - "num_warps": 1, - "num_stages": 1, + "num_warps": 32, + "num_stages": 5, "indexing": [ "pointer", "pointer", @@ -1845,22 +1890,22 @@ ], "pid_type": "flat" }, - "intermediate_8192_numtokens_24": { + "intermediate_11008_numtokens_24": { "block_sizes": [ - 16, - 32 + 1, + 1024 ], "loop_orders": [ [ - 0, - 1 + 1, + 0 ] ], "flatten_loops": [ false ], "l2_groupings": [ - 1 + 64 ], "range_unroll_factors": [ 0 @@ -1876,57 +1921,12 @@ null ], "load_eviction_policies": [ - "", + "last", "", "" ], "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_24": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, + "num_stages": 2, "indexing": [ "pointer", "pointer", @@ -1938,19 +1938,19 @@ "intermediate_14336_numtokens_24": { "block_sizes": [ 8, - 32 + 512 ], "loop_orders": [ [ - 1, - 0 + 0, + 1 ] ], "flatten_loops": [ true ], "l2_groupings": [ - 1 + 4 ], "range_unroll_factors": [ 0 @@ -1966,111 +1966,21 @@ null ], "load_eviction_policies": [ - "", - "", - "last" + "last", + "first", + "first" ], - "num_warps": 2, - "num_stages": 1, + "num_warps": 32, + "num_stages": 3, "indexing": [ "pointer", + "tensor_descriptor", "pointer", - "pointer", - "pointer" + "tensor_descriptor" ], "pid_type": "flat" }, "intermediate_2048_numtokens_32": { - "block_sizes": [ - 32, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_32": { - "block_sizes": [ - 32, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_32": { "block_sizes": [ 32, 16 @@ -2105,8 +2015,8 @@ "", "" ], - "num_warps": 1, - "num_stages": 2, + "num_warps": 4, + "num_stages": 1, "indexing": [ "pointer", "pointer", @@ -2115,10 +2025,10 @@ ], "pid_type": "flat" }, - "intermediate_8192_numtokens_32": { + "intermediate_2880_numtokens_32": { "block_sizes": [ - 32, - 128 + 4, + 256 ], "loop_orders": [ [ @@ -2148,10 +2058,10 @@ "load_eviction_policies": [ "", "", - "last" + "" ], - "num_warps": 2, - "num_stages": 2, + "num_warps": 4, + "num_stages": 1, "indexing": [ "pointer", "pointer", @@ -2160,22 +2070,67 @@ ], "pid_type": "flat" }, - "intermediate_11008_numtokens_32": { + "intermediate_4096_numtokens_32": { "block_sizes": [ - 16, - 8 + 4, + 4096 ], "loop_orders": [ [ - 1, - 0 + 0, + 1 ] ], "flatten_loops": [ true ], "l2_groupings": [ - 1 + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "" + ], + "num_warps": 32, + "num_stages": 5, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_32": { + "block_sizes": [ + 4, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 ], "range_unroll_factors": [ 0 @@ -2192,23 +2147,23 @@ ], "load_eviction_policies": [ "", - "", + "last", "" ], - "num_warps": 16, - "num_stages": 1, + "num_warps": 2, + "num_stages": 3, "indexing": [ - "pointer", - "pointer", + "tensor_descriptor", + "tensor_descriptor", "pointer", "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_14336_numtokens_32": { + "intermediate_11008_numtokens_32": { "block_sizes": [ - 32, - 8 + 2, + 2048 ], "loop_orders": [ [ @@ -2241,7 +2196,7 @@ "" ], "num_warps": 8, - "num_stages": 2, + "num_stages": 1, "indexing": [ "pointer", "pointer", @@ -2250,9 +2205,54 @@ ], "pid_type": "flat" }, + "intermediate_14336_numtokens_32": { + "block_sizes": [ + 1, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "" + ], + "num_warps": 4, + "num_stages": 3, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, "intermediate_2048_numtokens_40": { "block_sizes": [ - 64, + 32, 32 ], "loop_orders": [ @@ -2265,7 +2265,7 @@ true ], "l2_groupings": [ - 2 + 1 ], "range_unroll_factors": [ 0 @@ -2285,7 +2285,7 @@ "", "" ], - "num_warps": 2, + "num_warps": 4, "num_stages": 1, "indexing": [ "pointer", @@ -2297,20 +2297,20 @@ }, "intermediate_2880_numtokens_40": { "block_sizes": [ - 64, - 32 + 1, + 4096 ], "loop_orders": [ [ - 0, - 1 + 1, + 0 ] ], "flatten_loops": [ - true + false ], "l2_groupings": [ - 1 + 32 ], "range_unroll_factors": [ 0 @@ -2327,13 +2327,13 @@ ], "load_eviction_policies": [ "", - "", + "last", "last" ], "num_warps": 8, - "num_stages": 2, + "num_stages": 4, "indexing": [ - "pointer", + "tensor_descriptor", "pointer", "pointer", "pointer" @@ -2343,12 +2343,12 @@ "intermediate_4096_numtokens_40": { "block_sizes": [ 32, - 32 + 64 ], "loop_orders": [ [ - 1, - 0 + 0, + 1 ] ], "flatten_loops": [ @@ -2378,7 +2378,7 @@ "num_warps": 8, "num_stages": 1, "indexing": [ - "tensor_descriptor", + "pointer", "pointer", "pointer", "pointer" @@ -2387,8 +2387,8 @@ }, "intermediate_8192_numtokens_40": { "block_sizes": [ - 64, - 128 + 2, + 256 ], "loop_orders": [ [ @@ -2400,7 +2400,7 @@ true ], "l2_groupings": [ - 2 + 16 ], "range_unroll_factors": [ 0 @@ -2417,23 +2417,23 @@ ], "load_eviction_policies": [ "", - "", - "" + "last", + "last" ], - "num_warps": 4, - "num_stages": 1, + "num_warps": 2, + "num_stages": 2, "indexing": [ "pointer", "pointer", "pointer", - "pointer" + "tensor_descriptor" ], "pid_type": "flat" }, "intermediate_11008_numtokens_40": { "block_sizes": [ - 64, - 32 + 16, + 256 ], "loop_orders": [ [ @@ -2465,7 +2465,7 @@ "", "" ], - "num_warps": 4, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -2476,6 +2476,53 @@ "pid_type": "flat" }, "intermediate_14336_numtokens_40": { + "block_sizes": [ + 4, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 1 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "last" + ], + "num_warps": 16, + "num_stages": 5, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "persistent_interleaved", + "num_sm_multiplier": 32, + "maxnreg": 32 + }, + "intermediate_2048_numtokens_48": { "block_sizes": [ 32, 32 @@ -2510,7 +2557,7 @@ "", "" ], - "num_warps": 2, + "num_warps": 4, "num_stages": 1, "indexing": [ "pointer", @@ -2520,55 +2567,10 @@ ], "pid_type": "flat" }, - "intermediate_2048_numtokens_48": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "first", - "" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, "intermediate_2880_numtokens_48": { "block_sizes": [ - 8, - 8 + 16, + 64 ], "loop_orders": [ [ @@ -2600,12 +2602,12 @@ "", "" ], - "num_warps": 1, + "num_warps": 4, "num_stages": 1, "indexing": [ "pointer", "pointer", - "tensor_descriptor", + "pointer", "pointer" ], "pid_type": "flat" @@ -2645,52 +2647,7 @@ "", "" ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_48": { - "block_sizes": [ - 64, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -2700,10 +2657,10 @@ ], "pid_type": "flat" }, - "intermediate_11008_numtokens_48": { + "intermediate_8192_numtokens_48": { "block_sizes": [ - 16, - 256 + 1, + 4096 ], "loop_orders": [ [ @@ -2731,24 +2688,24 @@ null ], "load_eviction_policies": [ + "first", "", - "", - "" + "last" ], - "num_warps": 32, - "num_stages": 1, + "num_warps": 4, + "num_stages": 2, "indexing": [ "pointer", - "pointer", - "pointer", - "tensor_descriptor" + "tensor_descriptor", + "tensor_descriptor", + "pointer" ], "pid_type": "flat" }, - "intermediate_14336_numtokens_48": { + "intermediate_11008_numtokens_48": { "block_sizes": [ - 64, - 4 + 16, + 256 ], "loop_orders": [ [ @@ -2780,10 +2737,55 @@ "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ - "tensor_descriptor", + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_48": { + "block_sizes": [ + 32, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", "pointer", "pointer", "pointer" @@ -2792,52 +2794,7 @@ }, "intermediate_2048_numtokens_56": { "block_sizes": [ - 2, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_56": { - "block_sizes": [ - 8, + 32, 32 ], "loop_orders": [ @@ -2880,10 +2837,10 @@ ], "pid_type": "flat" }, - "intermediate_4096_numtokens_56": { + "intermediate_2880_numtokens_56": { "block_sizes": [ - 32, - 4 + 1, + 2048 ], "loop_orders": [ [ @@ -2912,15 +2869,60 @@ ], "load_eviction_policies": [ "", - "first", + "", "" ], - "num_warps": 4, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", "pointer", - "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_56": { + "block_sizes": [ + 32, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", "pointer" ], "pid_type": "flat" @@ -2960,7 +2962,7 @@ "", "" ], - "num_warps": 1, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -2972,8 +2974,8 @@ }, "intermediate_11008_numtokens_56": { "block_sizes": [ - 32, - 8 + 1, + 8192 ], "loop_orders": [ [ @@ -3002,10 +3004,10 @@ ], "load_eviction_policies": [ "", - "first", - "last" + "", + "" ], - "num_warps": 1, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -3016,144 +3018,9 @@ "pid_type": "flat" }, "intermediate_14336_numtokens_56": { - "block_sizes": [ - 64, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_64": { - "block_sizes": [ - 16, - 128 - ], - "loop_orders": [ - [ - 1, - 0 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_64": { - "block_sizes": [ - 4, - 64 - ], - "loop_orders": [ - [ - 1, - 0 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_64": { "block_sizes": [ 2, - 16 + 4096 ], "loop_orders": [ [ @@ -3162,10 +3029,10 @@ ] ], "flatten_loops": [ - true + false ], "l2_groupings": [ - 1 + 32 ], "range_unroll_factors": [ 0 @@ -3182,608 +3049,23 @@ ], "load_eviction_policies": [ "first", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_64": { - "block_sizes": [ - 8, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_64": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", + "first", "" ], "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_64": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_72": { - "block_sizes": [ - 4, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "last" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_72": { - "block_sizes": [ - 64, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "first" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_72": { - "block_sizes": [ - 64, - 16 - ], - "loop_orders": [ - [ - 1, - 0 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "last", - "", - "" - ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_72": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_72": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_72": { - "block_sizes": [ - 128, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_80": { - "block_sizes": [ - 32, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_80": { - "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_80": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_80": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 4 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, + "num_stages": 4, "indexing": [ "tensor_descriptor", "pointer", - "pointer", - "pointer" + "tensor_descriptor", + "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_11008_numtokens_80": { + "intermediate_2048_numtokens_64": { "block_sizes": [ 64, - 8 + 32 ], "loop_orders": [ [ @@ -3825,9 +3107,189 @@ ], "pid_type": "flat" }, - "intermediate_14336_numtokens_80": { + "intermediate_2880_numtokens_64": { "block_sizes": [ - 32, + 1, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_64": { + "block_sizes": [ + 64, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_64": { + "block_sizes": [ + 64, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_64": { + "block_sizes": [ + 1, + 8192 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_64": { + "block_sizes": [ + 16, 512 ], "loop_orders": [ @@ -3840,7 +3302,7 @@ true ], "l2_groupings": [ - 2 + 16 ], "range_unroll_factors": [ 0 @@ -3860,156 +3322,21 @@ "first", "" ], - "num_warps": 16, + "num_warps": 4, "num_stages": 1, "indexing": [ "pointer", "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_88": { - "block_sizes": [ - 32, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_88": { - "block_sizes": [ - 16, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ "tensor_descriptor", - "pointer", - "pointer", "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_4096_numtokens_88": { + "intermediate_2048_numtokens_72": { "block_sizes": [ 64, 32 ], - "loop_orders": [ - [ - 1, - 0 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "tensor_descriptor", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_88": { - "block_sizes": [ - 128, - 64 - ], "loop_orders": [ [ 0, @@ -4035,101 +3362,11 @@ "range_flattens": [ null ], - "load_eviction_policies": [ - "first", - "first", - "last" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_88": { - "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 4 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], "load_eviction_policies": [ "", "", "" ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_88": { - "block_sizes": [ - 16, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "first", - "last", - "" - ], "num_warps": 8, "num_stages": 1, "indexing": [ @@ -4140,248 +3377,68 @@ ], "pid_type": "flat" }, - "intermediate_2048_numtokens_96": { + "intermediate_2880_numtokens_72": { "block_sizes": [ - 128, + 32, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_72": { + "block_sizes": [ + 1, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ 4 ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_96": { - "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_96": { - "block_sizes": [ - 16, - 256 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 3, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_96": { - "block_sizes": [ - 64, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "first", - "", - "last" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_96": { - "block_sizes": [ - 64, - 256 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "first", - "", - "" - ], - "num_warps": 2, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_96": { - "block_sizes": [ - 32, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], "range_unroll_factors": [ 0 ], @@ -4400,20 +3457,20 @@ "", "first" ], - "num_warps": 1, - "num_stages": 1, + "num_warps": 16, + "num_stages": 2, "indexing": [ - "pointer", - "pointer", "tensor_descriptor", - "pointer" + "pointer", + "pointer", + "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_2048_numtokens_104": { + "intermediate_8192_numtokens_72": { "block_sizes": [ - 32, - 8 + 64, + 128 ], "loop_orders": [ [ @@ -4445,17 +3502,197 @@ "", "" ], - "num_warps": 16, - "num_stages": 2, + "num_warps": 8, + "num_stages": 1, "indexing": [ - "tensor_descriptor", + "pointer", "pointer", "pointer", "pointer" ], "pid_type": "flat" }, - "intermediate_2880_numtokens_104": { + "intermediate_11008_numtokens_72": { + "block_sizes": [ + 4, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 32, + "num_stages": 5, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_72": { + "block_sizes": [ + 32, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "", + "last" + ], + "num_warps": 8, + "num_stages": 3, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_80": { + "block_sizes": [ + 64, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_80": { + "block_sizes": [ + 32, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_80": { "block_sizes": [ 64, 64 @@ -4490,7 +3727,772 @@ "", "" ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_80": { + "block_sizes": [ + 4, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "last" + ], + "num_warps": 16, + "num_stages": 2, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_80": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "" + ], + "num_warps": 16, + "num_stages": 2, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_80": { + "block_sizes": [ + 2, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "" + ], + "num_warps": 4, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_88": { + "block_sizes": [ + 64, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_88": { + "block_sizes": [ + 8, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_88": { + "block_sizes": [ + 64, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_88": { + "block_sizes": [ + 64, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_88": { + "block_sizes": [ + 16, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "first", + "" + ], "num_warps": 32, + "num_stages": 2, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_88": { + "block_sizes": [ + 4, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 32, + "num_stages": 5, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_96": { + "block_sizes": [ + 64, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_96": { + "block_sizes": [ + 32, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_96": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 4, + "num_stages": 6, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_96": { + "block_sizes": [ + 64, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_96": { + "block_sizes": [ + 1, + 2048 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "" + ], + "num_warps": 32, + "num_stages": 2, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_96": { + "block_sizes": [ + 4, + 4096 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "first", + "" + ], + "num_warps": 4, + "num_stages": 1, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_104": { + "block_sizes": [ + 64, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_104": { + "block_sizes": [ + 8, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -4502,8 +4504,8 @@ }, "intermediate_4096_numtokens_104": { "block_sizes": [ - 32, - 32 + 64, + 64 ], "loop_orders": [ [ @@ -4512,7 +4514,7 @@ ] ], "flatten_loops": [ - false + true ], "l2_groupings": [ 1 @@ -4546,9 +4548,99 @@ "pid_type": "flat" }, "intermediate_8192_numtokens_104": { + "block_sizes": [ + 64, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_104": { + "block_sizes": [ + 2, + 8192 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "first" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_104": { "block_sizes": [ 8, - 8 + 512 ], "loop_orders": [ [ @@ -4575,6 +4667,51 @@ "range_flattens": [ null ], + "load_eviction_policies": [ + "last", + "last", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_112": { + "block_sizes": [ + 64, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], "load_eviction_policies": [ "", "", @@ -4586,14 +4723,14 @@ "pointer", "pointer", "pointer", - "tensor_descriptor" + "pointer" ], "pid_type": "flat" }, - "intermediate_11008_numtokens_104": { + "intermediate_2880_numtokens_112": { "block_sizes": [ - 128, - 16 + 2, + 2048 ], "loop_orders": [ [ @@ -4625,20 +4762,20 @@ "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", - "tensor_descriptor", + "pointer", "pointer", "pointer" ], "pid_type": "flat" }, - "intermediate_14336_numtokens_104": { + "intermediate_4096_numtokens_112": { "block_sizes": [ - 32, - 16 + 64, + 64 ], "loop_orders": [ [ @@ -4670,20 +4807,20 @@ "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", "pointer", "pointer", - "tensor_descriptor" + "pointer" ], "pid_type": "flat" }, - "intermediate_2048_numtokens_112": { + "intermediate_8192_numtokens_112": { "block_sizes": [ - 32, - 1024 + 4, + 512 ], "loop_orders": [ [ @@ -4695,7 +4832,7 @@ true ], "l2_groupings": [ - 1 + 16 ], "range_unroll_factors": [ 0 @@ -4712,203 +4849,23 @@ ], "load_eviction_policies": [ "first", - "", - "" - ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_112": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_112": { - "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ "last", - "", - "" + "last" ], "num_warps": 16, - "num_stages": 1, + "num_stages": 3, "indexing": [ "pointer", "pointer", - "pointer", + "tensor_descriptor", "pointer" ], "pid_type": "flat" }, - "intermediate_8192_numtokens_112": { - "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, "intermediate_11008_numtokens_112": { "block_sizes": [ - 16, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_112": { - "block_sizes": [ - 32, - 8 + 4, + 1024 ], "loop_orders": [ [ @@ -4937,23 +4894,23 @@ ], "load_eviction_policies": [ "", - "", - "" + "last", + "last" ], - "num_warps": 1, - "num_stages": 2, + "num_warps": 32, + "num_stages": 3, "indexing": [ "pointer", "pointer", - "pointer", - "pointer" + "tensor_descriptor", + "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_2048_numtokens_120": { + "intermediate_14336_numtokens_112": { "block_sizes": [ - 32, - 64 + 64, + 256 ], "loop_orders": [ [ @@ -4985,7 +4942,7 @@ "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -4995,19 +4952,64 @@ ], "pid_type": "flat" }, - "intermediate_2880_numtokens_120": { + "intermediate_2048_numtokens_120": { "block_sizes": [ - 32, - 16 + 8, + 256 ], "loop_orders": [ [ - 1, - 0 + 0, + 1 ] ], "flatten_loops": [ - false + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "" + ], + "num_warps": 4, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_120": { + "block_sizes": [ + 2, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true ], "l2_groupings": [ 1 @@ -5030,7 +5032,7 @@ "", "" ], - "num_warps": 4, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -5042,8 +5044,8 @@ }, "intermediate_4096_numtokens_120": { "block_sizes": [ - 32, - 16 + 64, + 64 ], "loop_orders": [ [ @@ -5055,7 +5057,7 @@ true ], "l2_groupings": [ - 2 + 1 ], "range_unroll_factors": [ 0 @@ -5075,7 +5077,7 @@ "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -5088,97 +5090,7 @@ "intermediate_8192_numtokens_120": { "block_sizes": [ 64, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_120": { - "block_sizes": [ - 64, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_120": { - "block_sizes": [ - 128, - 32 + 128 ], "loop_orders": [ [ @@ -5220,10 +5132,100 @@ ], "pid_type": "flat" }, - "intermediate_2048_numtokens_128": { + "intermediate_11008_numtokens_120": { + "block_sizes": [ + 1, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 1, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_120": { "block_sizes": [ 32, - 64 + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "first", + "last" + ], + "num_warps": 16, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_128": { + "block_sizes": [ + 128, + 16 ], "loop_orders": [ [ @@ -5255,8 +5257,8 @@ "", "" ], - "num_warps": 2, - "num_stages": 2, + "num_warps": 8, + "num_stages": 1, "indexing": [ "pointer", "pointer", @@ -5267,8 +5269,8 @@ }, "intermediate_2880_numtokens_128": { "block_sizes": [ - 128, - 64 + 2, + 2048 ], "loop_orders": [ [ @@ -5297,10 +5299,10 @@ ], "load_eviction_policies": [ "", - "last", + "", "" ], - "num_warps": 4, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -5340,101 +5342,11 @@ "range_flattens": [ null ], - "load_eviction_policies": [ - "", - "", - "last" - ], - "num_warps": 32, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_128": { - "block_sizes": [ - 32, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], "load_eviction_policies": [ "", "", "" ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_128": { - "block_sizes": [ - 128, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "last" - ], "num_warps": 8, "num_stages": 1, "indexing": [ @@ -5445,99 +5357,9 @@ ], "pid_type": "flat" }, - "intermediate_14336_numtokens_128": { - "block_sizes": [ - 16, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 3, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_136": { + "intermediate_8192_numtokens_128": { "block_sizes": [ 128, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 3, - "indexing": [ - "tensor_descriptor", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_136": { - "block_sizes": [ - 8, 64 ], "loop_orders": [ @@ -5567,7 +5389,187 @@ ], "load_eviction_policies": [ "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_128": { + "block_sizes": [ + 2, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ "first", + "", + "last" + ], + "num_warps": 2, + "num_stages": 1, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_128": { + "block_sizes": [ + 4, + 4096 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_136": { + "block_sizes": [ + 128, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_136": { + "block_sizes": [ + 64, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", "" ], "num_warps": 8, @@ -5582,8 +5584,8 @@ }, "intermediate_4096_numtokens_136": { "block_sizes": [ - 32, - 16 + 128, + 64 ], "loop_orders": [ [ @@ -5615,7 +5617,7 @@ "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -5627,8 +5629,8 @@ }, "intermediate_8192_numtokens_136": { "block_sizes": [ - 32, - 128 + 2, + 512 ], "loop_orders": [ [ @@ -5640,7 +5642,7 @@ true ], "l2_groupings": [ - 1 + 32 ], "range_unroll_factors": [ 0 @@ -5656,204 +5658,24 @@ null ], "load_eviction_policies": [ - "", + "first", "", "" ], - "num_warps": 4, - "num_stages": 1, + "num_warps": 32, + "num_stages": 7, "indexing": [ + "tensor_descriptor", "pointer", "pointer", - "pointer", - "pointer" + "tensor_descriptor" ], "pid_type": "flat" }, "intermediate_11008_numtokens_136": { "block_sizes": [ - 16, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 3, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_136": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_144": { - "block_sizes": [ - 8, - 16 - ], - "loop_orders": [ - [ - 1, - 0 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_144": { - "block_sizes": [ - 256, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_144": { - "block_sizes": [ - 128, - 32 + 4, + 8192 ], "loop_orders": [ [ @@ -5865,97 +5687,7 @@ false ], "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_144": { - "block_sizes": [ - 128, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_144": { - "block_sizes": [ - 32, - 4 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 + 8 ], "range_unroll_factors": [ 0 @@ -5973,10 +5705,10 @@ "load_eviction_policies": [ "", "last", - "" + "last" ], "num_warps": 8, - "num_stages": 1, + "num_stages": 4, "indexing": [ "pointer", "pointer", @@ -5985,11 +5717,101 @@ ], "pid_type": "flat" }, - "intermediate_14336_numtokens_144": { + "intermediate_14336_numtokens_136": { "block_sizes": [ - 32, + 4, + 16384 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ 8 ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "last" + ], + "num_warps": 32, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_144": { + "block_sizes": [ + 1, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "first" + ], + "num_warps": 16, + "num_stages": 7, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_144": { + "block_sizes": [ + 64, + 64 + ], "loop_orders": [ [ 0, @@ -6015,25 +5837,790 @@ "range_flattens": [ null ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_144": { + "block_sizes": [ + 128, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_144": { + "block_sizes": [ + 1, + 2048 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "first" + ], + "num_warps": 1, + "num_stages": 4, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_144": { + "block_sizes": [ + 256, + 16 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "first" + ], + "num_warps": 16, + "num_stages": 2, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_144": { + "block_sizes": [ + 64, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "", + "last" + ], + "num_warps": 16, + "num_stages": 8, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_152": { + "block_sizes": [ + 4, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "first" + ], + "num_warps": 8, + "num_stages": 7, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_152": { + "block_sizes": [ + 64, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_152": { + "block_sizes": [ + 128, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_152": { + "block_sizes": [ + 64, + 16 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "first" + ], + "num_warps": 1, + "num_stages": 2, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_152": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "first" + ], + "num_warps": 4, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_152": { + "block_sizes": [ + 2, + 16384 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "" + ], + "num_warps": 16, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_160": { + "block_sizes": [ + 4, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "last" + ], + "num_warps": 1, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_160": { + "block_sizes": [ + 64, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_160": { + "block_sizes": [ + 128, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_160": { + "block_sizes": [ + 64, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 32, + "num_stages": 4, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_160": { + "block_sizes": [ + 128, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "first", + "last" + ], + "num_warps": 32, + "num_stages": 8, + "indexing": [ + "pointer", + "pointer", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_160": { + "block_sizes": [ + 1, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], "load_eviction_policies": [ "", "", "first" ], - "num_warps": 1, - "num_stages": 2, + "num_warps": 8, + "num_stages": 8, "indexing": [ - "pointer", "tensor_descriptor", "pointer", - "pointer" + "tensor_descriptor", + "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_2048_numtokens_152": { + "intermediate_2048_numtokens_168": { "block_sizes": [ - 32, - 8 + 4, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "", + "last" + ], + "num_warps": 32, + "num_stages": 8, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_168": { + "block_sizes": [ + 8, + 512 ], "loop_orders": [ [ @@ -6065,7 +6652,7 @@ "", "" ], - "num_warps": 16, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -6075,9 +6662,9 @@ ], "pid_type": "flat" }, - "intermediate_2880_numtokens_152": { + "intermediate_4096_numtokens_168": { "block_sizes": [ - 16, + 128, 64 ], "loop_orders": [ @@ -6110,141 +6697,6 @@ "", "" ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_152": { - "block_sizes": [ - 64, - 4 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_152": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "first", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_152": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], "num_warps": 8, "num_stages": 1, "indexing": [ @@ -6255,460 +6707,10 @@ ], "pid_type": "flat" }, - "intermediate_14336_numtokens_152": { - "block_sizes": [ - 64, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_160": { - "block_sizes": [ - 32, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_160": { - "block_sizes": [ - 128, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_160": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_160": { - "block_sizes": [ - 64, - 4 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "first", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_160": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_160": { - "block_sizes": [ - 128, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "tensor_descriptor", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_168": { - "block_sizes": [ - 128, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_168": { - "block_sizes": [ - 32, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_168": { - "block_sizes": [ - 64, - 32 - ], - "loop_orders": [ - [ - 1, - 0 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "last" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, "intermediate_8192_numtokens_168": { "block_sizes": [ - 64, - 8 + 128, + 128 ], "loop_orders": [ [ @@ -6753,7 +6755,7 @@ "intermediate_11008_numtokens_168": { "block_sizes": [ 64, - 4 + 256 ], "loop_orders": [ [ @@ -6785,7 +6787,7 @@ "", "" ], - "num_warps": 16, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -6797,8 +6799,8 @@ }, "intermediate_14336_numtokens_168": { "block_sizes": [ - 32, - 512 + 64, + 32 ], "loop_orders": [ [ @@ -6810,7 +6812,7 @@ true ], "l2_groupings": [ - 2 + 64 ], "range_unroll_factors": [ 0 @@ -6826,68 +6828,23 @@ null ], "load_eviction_policies": [ - "", - "", - "" + "first", + "last", + "first" ], "num_warps": 2, - "num_stages": 1, + "num_stages": 6, "indexing": [ + "tensor_descriptor", "pointer", - "pointer", - "pointer", - "tensor_descriptor" + "tensor_descriptor", + "pointer" ], "pid_type": "flat" }, "intermediate_2048_numtokens_176": { "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 1, - 0 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_176": { - "block_sizes": [ - 32, + 128, 32 ], "loop_orders": [ @@ -6920,7 +6877,7 @@ "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -6930,10 +6887,10 @@ ], "pid_type": "flat" }, - "intermediate_4096_numtokens_176": { + "intermediate_2880_numtokens_176": { "block_sizes": [ - 4, - 8 + 16, + 256 ], "loop_orders": [ [ @@ -6960,15 +6917,60 @@ "range_flattens": [ null ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_176": { + "block_sizes": [ + 128, + 4 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], "load_eviction_policies": [ "", "", "first" ], - "num_warps": 8, - "num_stages": 1, + "num_warps": 4, + "num_stages": 6, "indexing": [ - "pointer", + "tensor_descriptor", "pointer", "pointer", "pointer" @@ -6977,9 +6979,54 @@ }, "intermediate_8192_numtokens_176": { "block_sizes": [ - 8, + 1, + 8192 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ 16 ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "first" + ], + "num_warps": 16, + "num_stages": 5, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_176": { + "block_sizes": [ + 64, + 256 + ], "loop_orders": [ [ 0, @@ -7010,52 +7057,7 @@ "", "" ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_176": { - "block_sizes": [ - 8, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -7067,8 +7069,8 @@ }, "intermediate_14336_numtokens_176": { "block_sizes": [ - 8, - 32 + 128, + 256 ], "loop_orders": [ [ @@ -7100,7 +7102,7 @@ "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -7112,53 +7114,8 @@ }, "intermediate_2048_numtokens_184": { "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_184": { - "block_sizes": [ - 8, - 16 + 2, + 256 ], "loop_orders": [ [ @@ -7170,7 +7127,7 @@ true ], "l2_groupings": [ - 1 + 2 ], "range_unroll_factors": [ 0 @@ -7191,19 +7148,64 @@ "" ], "num_warps": 16, - "num_stages": 2, + "num_stages": 6, "indexing": [ "pointer", "pointer", "pointer", - "tensor_descriptor" + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_184": { + "block_sizes": [ + 64, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" ], "pid_type": "flat" }, "intermediate_4096_numtokens_184": { "block_sizes": [ - 32, - 8 + 128, + 64 ], "loop_orders": [ [ @@ -7247,7 +7249,7 @@ }, "intermediate_8192_numtokens_184": { "block_sizes": [ - 8, + 64, 64 ], "loop_orders": [ @@ -7260,7 +7262,52 @@ true ], "l2_groupings": [ - 2 + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 32, + "num_stages": 8, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_184": { + "block_sizes": [ + 64, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 ], "range_unroll_factors": [ 0 @@ -7290,203 +7337,23 @@ ], "pid_type": "flat" }, - "intermediate_11008_numtokens_184": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, "intermediate_14336_numtokens_184": { "block_sizes": [ - 16, + 64, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ 64 ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "last" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_192": { - "block_sizes": [ - 32, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_192": { - "block_sizes": [ - 8, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_192": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 4 - ], "range_unroll_factors": [ 0 ], @@ -7503,21 +7370,21 @@ "load_eviction_policies": [ "first", "", - "" + "last" ], - "num_warps": 16, - "num_stages": 2, + "num_warps": 8, + "num_stages": 3, "indexing": [ + "tensor_descriptor", "pointer", "pointer", - "pointer", - "pointer" + "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_8192_numtokens_192": { + "intermediate_2048_numtokens_192": { "block_sizes": [ - 4, + 128, 32 ], "loop_orders": [ @@ -7530,7 +7397,7 @@ true ], "l2_groupings": [ - 2 + 1 ], "range_unroll_factors": [ 0 @@ -7560,9 +7427,144 @@ ], "pid_type": "flat" }, - "intermediate_11008_numtokens_192": { + "intermediate_2880_numtokens_192": { + "block_sizes": [ + 64, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_192": { + "block_sizes": [ + 8, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "first", + "first" + ], + "num_warps": 16, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_192": { "block_sizes": [ 32, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "" + ], + "num_warps": 32, + "num_stages": 1, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_192": { + "block_sizes": [ + 16, 256 ], "loop_orders": [ @@ -7593,22 +7595,22 @@ "load_eviction_policies": [ "", "", - "" + "first" ], - "num_warps": 2, - "num_stages": 2, + "num_warps": 32, + "num_stages": 7, "indexing": [ + "tensor_descriptor", "pointer", - "pointer", - "pointer", - "pointer" + "tensor_descriptor", + "tensor_descriptor" ], "pid_type": "flat" }, "intermediate_14336_numtokens_192": { "block_sizes": [ - 8, - 16 + 128, + 256 ], "loop_orders": [ [ @@ -7640,11 +7642,11 @@ "", "" ], - "num_warps": 4, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", - "tensor_descriptor", + "pointer", "pointer", "pointer" ], @@ -7652,7 +7654,7 @@ }, "intermediate_2048_numtokens_200": { "block_sizes": [ - 32, + 128, 32 ], "loop_orders": [ @@ -7665,7 +7667,7 @@ true ], "l2_groupings": [ - 2 + 1 ], "range_unroll_factors": [ 0 @@ -7685,7 +7687,7 @@ "", "" ], - "num_warps": 1, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -7697,8 +7699,8 @@ }, "intermediate_2880_numtokens_200": { "block_sizes": [ - 32, - 32 + 8, + 512 ], "loop_orders": [ [ @@ -7727,10 +7729,10 @@ ], "load_eviction_policies": [ "", - "last", + "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -7742,8 +7744,8 @@ }, "intermediate_4096_numtokens_200": { "block_sizes": [ - 64, - 32 + 4, + 512 ], "loop_orders": [ [ @@ -7755,7 +7757,7 @@ true ], "l2_groupings": [ - 1 + 16 ], "range_unroll_factors": [ 0 @@ -7771,24 +7773,24 @@ null ], "load_eviction_policies": [ + "last", "", - "", - "" + "first" ], - "num_warps": 2, + "num_warps": 1, "num_stages": 1, "indexing": [ "pointer", "pointer", "pointer", - "pointer" + "tensor_descriptor" ], "pid_type": "flat" }, "intermediate_8192_numtokens_200": { "block_sizes": [ - 32, - 32 + 128, + 128 ], "loop_orders": [ [ @@ -7820,7 +7822,7 @@ "", "" ], - "num_warps": 1, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -7832,17 +7834,17 @@ }, "intermediate_11008_numtokens_200": { "block_sizes": [ - 8, - 32 + 1, + 1024 ], "loop_orders": [ [ - 0, - 1 + 1, + 0 ] ], "flatten_loops": [ - true + false ], "l2_groupings": [ 1 @@ -7861,15 +7863,15 @@ null ], "load_eviction_policies": [ - "", - "", - "" + "first", + "first", + "first" ], - "num_warps": 1, - "num_stages": 1, + "num_warps": 32, + "num_stages": 3, "indexing": [ "pointer", - "pointer", + "tensor_descriptor", "pointer", "pointer" ], @@ -7878,7 +7880,7 @@ "intermediate_14336_numtokens_200": { "block_sizes": [ 16, - 8 + 128 ], "loop_orders": [ [ @@ -7890,7 +7892,7 @@ true ], "l2_groupings": [ - 1 + 64 ], "range_unroll_factors": [ 0 @@ -7910,154 +7912,19 @@ "", "first" ], - "num_warps": 4, - "num_stages": 1, + "num_warps": 32, + "num_stages": 6, "indexing": [ "pointer", "pointer", - "pointer", + "tensor_descriptor", "pointer" ], "pid_type": "flat" }, "intermediate_2048_numtokens_208": { "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_208": { - "block_sizes": [ - 64, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_208": { - "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_208": { - "block_sizes": [ - 256, + 128, 32 ], "loop_orders": [ @@ -8090,20 +7957,20 @@ "", "" ], - "num_warps": 16, - "num_stages": 2, + "num_warps": 8, + "num_stages": 1, "indexing": [ - "tensor_descriptor", - "tensor_descriptor", + "pointer", + "pointer", "pointer", "pointer" ], "pid_type": "flat" }, - "intermediate_11008_numtokens_208": { + "intermediate_2880_numtokens_208": { "block_sizes": [ - 64, - 64 + 256, + 16 ], "loop_orders": [ [ @@ -8115,7 +7982,7 @@ true ], "l2_groupings": [ - 1 + 8 ], "range_unroll_factors": [ 0 @@ -8131,24 +7998,24 @@ null ], "load_eviction_policies": [ - "last", - "last", - "" + "first", + "", + "last" ], - "num_warps": 8, - "num_stages": 1, + "num_warps": 32, + "num_stages": 8, "indexing": [ "pointer", "pointer", - "pointer", + "tensor_descriptor", "pointer" ], "pid_type": "flat" }, - "intermediate_14336_numtokens_208": { + "intermediate_4096_numtokens_208": { "block_sizes": [ - 16, - 128 + 256, + 32 ], "loop_orders": [ [ @@ -8175,12 +8042,147 @@ "range_flattens": [ null ], + "load_eviction_policies": [ + "last", + "first", + "last" + ], + "num_warps": 4, + "num_stages": 2, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_208": { + "block_sizes": [ + 128, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], "load_eviction_policies": [ "", "", "" ], - "num_warps": 32, + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_208": { + "block_sizes": [ + 32, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "first", + "first" + ], + "num_warps": 8, + "num_stages": 5, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_208": { + "block_sizes": [ + 128, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -8202,10 +8204,100 @@ ] ], "flatten_loops": [ - false + true ], "l2_groupings": [ - 2 + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "last" + ], + "num_warps": 8, + "num_stages": 4, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_216": { + "block_sizes": [ + 4, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "first", + "first" + ], + "num_warps": 8, + "num_stages": 7, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_216": { + "block_sizes": [ + 128, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 ], "range_unroll_factors": [ 0 @@ -8225,7 +8317,7 @@ "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -8235,10 +8327,10 @@ ], "pid_type": "flat" }, - "intermediate_2880_numtokens_216": { + "intermediate_8192_numtokens_216": { "block_sizes": [ - 16, - 128 + 1, + 8192 ], "loop_orders": [ [ @@ -8265,22 +8357,292 @@ "range_flattens": [ null ], + "load_eviction_policies": [ + "", + "", + "last" + ], + "num_warps": 2, + "num_stages": 7, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_216": { + "block_sizes": [ + 1, + 16384 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "last" + ], + "num_warps": 4, + "num_stages": 4, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_216": { + "block_sizes": [ + 128, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_224": { + "block_sizes": [ + 32, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "first", + "first" + ], + "num_warps": 16, + "num_stages": 5, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_224": { + "block_sizes": [ + 4, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_224": { + "block_sizes": [ + 128, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_224": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], "load_eviction_policies": [ "", "last", - "" + "first" ], "num_warps": 32, "num_stages": 1, "indexing": [ - "pointer", + "tensor_descriptor", "pointer", "pointer", "pointer" ], "pid_type": "flat" }, - "intermediate_4096_numtokens_216": { + "intermediate_11008_numtokens_224": { "block_sizes": [ 32, 32 @@ -8295,143 +8657,8 @@ true ], "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_216": { - "block_sizes": [ - 16, 32 ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_216": { - "block_sizes": [ - 32, - 4 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "first" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_216": { - "block_sizes": [ - 64, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], "range_unroll_factors": [ 0 ], @@ -8450,20 +8677,20 @@ "", "last" ], - "num_warps": 32, - "num_stages": 1, + "num_warps": 2, + "num_stages": 3, "indexing": [ "pointer", "pointer", - "pointer", + "tensor_descriptor", "pointer" ], "pid_type": "flat" }, - "intermediate_2048_numtokens_224": { + "intermediate_14336_numtokens_224": { "block_sizes": [ - 32, - 16 + 1, + 2048 ], "loop_orders": [ [ @@ -8475,7 +8702,7 @@ false ], "l2_groupings": [ - 1 + 32 ], "range_unroll_factors": [ 0 @@ -8492,126 +8719,36 @@ ], "load_eviction_policies": [ "", - "last", + "", "" ], - "num_warps": 2, - "num_stages": 1, + "num_warps": 8, + "num_stages": 8, "indexing": [ "pointer", + "tensor_descriptor", "pointer", - "pointer", - "pointer" + "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_2880_numtokens_224": { + "intermediate_2048_numtokens_232": { "block_sizes": [ - 64, + 1, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ 64 ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_224": { - "block_sizes": [ - 64, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 4 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_224": { - "block_sizes": [ - 16, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], "range_unroll_factors": [ 0 ], @@ -8632,51 +8769,6 @@ ], "num_warps": 16, "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_224": { - "block_sizes": [ - 256, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 1, "indexing": [ "tensor_descriptor", "pointer", @@ -8685,9 +8777,9 @@ ], "pid_type": "flat" }, - "intermediate_14336_numtokens_224": { + "intermediate_2880_numtokens_232": { "block_sizes": [ - 32, + 256, 8 ], "loop_orders": [ @@ -8700,7 +8792,7 @@ true ], "l2_groupings": [ - 1 + 4 ], "range_unroll_factors": [ 0 @@ -8717,10 +8809,10 @@ ], "load_eviction_policies": [ "last", - "", - "first" + "first", + "last" ], - "num_warps": 4, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -8730,100 +8822,10 @@ ], "pid_type": "flat" }, - "intermediate_2048_numtokens_232": { - "block_sizes": [ - 16, - 8 - ], - "loop_orders": [ - [ - 1, - 0 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 4 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_232": { - "block_sizes": [ - 64, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, "intermediate_4096_numtokens_232": { "block_sizes": [ - 16, - 4 + 128, + 64 ], "loop_orders": [ [ @@ -8867,20 +8869,20 @@ }, "intermediate_8192_numtokens_232": { "block_sizes": [ - 32, - 32 + 256, + 8 ], "loop_orders": [ [ - 1, - 0 + 0, + 1 ] ], "flatten_loops": [ - false + true ], "l2_groupings": [ - 1 + 2 ], "range_unroll_factors": [ 0 @@ -8898,9 +8900,9 @@ "load_eviction_policies": [ "", "", - "" + "first" ], - "num_warps": 4, + "num_warps": 16, "num_stages": 1, "indexing": [ "pointer", @@ -8912,20 +8914,20 @@ }, "intermediate_11008_numtokens_232": { "block_sizes": [ - 16, - 32 + 4, + 1024 ], "loop_orders": [ [ - 0, - 1 + 1, + 0 ] ], "flatten_loops": [ - true + false ], "l2_groupings": [ - 1 + 32 ], "range_unroll_factors": [ 0 @@ -8941,24 +8943,24 @@ null ], "load_eviction_policies": [ - "", + "first", "last", - "" + "first" ], - "num_warps": 2, - "num_stages": 2, + "num_warps": 1, + "num_stages": 8, "indexing": [ "pointer", + "tensor_descriptor", "pointer", - "pointer", - "pointer" + "tensor_descriptor" ], "pid_type": "flat" }, "intermediate_14336_numtokens_232": { "block_sizes": [ - 32, - 8 + 8, + 4096 ], "loop_orders": [ [ @@ -8990,7 +8992,7 @@ "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -9002,142 +9004,7 @@ }, "intermediate_2048_numtokens_240": { "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_240": { - "block_sizes": [ - 8, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_240": { - "block_sizes": [ - 16, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "tensor_descriptor", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_240": { - "block_sizes": [ - 32, + 64, 8 ], "loop_orders": [ @@ -9149,231 +9016,6 @@ "flatten_loops": [ true ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_240": { - "block_sizes": [ - 8, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_240": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_248": { - "block_sizes": [ - 16, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_248": { - "block_sizes": [ - 16, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_248": { - "block_sizes": [ - 256, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], "l2_groupings": [ 2 ], @@ -9393,9 +9035,54 @@ "load_eviction_policies": [ "first", "", + "last" + ], + "num_warps": 4, + "num_stages": 5, + "indexing": [ + "pointer", + "pointer", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_240": { + "block_sizes": [ + 4, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", "" ], - "num_warps": 16, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -9405,9 +9092,189 @@ ], "pid_type": "flat" }, - "intermediate_8192_numtokens_248": { + "intermediate_4096_numtokens_240": { "block_sizes": [ - 64, + 4, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "" + ], + "num_warps": 8, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_240": { + "block_sizes": [ + 1, + 1024 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "first", + "last" + ], + "num_warps": 4, + "num_stages": 8, + "indexing": [ + "pointer", + "pointer", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_240": { + "block_sizes": [ + 4, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "first" + ], + "num_warps": 32, + "num_stages": 7, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_240": { + "block_sizes": [ + 1, + 8192 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "last" + ], + "num_warps": 32, + "num_stages": 2, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_248": { + "block_sizes": [ + 128, 32 ], "loop_orders": [ @@ -9438,10 +9305,10 @@ "load_eviction_policies": [ "", "", - "first" + "" ], - "num_warps": 4, - "num_stages": 2, + "num_warps": 8, + "num_stages": 1, "indexing": [ "pointer", "pointer", @@ -9450,10 +9317,10 @@ ], "pid_type": "flat" }, - "intermediate_11008_numtokens_248": { + "intermediate_2880_numtokens_248": { "block_sizes": [ - 64, - 4 + 4, + 2048 ], "loop_orders": [ [ @@ -9481,11 +9348,146 @@ null ], "load_eviction_policies": [ - "last", + "", "", "" ], - "num_warps": 1, + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_248": { + "block_sizes": [ + 128, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_248": { + "block_sizes": [ + 256, + 16 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 2, + "num_stages": 4, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_248": { + "block_sizes": [ + 4, + 8192 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -9497,8 +9499,8 @@ }, "intermediate_14336_numtokens_248": { "block_sizes": [ - 64, - 256 + 8, + 4096 ], "loop_orders": [ [ @@ -9541,456 +9543,6 @@ "pid_type": "flat" }, "intermediate_2048_numtokens_272": { - "block_sizes": [ - 128, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_272": { - "block_sizes": [ - 8, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_272": { - "block_sizes": [ - 128, - 32 - ], - "loop_orders": [ - [ - 1, - 0 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_272": { - "block_sizes": [ - 128, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_272": { - "block_sizes": [ - 16, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_272": { - "block_sizes": [ - 64, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_288": { - "block_sizes": [ - 4, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "last", - "", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_288": { - "block_sizes": [ - 8, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "tensor_descriptor", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_288": { - "block_sizes": [ - 64, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_288": { - "block_sizes": [ - 128, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_288": { "block_sizes": [ 256, 32 @@ -10004,51 +9556,6 @@ "flatten_loops": [ true ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_288": { - "block_sizes": [ - 16, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], "l2_groupings": [ 1 ], @@ -10070,7 +9577,7 @@ "", "" ], - "num_warps": 1, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -10080,9 +9587,9 @@ ], "pid_type": "flat" }, - "intermediate_2048_numtokens_304": { + "intermediate_2880_numtokens_272": { "block_sizes": [ - 8, + 128, 64 ], "loop_orders": [ @@ -10115,96 +9622,6 @@ "", "" ], - "num_warps": 32, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_304": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_304": { - "block_sizes": [ - 128, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], "num_warps": 8, "num_stages": 1, "indexing": [ @@ -10215,22 +9632,22 @@ ], "pid_type": "flat" }, - "intermediate_8192_numtokens_304": { + "intermediate_4096_numtokens_272": { "block_sizes": [ - 8, - 64 + 1, + 4096 ], "loop_orders": [ [ - 0, - 1 + 1, + 0 ] ], "flatten_loops": [ - true + false ], "l2_groupings": [ - 1 + 64 ], "range_unroll_factors": [ 0 @@ -10248,157 +9665,22 @@ "load_eviction_policies": [ "", "last", - "" + "first" ], - "num_warps": 16, + "num_warps": 32, "num_stages": 1, "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_304": { - "block_sizes": [ - 64, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_304": { - "block_sizes": [ - 64, - 4 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_320": { - "block_sizes": [ - 128, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", + "tensor_descriptor", "pointer", "tensor_descriptor", - "pointer" + "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_2880_numtokens_320": { + "intermediate_8192_numtokens_272": { "block_sizes": [ - 64, - 32 + 8, + 512 ], "loop_orders": [ [ @@ -10410,7 +9692,97 @@ true ], "l2_groupings": [ - 1 + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "", + "last" + ], + "num_warps": 2, + "num_stages": 6, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_272": { + "block_sizes": [ + 8, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "", + "last" + ], + "num_warps": 4, + "num_stages": 8, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_272": { + "block_sizes": [ + 512, + 16 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 32 ], "range_unroll_factors": [ 0 @@ -10428,19 +9800,109 @@ "load_eviction_policies": [ "", "", - "" + "last" ], - "num_warps": 4, - "num_stages": 1, + "num_warps": 8, + "num_stages": 8, "indexing": [ "pointer", "pointer", "pointer", - "pointer" + "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_4096_numtokens_320": { + "intermediate_2048_numtokens_288": { + "block_sizes": [ + 64, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "last" + ], + "num_warps": 32, + "num_stages": 4, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_288": { + "block_sizes": [ + 8, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "last" + ], + "num_warps": 2, + "num_stages": 1, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_288": { "block_sizes": [ 512, 4 @@ -10455,7 +9917,97 @@ true ], "l2_groupings": [ - 1 + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "first" + ], + "num_warps": 1, + "num_stages": 2, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_288": { + "block_sizes": [ + 1, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "first", + "" + ], + "num_warps": 2, + "num_stages": 3, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_288": { + "block_sizes": [ + 1, + 8192 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 2 ], "range_unroll_factors": [ 0 @@ -10473,10 +10025,10 @@ "load_eviction_policies": [ "last", "", - "" + "last" ], - "num_warps": 1, - "num_stages": 2, + "num_warps": 8, + "num_stages": 3, "indexing": [ "pointer", "pointer", @@ -10485,10 +10037,55 @@ ], "pid_type": "flat" }, - "intermediate_8192_numtokens_320": { + "intermediate_14336_numtokens_288": { "block_sizes": [ - 64, - 128 + 1, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "last" + ], + "num_warps": 1, + "num_stages": 5, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_304": { + "block_sizes": [ + 256, + 32 ], "loop_orders": [ [ @@ -10530,11 +10127,103 @@ ], "pid_type": "flat" }, - "intermediate_11008_numtokens_320": { + "intermediate_2880_numtokens_304": { "block_sizes": [ - 32, + 1, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 2 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 2 + ], + "range_multi_buffers": [ + false + ], + "range_flattens": [ + true + ], + "load_eviction_policies": [ + "last", + "", + "last" + ], + "num_warps": 16, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "persistent_blocked", + "num_sm_multiplier": 2, + "maxnreg": 64 + }, + "intermediate_4096_numtokens_304": { + "block_sizes": [ + 16, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ 32 ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "last" + ], + "num_warps": 16, + "num_stages": 2, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_304": { + "block_sizes": [ + 1, + 1024 + ], "loop_orders": [ [ 0, @@ -10562,23 +10251,23 @@ ], "load_eviction_policies": [ "", - "", - "" + "last", + "last" ], - "num_warps": 8, - "num_stages": 1, + "num_warps": 32, + "num_stages": 4, "indexing": [ - "tensor_descriptor", "tensor_descriptor", "pointer", - "pointer" + "tensor_descriptor", + "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_14336_numtokens_320": { + "intermediate_11008_numtokens_304": { "block_sizes": [ 128, - 16 + 256 ], "loop_orders": [ [ @@ -10610,8 +10299,323 @@ "", "" ], - "num_warps": 32, + "num_warps": 8, "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_304": { + "block_sizes": [ + 4, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "" + ], + "num_warps": 16, + "num_stages": 6, + "indexing": [ + "pointer", + "pointer", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_320": { + "block_sizes": [ + 1, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "" + ], + "num_warps": 2, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_320": { + "block_sizes": [ + 128, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_320": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "last" + ], + "num_warps": 8, + "num_stages": 7, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_320": { + "block_sizes": [ + 1, + 8192 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "" + ], + "num_warps": 16, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_320": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "first", + "" + ], + "num_warps": 2, + "num_stages": 3, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_320": { + "block_sizes": [ + 8, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "first" + ], + "num_warps": 32, + "num_stages": 3, "indexing": [ "pointer", "tensor_descriptor", @@ -10622,7 +10626,7 @@ }, "intermediate_2048_numtokens_336": { "block_sizes": [ - 2, + 256, 32 ], "loop_orders": [ @@ -10660,3029 +10664,14 @@ "indexing": [ "pointer", "pointer", - "tensor_descriptor", + "pointer", "pointer" ], "pid_type": "flat" }, "intermediate_2880_numtokens_336": { - "block_sizes": [ - 8, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_336": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_336": { - "block_sizes": [ - 64, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "last" - ], - "num_warps": 8, - "num_stages": 2, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_336": { - "block_sizes": [ - 8, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "last", - "", - "" - ], - "num_warps": 2, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_336": { - "block_sizes": [ - 8, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_352": { - "block_sizes": [ - 32, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_352": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_352": { "block_sizes": [ 16, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "last" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_352": { - "block_sizes": [ - 64, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_352": { - "block_sizes": [ - 8, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_352": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "last" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_368": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_368": { - "block_sizes": [ - 8, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_368": { - "block_sizes": [ - 8, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_368": { - "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_368": { - "block_sizes": [ - 32, - 4 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_368": { - "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "first", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_384": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_384": { - "block_sizes": [ - 64, - 256 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "first" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_384": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "last" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_384": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_384": { - "block_sizes": [ - 8, - 256 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_384": { - "block_sizes": [ - 32, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_400": { - "block_sizes": [ - 64, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "first" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_400": { - "block_sizes": [ - 8, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_400": { - "block_sizes": [ - 64, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_400": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_400": { - "block_sizes": [ - 256, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "last" - ], - "num_warps": 32, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_400": { - "block_sizes": [ - 8, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_416": { - "block_sizes": [ - 128, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_416": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "first" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_416": { - "block_sizes": [ - 64, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "first", - "", - "last" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_416": { - "block_sizes": [ - 128, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_416": { - "block_sizes": [ - 64, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "first", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_416": { - "block_sizes": [ - 32, - 256 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_432": { - "block_sizes": [ - 16, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_432": { - "block_sizes": [ - 32, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_432": { - "block_sizes": [ - 16, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 4, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_432": { - "block_sizes": [ - 16, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_432": { - "block_sizes": [ - 16, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_432": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 2, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_448": { - "block_sizes": [ - 4, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_448": { - "block_sizes": [ - 8, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_448": { - "block_sizes": [ - 4, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_448": { - "block_sizes": [ - 32, - 256 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "last", - "", - "" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_448": { - "block_sizes": [ - 16, - 256 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_448": { - "block_sizes": [ - 16, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_464": { - "block_sizes": [ - 32, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "last" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_464": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_464": { - "block_sizes": [ - 16, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_464": { - "block_sizes": [ - 8, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_464": { - "block_sizes": [ - 128, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "first" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_464": { - "block_sizes": [ - 128, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "first", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_480": { - "block_sizes": [ - 4, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "last", - "", - "" - ], - "num_warps": 2, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_480": { - "block_sizes": [ - 4, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "first", - "first" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_480": { - "block_sizes": [ - 8, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "last", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_480": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_480": { - "block_sizes": [ - 64, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "last", - "", - "last" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_480": { - "block_sizes": [ - 16, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 4 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "last", - "", - "" - ], - "num_warps": 8, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_496": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_496": { - "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "first", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_496": { - "block_sizes": [ - 16, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 4 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_496": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_496": { - "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_496": { - "block_sizes": [ - 256, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_512": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "last" - ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_512": { - "block_sizes": [ - 16, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_512": { - "block_sizes": [ - 128, 512 ], "loop_orders": [ @@ -13715,8 +10704,8 @@ "", "" ], - "num_warps": 16, - "num_stages": 2, + "num_warps": 8, + "num_stages": 1, "indexing": [ "pointer", "pointer", @@ -13725,9 +10714,54 @@ ], "pid_type": "flat" }, - "intermediate_8192_numtokens_512": { + "intermediate_4096_numtokens_336": { "block_sizes": [ - 32, + 16, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "", + "first" + ], + "num_warps": 2, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_336": { + "block_sizes": [ + 256, 128 ], "loop_orders": [ @@ -13736,6 +10770,276 @@ 1 ] ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_336": { + "block_sizes": [ + 4, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "" + ], + "num_warps": 4, + "num_stages": 7, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_336": { + "block_sizes": [ + 256, + 8 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "last" + ], + "num_warps": 16, + "num_stages": 8, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_352": { + "block_sizes": [ + 512, + 1 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "first" + ], + "num_warps": 1, + "num_stages": 4, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_352": { + "block_sizes": [ + 4, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "" + ], + "num_warps": 32, + "num_stages": 7, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_352": { + "block_sizes": [ + 512, + 4 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 16, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_352": { + "block_sizes": [ + 1, + 8192 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], "flatten_loops": [ false ], @@ -13755,12 +11059,102 @@ "range_flattens": [ null ], + "load_eviction_policies": [ + "", + "last", + "first" + ], + "num_warps": 16, + "num_stages": 2, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_352": { + "block_sizes": [ + 16, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "" + ], + "num_warps": 16, + "num_stages": 2, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_352": { + "block_sizes": [ + 32, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], "load_eviction_policies": [ "", "", "" ], - "num_warps": 32, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -13770,9 +11164,1764 @@ ], "pid_type": "flat" }, - "intermediate_11008_numtokens_512": { + "intermediate_2048_numtokens_368": { + "block_sizes": [ + 4, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "first", + "first" + ], + "num_warps": 8, + "num_stages": 4, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_368": { + "block_sizes": [ + 128, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "first", + "" + ], + "num_warps": 4, + "num_stages": 4, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_368": { + "block_sizes": [ + 64, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 32, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_368": { + "block_sizes": [ + 2, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "last" + ], + "num_warps": 1, + "num_stages": 4, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_368": { + "block_sizes": [ + 128, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_368": { "block_sizes": [ 32, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_384": { + "block_sizes": [ + 256, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_384": { + "block_sizes": [ + 512, + 2 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "first", + "last" + ], + "num_warps": 8, + "num_stages": 3, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_384": { + "block_sizes": [ + 4, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "last" + ], + "num_warps": 8, + "num_stages": 5, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_384": { + "block_sizes": [ + 128, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "first" + ], + "num_warps": 4, + "num_stages": 2, + "indexing": [ + "pointer", + "pointer", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_384": { + "block_sizes": [ + 1, + 8192 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "first" + ], + "num_warps": 4, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_384": { + "block_sizes": [ + 128, + 16 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "first" + ], + "num_warps": 32, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_400": { + "block_sizes": [ + 1, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 4, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_400": { + "block_sizes": [ + 16, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_400": { + "block_sizes": [ + 1, + 1024 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "first" + ], + "num_warps": 1, + "num_stages": 1, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_400": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "", + "last" + ], + "num_warps": 8, + "num_stages": 4, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_400": { + "block_sizes": [ + 2, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "last" + ], + "num_warps": 4, + "num_stages": 3, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_400": { + "block_sizes": [ + 4, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "first" + ], + "num_warps": 8, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_416": { + "block_sizes": [ + 256, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_416": { + "block_sizes": [ + 32, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_416": { + "block_sizes": [ + 512, + 8 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "" + ], + "num_warps": 8, + "num_stages": 7, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_416": { + "block_sizes": [ + 1, + 2048 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "first" + ], + "num_warps": 8, + "num_stages": 8, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_416": { + "block_sizes": [ + 256, + 8 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "", + "last" + ], + "num_warps": 4, + "num_stages": 7, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_416": { + "block_sizes": [ + 128, + 16 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "first", + "first" + ], + "num_warps": 16, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_432": { + "block_sizes": [ + 256, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_432": { + "block_sizes": [ + 8, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_432": { + "block_sizes": [ + 64, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_432": { + "block_sizes": [ + 256, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "" + ], + "num_warps": 32, + "num_stages": 5, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_432": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "first", + "first" + ], + "num_warps": 1, + "num_stages": 8, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_432": { + "block_sizes": [ + 512, + 4 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "last" + ], + "num_warps": 1, + "num_stages": 7, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_448": { + "block_sizes": [ + 256, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_448": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "" + ], + "num_warps": 2, + "num_stages": 6, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_448": { + "block_sizes": [ + 8, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "last" + ], + "num_warps": 16, + "num_stages": 2, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_448": { + "block_sizes": [ + 128, + 8 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "first", + "last" + ], + "num_warps": 32, + "num_stages": 3, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_448": { + "block_sizes": [ + 1, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 4, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_448": { + "block_sizes": [ + 64, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 32, + "num_stages": 8, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_464": { + "block_sizes": [ + 256, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_464": { + "block_sizes": [ + 8, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_464": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 1, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_464": { + "block_sizes": [ + 256, 128 ], "loop_orders": [ @@ -13805,8 +12954,278 @@ "", "" ], - "num_warps": 1, + "num_warps": 8, "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_464": { + "block_sizes": [ + 1, + 16384 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 32, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_464": { + "block_sizes": [ + 64, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "" + ], + "num_warps": 32, + "num_stages": 7, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_480": { + "block_sizes": [ + 16, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "first", + "" + ], + "num_warps": 16, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_480": { + "block_sizes": [ + 128, + 16 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "", + "" + ], + "num_warps": 8, + "num_stages": 5, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_480": { + "block_sizes": [ + 64, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "first" + ], + "num_warps": 2, + "num_stages": 1, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_480": { + "block_sizes": [ + 1, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "first", + "" + ], + "num_warps": 1, + "num_stages": 2, "indexing": [ "tensor_descriptor", "pointer", @@ -13815,9 +13234,459 @@ ], "pid_type": "flat" }, - "intermediate_14336_numtokens_512": { + "intermediate_11008_numtokens_480": { "block_sizes": [ - 16, + 1, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 4, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_480": { + "block_sizes": [ + 1, + 16384 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "first" + ], + "num_warps": 32, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_496": { + "block_sizes": [ + 1, + 2048 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "last" + ], + "num_warps": 16, + "num_stages": 7, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_496": { + "block_sizes": [ + 8, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 4, + "num_stages": 8, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_496": { + "block_sizes": [ + 256, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_496": { + "block_sizes": [ + 256, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_496": { + "block_sizes": [ + 1, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 8, + "num_stages": 4, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_496": { + "block_sizes": [ + 4, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "first" + ], + "num_warps": 4, + "num_stages": 4, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_512": { + "block_sizes": [ + 512, + 16 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_512": { + "block_sizes": [ + 8, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_512": { + "block_sizes": [ + 8, 128 ], "loop_orders": [ @@ -13835,6 +13704,97 @@ "range_unroll_factors": [ 0 ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "last" + ], + "num_warps": 16, + "num_stages": 2, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_512": { + "block_sizes": [ + 1, + 2048 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "last" + ], + "num_warps": 4, + "num_stages": 4, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_512": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], "range_num_stages": [ 0 ], @@ -13847,18 +13807,62 @@ "load_eviction_policies": [ "first", "", + "first" + ], + "num_warps": 16, + "num_stages": 7, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_512": { + "block_sizes": [ + 128, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", "" ], - "num_warps": 1, - "num_stages": 2, + "num_warps": 2, + "num_stages": 7, "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" ], - "pid_type": "flat", - "range_warp_specializes": [] + "pid_type": "flat" } }, "nvidia_h100": { @@ -13909,8 +13913,8 @@ }, "intermediate_4096_numtokens_256": { "block_sizes": [ - 32, - 512 + 256, + 32 ], "loop_orders": [ [ @@ -13922,7 +13926,7 @@ true ], "l2_groupings": [ - 2 + 1 ], "range_unroll_factors": [ 0 @@ -13942,11 +13946,11 @@ "", "" ], - "num_warps": 2, - "num_stages": 2, + "num_warps": 8, + "num_stages": 1, "indexing": [ "pointer", - "tensor_descriptor", + "pointer", "pointer", "pointer" ], @@ -13999,8 +14003,8 @@ }, "intermediate_8192_numtokens_256": { "block_sizes": [ - 32, - 8 + 256, + 64 ], "loop_orders": [ [ @@ -14029,23 +14033,23 @@ ], "load_eviction_policies": [ "", - "first", + "", "" ], - "num_warps": 4, + "num_warps": 8, "num_stages": 1, "indexing": [ - "tensor_descriptor", "pointer", "pointer", - "tensor_descriptor" + "pointer", + "pointer" ], "pid_type": "flat" }, "intermediate_14336_numtokens_256": { "block_sizes": [ - 16, - 32 + 8, + 4096 ], "loop_orders": [ [ @@ -14077,13 +14081,13 @@ "", "" ], - "num_warps": 1, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", "pointer", "pointer", - "tensor_descriptor" + "pointer" ], "pid_type": "flat" }, @@ -14134,8 +14138,8 @@ }, "intermediate_7688_numtokens_256": { "block_sizes": [ - 8, - 16 + 32, + 512 ], "loop_orders": [ [ @@ -14164,10 +14168,10 @@ ], "load_eviction_policies": [ "", - "last", + "", "" ], - "num_warps": 1, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -14225,187 +14229,7 @@ "intermediate_2048_numtokens_1": { "block_sizes": [ 1, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_1": { - "block_sizes": [ - 1, - 1 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_1": { - "block_sizes": [ - 1, - 32 - ], - "loop_orders": [ - [ - 1, - 0 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_1": { - "block_sizes": [ - 1, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_1": { - "block_sizes": [ - 1, - 32 + 256 ], "loop_orders": [ [ @@ -14432,96 +14256,6 @@ "range_flattens": [ null ], - "load_eviction_policies": [ - "first", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_1": { - "block_sizes": [ - 1, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "first" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_2": { - "block_sizes": [ - 2, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], "load_eviction_policies": [ "", "", @@ -14532,15 +14266,15 @@ "indexing": [ "pointer", "pointer", - "tensor_descriptor", + "pointer", "pointer" ], "pid_type": "flat" }, - "intermediate_2880_numtokens_2": { + "intermediate_2880_numtokens_1": { "block_sizes": [ 1, - 4 + 128 ], "loop_orders": [ [ @@ -14549,7 +14283,7 @@ ] ], "flatten_loops": [ - false + true ], "l2_groupings": [ 1 @@ -14568,11 +14302,281 @@ null ], "load_eviction_policies": [ - "first", + "", "", "" ], - "num_warps": 32, + "num_warps": 4, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_1": { + "block_sizes": [ + 1, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 2, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_1": { + "block_sizes": [ + 1, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 2, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_1": { + "block_sizes": [ + 1, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 2, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_1": { + "block_sizes": [ + 1, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 2, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_2": { + "block_sizes": [ + 2, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 2, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_2": { + "block_sizes": [ + 2, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 4, "num_stages": 1, "indexing": [ "pointer", @@ -14585,7 +14589,7 @@ "intermediate_4096_numtokens_2": { "block_sizes": [ 2, - 32 + 128 ], "loop_orders": [ [ @@ -14617,8 +14621,8 @@ "", "" ], - "num_warps": 8, - "num_stages": 2, + "num_warps": 2, + "num_stages": 1, "indexing": [ "pointer", "pointer", @@ -14629,8 +14633,8 @@ }, "intermediate_8192_numtokens_2": { "block_sizes": [ - 1, - 32 + 2, + 128 ], "loop_orders": [ [ @@ -14639,7 +14643,7 @@ ] ], "flatten_loops": [ - false + true ], "l2_groupings": [ 1 @@ -14658,14 +14662,14 @@ null ], "load_eviction_policies": [ - "first", + "", "", "" ], - "num_warps": 8, - "num_stages": 2, + "num_warps": 2, + "num_stages": 1, "indexing": [ - "tensor_descriptor", + "pointer", "pointer", "pointer", "pointer" @@ -14675,200 +14679,20 @@ "intermediate_11008_numtokens_2": { "block_sizes": [ 1, - 256 + 16384 ], "loop_orders": [ [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "first", - "" - ], - "num_warps": 4, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_2": { - "block_sizes": [ - 1, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_4": { - "block_sizes": [ - 1, - 256 - ], - "loop_orders": [ - [ - 0, - 1 + 1, + 0 ] ], "flatten_loops": [ true ], "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_4": { - "block_sizes": [ - 1, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_4": { - "block_sizes": [ - 4, 16 ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], "range_unroll_factors": [ 0 ], @@ -14884,10 +14708,190 @@ ], "load_eviction_policies": [ "last", + "first", + "last" + ], + "num_warps": 32, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "tensor_descriptor" + ], + "pid_type": "xyz" + }, + "intermediate_14336_numtokens_2": { + "block_sizes": [ + 2, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", "", "" ], - "num_warps": 4, + "num_warps": 2, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_4": { + "block_sizes": [ + 4, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 2, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_4": { + "block_sizes": [ + 4, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "first" + ], + "num_warps": 8, + "num_stages": 7, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "xyz" + }, + "intermediate_4096_numtokens_4": { + "block_sizes": [ + 4, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 2, "num_stages": 1, "indexing": [ "pointer", @@ -14900,52 +14904,7 @@ "intermediate_8192_numtokens_4": { "block_sizes": [ 1, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_4": { - "block_sizes": [ - 1, - 32 + 2048 ], "loop_orders": [ [ @@ -14974,23 +14933,23 @@ ], "load_eviction_policies": [ "", - "first", + "", "" ], - "num_warps": 16, - "num_stages": 1, + "num_warps": 8, + "num_stages": 8, "indexing": [ - "pointer", "pointer", "tensor_descriptor", + "pointer", "pointer" ], "pid_type": "flat" }, - "intermediate_14336_numtokens_4": { + "intermediate_11008_numtokens_4": { "block_sizes": [ 4, - 16 + 2048 ], "loop_orders": [ [ @@ -15001,6 +14960,96 @@ "flatten_loops": [ true ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 8, + "num_stages": 6, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "xyz" + }, + "intermediate_14336_numtokens_4": { + "block_sizes": [ + 4, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 2, + "num_stages": 2, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "xyz" + }, + "intermediate_2048_numtokens_8": { + "block_sizes": [ + 8, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], "l2_groupings": [ 1 ], @@ -15022,7 +15071,7 @@ "", "" ], - "num_warps": 16, + "num_warps": 2, "num_stages": 1, "indexing": [ "pointer", @@ -15032,22 +15081,22 @@ ], "pid_type": "flat" }, - "intermediate_2048_numtokens_8": { + "intermediate_2880_numtokens_8": { "block_sizes": [ - 8, - 256 + 4, + 2048 ], "loop_orders": [ [ - 1, - 0 + 0, + 1 ] ], "flatten_loops": [ true ], "l2_groupings": [ - 1 + 4 ], "range_unroll_factors": [ 0 @@ -15064,20 +15113,20 @@ ], "load_eviction_policies": [ "last", - "", + "last", "" ], - "num_warps": 32, - "num_stages": 1, + "num_warps": 16, + "num_stages": 5, "indexing": [ + "pointer", + "tensor_descriptor", "tensor_descriptor", - "pointer", - "pointer", "pointer" ], - "pid_type": "flat" + "pid_type": "xyz" }, - "intermediate_2880_numtokens_8": { + "intermediate_4096_numtokens_8": { "block_sizes": [ 8, 32 @@ -15112,52 +15161,7 @@ "", "" ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_8": { - "block_sizes": [ - 2, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, + "num_warps": 2, "num_stages": 1, "indexing": [ "pointer", @@ -15169,8 +15173,8 @@ }, "intermediate_8192_numtokens_8": { "block_sizes": [ - 4, - 64 + 2, + 1024 ], "loop_orders": [ [ @@ -15179,10 +15183,10 @@ ] ], "flatten_loops": [ - false + true ], "l2_groupings": [ - 1 + 16 ], "range_unroll_factors": [ 0 @@ -15198,36 +15202,36 @@ null ], "load_eviction_policies": [ - "", - "", + "first", + "last", "" ], "num_warps": 1, - "num_stages": 1, + "num_stages": 8, "indexing": [ + "tensor_descriptor", "pointer", - "pointer", - "pointer", + "tensor_descriptor", "pointer" ], "pid_type": "flat" }, "intermediate_11008_numtokens_8": { "block_sizes": [ - 8, - 128 + 1, + 1024 ], "loop_orders": [ [ - 0, - 1 + 1, + 0 ] ], "flatten_loops": [ - false + true ], "l2_groupings": [ - 1 + 4 ], "range_unroll_factors": [ 0 @@ -15245,22 +15249,22 @@ "load_eviction_policies": [ "last", "", - "" + "first" ], - "num_warps": 4, - "num_stages": 2, + "num_warps": 2, + "num_stages": 5, "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" ], "pid_type": "flat" }, "intermediate_14336_numtokens_8": { "block_sizes": [ 8, - 32 + 128 ], "loop_orders": [ [ @@ -15269,7 +15273,7 @@ ] ], "flatten_loops": [ - false + true ], "l2_groupings": [ 1 @@ -15289,7 +15293,7 @@ ], "load_eviction_policies": [ "", - "first", + "", "" ], "num_warps": 4, @@ -15304,8 +15308,8 @@ }, "intermediate_2048_numtokens_16": { "block_sizes": [ - 16, - 64 + 8, + 512 ], "loop_orders": [ [ @@ -15314,10 +15318,55 @@ ] ], "flatten_loops": [ - false + true ], "l2_groupings": [ - 2 + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "first" + ], + "num_warps": 16, + "num_stages": 2, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "xyz" + }, + "intermediate_2880_numtokens_16": { + "block_sizes": [ + 2, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 ], "range_unroll_factors": [ 0 @@ -15338,51 +15387,6 @@ "" ], "num_warps": 4, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_16": { - "block_sizes": [ - 16, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -15395,7 +15399,97 @@ "intermediate_4096_numtokens_16": { "block_sizes": [ 16, - 32 + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "", + "last" + ], + "num_warps": 32, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_16": { + "block_sizes": [ + 16, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 4, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_16": { + "block_sizes": [ + 1, + 2048 ], "loop_orders": [ [ @@ -15428,7 +15522,7 @@ "" ], "num_warps": 8, - "num_stages": 2, + "num_stages": 1, "indexing": [ "pointer", "pointer", @@ -15437,22 +15531,22 @@ ], "pid_type": "flat" }, - "intermediate_8192_numtokens_16": { + "intermediate_14336_numtokens_16": { "block_sizes": [ - 4, - 32 + 2, + 256 ], "loop_orders": [ [ - 0, - 1 + 1, + 0 ] ], "flatten_loops": [ - false + true ], "l2_groupings": [ - 1 + 64 ], "range_unroll_factors": [ 0 @@ -15468,24 +15562,24 @@ null ], "load_eviction_policies": [ - "", - "", - "" + "last", + "last", + "last" ], - "num_warps": 1, - "num_stages": 1, + "num_warps": 16, + "num_stages": 7, "indexing": [ - "pointer", "tensor_descriptor", "pointer", - "pointer" + "tensor_descriptor", + "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_11008_numtokens_16": { + "intermediate_2048_numtokens_24": { "block_sizes": [ - 8, - 32 + 1, + 1024 ], "loop_orders": [ [ @@ -15497,7 +15591,7 @@ true ], "l2_groupings": [ - 1 + 64 ], "range_unroll_factors": [ 0 @@ -15517,115 +15611,70 @@ "", "first" ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_16": { - "block_sizes": [ - 16, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, + "num_warps": 4, + "num_stages": 8, "indexing": [ "tensor_descriptor", "pointer", "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_24": { - "block_sizes": [ - 16, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "last" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" + "tensor_descriptor" ], "pid_type": "flat" }, "intermediate_2880_numtokens_24": { "block_sizes": [ - 32, + 4, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "first" + ], + "num_warps": 16, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_24": { + "block_sizes": [ + 16, 64 ], "loop_orders": [ [ - 1, - 0 + 0, + 1 ] ], "flatten_loops": [ @@ -15655,17 +15704,17 @@ "num_warps": 4, "num_stages": 1, "indexing": [ - "tensor_descriptor", "pointer", "pointer", - "tensor_descriptor" + "pointer", + "pointer" ], "pid_type": "flat" }, - "intermediate_4096_numtokens_24": { + "intermediate_8192_numtokens_24": { "block_sizes": [ - 32, - 32 + 1, + 2048 ], "loop_orders": [ [ @@ -15677,7 +15726,7 @@ true ], "l2_groupings": [ - 2 + 32 ], "range_unroll_factors": [ 0 @@ -15694,11 +15743,11 @@ ], "load_eviction_policies": [ "", - "first", - "" + "last", + "last" ], - "num_warps": 1, - "num_stages": 1, + "num_warps": 32, + "num_stages": 5, "indexing": [ "pointer", "pointer", @@ -15707,22 +15756,22 @@ ], "pid_type": "flat" }, - "intermediate_8192_numtokens_24": { + "intermediate_11008_numtokens_24": { "block_sizes": [ - 16, - 32 + 1, + 1024 ], "loop_orders": [ [ - 0, - 1 + 1, + 0 ] ], "flatten_loops": [ false ], "l2_groupings": [ - 1 + 64 ], "range_unroll_factors": [ 0 @@ -15738,57 +15787,12 @@ null ], "load_eviction_policies": [ - "", + "last", "", "" ], "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_24": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, + "num_stages": 2, "indexing": [ "pointer", "pointer", @@ -15800,19 +15804,19 @@ "intermediate_14336_numtokens_24": { "block_sizes": [ 8, - 32 + 512 ], "loop_orders": [ [ - 1, - 0 + 0, + 1 ] ], "flatten_loops": [ true ], "l2_groupings": [ - 1 + 4 ], "range_unroll_factors": [ 0 @@ -15828,111 +15832,21 @@ null ], "load_eviction_policies": [ - "", - "", - "last" + "last", + "first", + "first" ], - "num_warps": 2, - "num_stages": 1, + "num_warps": 32, + "num_stages": 3, "indexing": [ "pointer", + "tensor_descriptor", "pointer", - "pointer", - "pointer" + "tensor_descriptor" ], "pid_type": "flat" }, "intermediate_2048_numtokens_32": { - "block_sizes": [ - 32, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_32": { - "block_sizes": [ - 32, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_32": { "block_sizes": [ 32, 16 @@ -15967,8 +15881,8 @@ "", "" ], - "num_warps": 1, - "num_stages": 2, + "num_warps": 4, + "num_stages": 1, "indexing": [ "pointer", "pointer", @@ -15977,10 +15891,10 @@ ], "pid_type": "flat" }, - "intermediate_8192_numtokens_32": { + "intermediate_2880_numtokens_32": { "block_sizes": [ - 32, - 128 + 4, + 256 ], "loop_orders": [ [ @@ -16010,10 +15924,10 @@ "load_eviction_policies": [ "", "", - "last" + "" ], - "num_warps": 2, - "num_stages": 2, + "num_warps": 4, + "num_stages": 1, "indexing": [ "pointer", "pointer", @@ -16022,22 +15936,67 @@ ], "pid_type": "flat" }, - "intermediate_11008_numtokens_32": { + "intermediate_4096_numtokens_32": { "block_sizes": [ - 16, - 8 + 4, + 4096 ], "loop_orders": [ [ - 1, - 0 + 0, + 1 ] ], "flatten_loops": [ true ], "l2_groupings": [ - 1 + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "" + ], + "num_warps": 32, + "num_stages": 5, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_32": { + "block_sizes": [ + 4, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 ], "range_unroll_factors": [ 0 @@ -16054,23 +16013,23 @@ ], "load_eviction_policies": [ "", - "", + "last", "" ], - "num_warps": 16, - "num_stages": 1, + "num_warps": 2, + "num_stages": 3, "indexing": [ - "pointer", - "pointer", + "tensor_descriptor", + "tensor_descriptor", "pointer", "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_14336_numtokens_32": { + "intermediate_11008_numtokens_32": { "block_sizes": [ - 32, - 8 + 2, + 2048 ], "loop_orders": [ [ @@ -16103,7 +16062,7 @@ "" ], "num_warps": 8, - "num_stages": 2, + "num_stages": 1, "indexing": [ "pointer", "pointer", @@ -16112,9 +16071,54 @@ ], "pid_type": "flat" }, + "intermediate_14336_numtokens_32": { + "block_sizes": [ + 1, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "" + ], + "num_warps": 4, + "num_stages": 3, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, "intermediate_2048_numtokens_40": { "block_sizes": [ - 64, + 32, 32 ], "loop_orders": [ @@ -16127,7 +16131,7 @@ true ], "l2_groupings": [ - 2 + 1 ], "range_unroll_factors": [ 0 @@ -16147,7 +16151,7 @@ "", "" ], - "num_warps": 2, + "num_warps": 4, "num_stages": 1, "indexing": [ "pointer", @@ -16159,20 +16163,20 @@ }, "intermediate_2880_numtokens_40": { "block_sizes": [ - 64, - 32 + 1, + 4096 ], "loop_orders": [ [ - 0, - 1 + 1, + 0 ] ], "flatten_loops": [ - true + false ], "l2_groupings": [ - 1 + 32 ], "range_unroll_factors": [ 0 @@ -16189,13 +16193,13 @@ ], "load_eviction_policies": [ "", - "", + "last", "last" ], "num_warps": 8, - "num_stages": 2, + "num_stages": 4, "indexing": [ - "pointer", + "tensor_descriptor", "pointer", "pointer", "pointer" @@ -16205,12 +16209,12 @@ "intermediate_4096_numtokens_40": { "block_sizes": [ 32, - 32 + 64 ], "loop_orders": [ [ - 1, - 0 + 0, + 1 ] ], "flatten_loops": [ @@ -16240,7 +16244,7 @@ "num_warps": 8, "num_stages": 1, "indexing": [ - "tensor_descriptor", + "pointer", "pointer", "pointer", "pointer" @@ -16249,8 +16253,8 @@ }, "intermediate_8192_numtokens_40": { "block_sizes": [ - 64, - 128 + 2, + 256 ], "loop_orders": [ [ @@ -16262,7 +16266,7 @@ true ], "l2_groupings": [ - 2 + 16 ], "range_unroll_factors": [ 0 @@ -16279,23 +16283,23 @@ ], "load_eviction_policies": [ "", - "", - "" + "last", + "last" ], - "num_warps": 4, - "num_stages": 1, + "num_warps": 2, + "num_stages": 2, "indexing": [ "pointer", "pointer", "pointer", - "pointer" + "tensor_descriptor" ], "pid_type": "flat" }, "intermediate_11008_numtokens_40": { "block_sizes": [ - 64, - 32 + 16, + 256 ], "loop_orders": [ [ @@ -16327,7 +16331,7 @@ "", "" ], - "num_warps": 4, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -16338,6 +16342,53 @@ "pid_type": "flat" }, "intermediate_14336_numtokens_40": { + "block_sizes": [ + 4, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 1 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "last" + ], + "num_warps": 16, + "num_stages": 5, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "persistent_interleaved", + "num_sm_multiplier": 32, + "maxnreg": 32 + }, + "intermediate_2048_numtokens_48": { "block_sizes": [ 32, 32 @@ -16372,7 +16423,7 @@ "", "" ], - "num_warps": 2, + "num_warps": 4, "num_stages": 1, "indexing": [ "pointer", @@ -16382,55 +16433,10 @@ ], "pid_type": "flat" }, - "intermediate_2048_numtokens_48": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "first", - "" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, "intermediate_2880_numtokens_48": { "block_sizes": [ - 8, - 8 + 16, + 64 ], "loop_orders": [ [ @@ -16462,12 +16468,12 @@ "", "" ], - "num_warps": 1, + "num_warps": 4, "num_stages": 1, "indexing": [ "pointer", "pointer", - "tensor_descriptor", + "pointer", "pointer" ], "pid_type": "flat" @@ -16507,52 +16513,7 @@ "", "" ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_48": { - "block_sizes": [ - 64, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -16562,10 +16523,10 @@ ], "pid_type": "flat" }, - "intermediate_11008_numtokens_48": { + "intermediate_8192_numtokens_48": { "block_sizes": [ - 16, - 256 + 1, + 4096 ], "loop_orders": [ [ @@ -16593,24 +16554,24 @@ null ], "load_eviction_policies": [ + "first", "", - "", - "" + "last" ], - "num_warps": 32, - "num_stages": 1, + "num_warps": 4, + "num_stages": 2, "indexing": [ "pointer", - "pointer", - "pointer", - "tensor_descriptor" + "tensor_descriptor", + "tensor_descriptor", + "pointer" ], "pid_type": "flat" }, - "intermediate_14336_numtokens_48": { + "intermediate_11008_numtokens_48": { "block_sizes": [ - 64, - 4 + 16, + 256 ], "loop_orders": [ [ @@ -16642,10 +16603,55 @@ "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ - "tensor_descriptor", + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_48": { + "block_sizes": [ + 32, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", "pointer", "pointer", "pointer" @@ -16654,52 +16660,7 @@ }, "intermediate_2048_numtokens_56": { "block_sizes": [ - 2, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_56": { - "block_sizes": [ - 8, + 32, 32 ], "loop_orders": [ @@ -16742,10 +16703,10 @@ ], "pid_type": "flat" }, - "intermediate_4096_numtokens_56": { + "intermediate_2880_numtokens_56": { "block_sizes": [ - 32, - 4 + 1, + 2048 ], "loop_orders": [ [ @@ -16774,15 +16735,60 @@ ], "load_eviction_policies": [ "", - "first", + "", "" ], - "num_warps": 4, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", "pointer", - "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_56": { + "block_sizes": [ + 32, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", "pointer" ], "pid_type": "flat" @@ -16822,7 +16828,7 @@ "", "" ], - "num_warps": 1, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -16834,8 +16840,8 @@ }, "intermediate_11008_numtokens_56": { "block_sizes": [ - 32, - 8 + 1, + 8192 ], "loop_orders": [ [ @@ -16864,10 +16870,10 @@ ], "load_eviction_policies": [ "", - "first", - "last" + "", + "" ], - "num_warps": 1, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -16878,144 +16884,9 @@ "pid_type": "flat" }, "intermediate_14336_numtokens_56": { - "block_sizes": [ - 64, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_64": { - "block_sizes": [ - 16, - 128 - ], - "loop_orders": [ - [ - 1, - 0 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_64": { - "block_sizes": [ - 4, - 64 - ], - "loop_orders": [ - [ - 1, - 0 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_64": { "block_sizes": [ 2, - 16 + 4096 ], "loop_orders": [ [ @@ -17024,10 +16895,10 @@ ] ], "flatten_loops": [ - true + false ], "l2_groupings": [ - 1 + 32 ], "range_unroll_factors": [ 0 @@ -17044,608 +16915,23 @@ ], "load_eviction_policies": [ "first", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_64": { - "block_sizes": [ - 8, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_64": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", + "first", "" ], "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_64": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_72": { - "block_sizes": [ - 4, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "last" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_72": { - "block_sizes": [ - 64, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "first" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_72": { - "block_sizes": [ - 64, - 16 - ], - "loop_orders": [ - [ - 1, - 0 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "last", - "", - "" - ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_72": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_72": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_72": { - "block_sizes": [ - 128, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_80": { - "block_sizes": [ - 32, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_80": { - "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_80": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_80": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 4 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, + "num_stages": 4, "indexing": [ "tensor_descriptor", "pointer", - "pointer", - "pointer" + "tensor_descriptor", + "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_11008_numtokens_80": { + "intermediate_2048_numtokens_64": { "block_sizes": [ 64, - 8 + 32 ], "loop_orders": [ [ @@ -17687,9 +16973,189 @@ ], "pid_type": "flat" }, - "intermediate_14336_numtokens_80": { + "intermediate_2880_numtokens_64": { "block_sizes": [ - 32, + 1, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_64": { + "block_sizes": [ + 64, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_64": { + "block_sizes": [ + 64, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_64": { + "block_sizes": [ + 1, + 8192 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_64": { + "block_sizes": [ + 16, 512 ], "loop_orders": [ @@ -17702,7 +17168,7 @@ true ], "l2_groupings": [ - 2 + 16 ], "range_unroll_factors": [ 0 @@ -17722,156 +17188,21 @@ "first", "" ], - "num_warps": 16, + "num_warps": 4, "num_stages": 1, "indexing": [ "pointer", "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_88": { - "block_sizes": [ - 32, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_88": { - "block_sizes": [ - 16, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ "tensor_descriptor", - "pointer", - "pointer", "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_4096_numtokens_88": { + "intermediate_2048_numtokens_72": { "block_sizes": [ 64, 32 ], - "loop_orders": [ - [ - 1, - 0 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "tensor_descriptor", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_88": { - "block_sizes": [ - 128, - 64 - ], "loop_orders": [ [ 0, @@ -17897,101 +17228,11 @@ "range_flattens": [ null ], - "load_eviction_policies": [ - "first", - "first", - "last" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_88": { - "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 4 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], "load_eviction_policies": [ "", "", "" ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_88": { - "block_sizes": [ - 16, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "first", - "last", - "" - ], "num_warps": 8, "num_stages": 1, "indexing": [ @@ -18002,248 +17243,68 @@ ], "pid_type": "flat" }, - "intermediate_2048_numtokens_96": { + "intermediate_2880_numtokens_72": { "block_sizes": [ - 128, + 32, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_72": { + "block_sizes": [ + 1, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ 4 ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_96": { - "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_96": { - "block_sizes": [ - 16, - 256 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 3, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_96": { - "block_sizes": [ - 64, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "first", - "", - "last" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_96": { - "block_sizes": [ - 64, - 256 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "first", - "", - "" - ], - "num_warps": 2, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_96": { - "block_sizes": [ - 32, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], "range_unroll_factors": [ 0 ], @@ -18262,20 +17323,20 @@ "", "first" ], - "num_warps": 1, - "num_stages": 1, + "num_warps": 16, + "num_stages": 2, "indexing": [ - "pointer", - "pointer", "tensor_descriptor", - "pointer" + "pointer", + "pointer", + "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_2048_numtokens_104": { + "intermediate_8192_numtokens_72": { "block_sizes": [ - 32, - 8 + 64, + 128 ], "loop_orders": [ [ @@ -18307,17 +17368,197 @@ "", "" ], - "num_warps": 16, - "num_stages": 2, + "num_warps": 8, + "num_stages": 1, "indexing": [ - "tensor_descriptor", + "pointer", "pointer", "pointer", "pointer" ], "pid_type": "flat" }, - "intermediate_2880_numtokens_104": { + "intermediate_11008_numtokens_72": { + "block_sizes": [ + 4, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 32, + "num_stages": 5, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_72": { + "block_sizes": [ + 32, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "", + "last" + ], + "num_warps": 8, + "num_stages": 3, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_80": { + "block_sizes": [ + 64, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_80": { + "block_sizes": [ + 32, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_80": { "block_sizes": [ 64, 64 @@ -18352,7 +17593,772 @@ "", "" ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_80": { + "block_sizes": [ + 4, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "last" + ], + "num_warps": 16, + "num_stages": 2, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_80": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "" + ], + "num_warps": 16, + "num_stages": 2, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_80": { + "block_sizes": [ + 2, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "" + ], + "num_warps": 4, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_88": { + "block_sizes": [ + 64, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_88": { + "block_sizes": [ + 8, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_88": { + "block_sizes": [ + 64, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_88": { + "block_sizes": [ + 64, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_88": { + "block_sizes": [ + 16, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "first", + "" + ], "num_warps": 32, + "num_stages": 2, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_88": { + "block_sizes": [ + 4, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 32, + "num_stages": 5, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_96": { + "block_sizes": [ + 64, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_96": { + "block_sizes": [ + 32, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_96": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 4, + "num_stages": 6, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_96": { + "block_sizes": [ + 64, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_96": { + "block_sizes": [ + 1, + 2048 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "" + ], + "num_warps": 32, + "num_stages": 2, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_96": { + "block_sizes": [ + 4, + 4096 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "first", + "" + ], + "num_warps": 4, + "num_stages": 1, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_104": { + "block_sizes": [ + 64, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_104": { + "block_sizes": [ + 8, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -18364,8 +18370,8 @@ }, "intermediate_4096_numtokens_104": { "block_sizes": [ - 32, - 32 + 64, + 64 ], "loop_orders": [ [ @@ -18374,7 +18380,7 @@ ] ], "flatten_loops": [ - false + true ], "l2_groupings": [ 1 @@ -18408,9 +18414,99 @@ "pid_type": "flat" }, "intermediate_8192_numtokens_104": { + "block_sizes": [ + 64, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_104": { + "block_sizes": [ + 2, + 8192 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "first" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_104": { "block_sizes": [ 8, - 8 + 512 ], "loop_orders": [ [ @@ -18437,6 +18533,51 @@ "range_flattens": [ null ], + "load_eviction_policies": [ + "last", + "last", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_112": { + "block_sizes": [ + 64, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], "load_eviction_policies": [ "", "", @@ -18448,14 +18589,14 @@ "pointer", "pointer", "pointer", - "tensor_descriptor" + "pointer" ], "pid_type": "flat" }, - "intermediate_11008_numtokens_104": { + "intermediate_2880_numtokens_112": { "block_sizes": [ - 128, - 16 + 2, + 2048 ], "loop_orders": [ [ @@ -18487,20 +18628,20 @@ "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", - "tensor_descriptor", + "pointer", "pointer", "pointer" ], "pid_type": "flat" }, - "intermediate_14336_numtokens_104": { + "intermediate_4096_numtokens_112": { "block_sizes": [ - 32, - 16 + 64, + 64 ], "loop_orders": [ [ @@ -18532,20 +18673,20 @@ "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", "pointer", "pointer", - "tensor_descriptor" + "pointer" ], "pid_type": "flat" }, - "intermediate_2048_numtokens_112": { + "intermediate_8192_numtokens_112": { "block_sizes": [ - 32, - 1024 + 4, + 512 ], "loop_orders": [ [ @@ -18557,7 +18698,7 @@ true ], "l2_groupings": [ - 1 + 16 ], "range_unroll_factors": [ 0 @@ -18574,203 +18715,23 @@ ], "load_eviction_policies": [ "first", - "", - "" - ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_112": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_112": { - "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ "last", - "", - "" + "last" ], "num_warps": 16, - "num_stages": 1, + "num_stages": 3, "indexing": [ "pointer", "pointer", - "pointer", + "tensor_descriptor", "pointer" ], "pid_type": "flat" }, - "intermediate_8192_numtokens_112": { - "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, "intermediate_11008_numtokens_112": { "block_sizes": [ - 16, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_112": { - "block_sizes": [ - 32, - 8 + 4, + 1024 ], "loop_orders": [ [ @@ -18799,23 +18760,23 @@ ], "load_eviction_policies": [ "", - "", - "" + "last", + "last" ], - "num_warps": 1, - "num_stages": 2, + "num_warps": 32, + "num_stages": 3, "indexing": [ "pointer", "pointer", - "pointer", - "pointer" + "tensor_descriptor", + "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_2048_numtokens_120": { + "intermediate_14336_numtokens_112": { "block_sizes": [ - 32, - 64 + 64, + 256 ], "loop_orders": [ [ @@ -18847,7 +18808,7 @@ "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -18857,19 +18818,64 @@ ], "pid_type": "flat" }, - "intermediate_2880_numtokens_120": { + "intermediate_2048_numtokens_120": { "block_sizes": [ - 32, - 16 + 8, + 256 ], "loop_orders": [ [ - 1, - 0 + 0, + 1 ] ], "flatten_loops": [ - false + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "" + ], + "num_warps": 4, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_120": { + "block_sizes": [ + 2, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true ], "l2_groupings": [ 1 @@ -18892,7 +18898,7 @@ "", "" ], - "num_warps": 4, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -18904,8 +18910,8 @@ }, "intermediate_4096_numtokens_120": { "block_sizes": [ - 32, - 16 + 64, + 64 ], "loop_orders": [ [ @@ -18917,7 +18923,7 @@ true ], "l2_groupings": [ - 2 + 1 ], "range_unroll_factors": [ 0 @@ -18937,7 +18943,7 @@ "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -18950,97 +18956,7 @@ "intermediate_8192_numtokens_120": { "block_sizes": [ 64, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_120": { - "block_sizes": [ - 64, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_120": { - "block_sizes": [ - 128, - 32 + 128 ], "loop_orders": [ [ @@ -19082,10 +18998,100 @@ ], "pid_type": "flat" }, - "intermediate_2048_numtokens_128": { + "intermediate_11008_numtokens_120": { + "block_sizes": [ + 1, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 1, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_120": { "block_sizes": [ 32, - 64 + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "first", + "last" + ], + "num_warps": 16, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_128": { + "block_sizes": [ + 128, + 16 ], "loop_orders": [ [ @@ -19117,8 +19123,8 @@ "", "" ], - "num_warps": 2, - "num_stages": 2, + "num_warps": 8, + "num_stages": 1, "indexing": [ "pointer", "pointer", @@ -19129,8 +19135,8 @@ }, "intermediate_2880_numtokens_128": { "block_sizes": [ - 128, - 64 + 2, + 2048 ], "loop_orders": [ [ @@ -19159,10 +19165,10 @@ ], "load_eviction_policies": [ "", - "last", + "", "" ], - "num_warps": 4, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -19202,101 +19208,11 @@ "range_flattens": [ null ], - "load_eviction_policies": [ - "", - "", - "last" - ], - "num_warps": 32, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_128": { - "block_sizes": [ - 32, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], "load_eviction_policies": [ "", "", "" ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_128": { - "block_sizes": [ - 128, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "last" - ], "num_warps": 8, "num_stages": 1, "indexing": [ @@ -19307,99 +19223,9 @@ ], "pid_type": "flat" }, - "intermediate_14336_numtokens_128": { - "block_sizes": [ - 16, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 3, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_136": { + "intermediate_8192_numtokens_128": { "block_sizes": [ 128, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 3, - "indexing": [ - "tensor_descriptor", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_136": { - "block_sizes": [ - 8, 64 ], "loop_orders": [ @@ -19429,7 +19255,187 @@ ], "load_eviction_policies": [ "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_128": { + "block_sizes": [ + 2, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ "first", + "", + "last" + ], + "num_warps": 2, + "num_stages": 1, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_128": { + "block_sizes": [ + 4, + 4096 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_136": { + "block_sizes": [ + 128, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_136": { + "block_sizes": [ + 64, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", "" ], "num_warps": 8, @@ -19444,8 +19450,8 @@ }, "intermediate_4096_numtokens_136": { "block_sizes": [ - 32, - 16 + 128, + 64 ], "loop_orders": [ [ @@ -19477,7 +19483,7 @@ "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -19489,8 +19495,8 @@ }, "intermediate_8192_numtokens_136": { "block_sizes": [ - 32, - 128 + 2, + 512 ], "loop_orders": [ [ @@ -19502,7 +19508,7 @@ true ], "l2_groupings": [ - 1 + 32 ], "range_unroll_factors": [ 0 @@ -19518,204 +19524,24 @@ null ], "load_eviction_policies": [ - "", + "first", "", "" ], - "num_warps": 4, - "num_stages": 1, + "num_warps": 32, + "num_stages": 7, "indexing": [ + "tensor_descriptor", "pointer", "pointer", - "pointer", - "pointer" + "tensor_descriptor" ], "pid_type": "flat" }, "intermediate_11008_numtokens_136": { "block_sizes": [ - 16, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 3, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_136": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_144": { - "block_sizes": [ - 8, - 16 - ], - "loop_orders": [ - [ - 1, - 0 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_144": { - "block_sizes": [ - 256, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_144": { - "block_sizes": [ - 128, - 32 + 4, + 8192 ], "loop_orders": [ [ @@ -19727,97 +19553,7 @@ false ], "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_144": { - "block_sizes": [ - 128, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_144": { - "block_sizes": [ - 32, - 4 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 + 8 ], "range_unroll_factors": [ 0 @@ -19835,10 +19571,10 @@ "load_eviction_policies": [ "", "last", - "" + "last" ], "num_warps": 8, - "num_stages": 1, + "num_stages": 4, "indexing": [ "pointer", "pointer", @@ -19847,11 +19583,101 @@ ], "pid_type": "flat" }, - "intermediate_14336_numtokens_144": { + "intermediate_14336_numtokens_136": { "block_sizes": [ - 32, + 4, + 16384 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ 8 ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "last" + ], + "num_warps": 32, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_144": { + "block_sizes": [ + 1, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "first" + ], + "num_warps": 16, + "num_stages": 7, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_144": { + "block_sizes": [ + 64, + 64 + ], "loop_orders": [ [ 0, @@ -19877,25 +19703,790 @@ "range_flattens": [ null ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_144": { + "block_sizes": [ + 128, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_144": { + "block_sizes": [ + 1, + 2048 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "first" + ], + "num_warps": 1, + "num_stages": 4, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_144": { + "block_sizes": [ + 256, + 16 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "first" + ], + "num_warps": 16, + "num_stages": 2, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_144": { + "block_sizes": [ + 64, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "", + "last" + ], + "num_warps": 16, + "num_stages": 8, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_152": { + "block_sizes": [ + 4, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "first" + ], + "num_warps": 8, + "num_stages": 7, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_152": { + "block_sizes": [ + 64, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_152": { + "block_sizes": [ + 128, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_152": { + "block_sizes": [ + 64, + 16 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "first" + ], + "num_warps": 1, + "num_stages": 2, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_152": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "first" + ], + "num_warps": 4, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_152": { + "block_sizes": [ + 2, + 16384 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "" + ], + "num_warps": 16, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_160": { + "block_sizes": [ + 4, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "last" + ], + "num_warps": 1, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_160": { + "block_sizes": [ + 64, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_160": { + "block_sizes": [ + 128, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_160": { + "block_sizes": [ + 64, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 32, + "num_stages": 4, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_160": { + "block_sizes": [ + 128, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "first", + "last" + ], + "num_warps": 32, + "num_stages": 8, + "indexing": [ + "pointer", + "pointer", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_160": { + "block_sizes": [ + 1, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], "load_eviction_policies": [ "", "", "first" ], - "num_warps": 1, - "num_stages": 2, + "num_warps": 8, + "num_stages": 8, "indexing": [ - "pointer", "tensor_descriptor", "pointer", - "pointer" + "tensor_descriptor", + "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_2048_numtokens_152": { + "intermediate_2048_numtokens_168": { "block_sizes": [ - 32, - 8 + 4, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "", + "last" + ], + "num_warps": 32, + "num_stages": 8, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_168": { + "block_sizes": [ + 8, + 512 ], "loop_orders": [ [ @@ -19927,7 +20518,7 @@ "", "" ], - "num_warps": 16, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -19937,9 +20528,9 @@ ], "pid_type": "flat" }, - "intermediate_2880_numtokens_152": { + "intermediate_4096_numtokens_168": { "block_sizes": [ - 16, + 128, 64 ], "loop_orders": [ @@ -19972,141 +20563,6 @@ "", "" ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_152": { - "block_sizes": [ - 64, - 4 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_152": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "first", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_152": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], "num_warps": 8, "num_stages": 1, "indexing": [ @@ -20117,460 +20573,10 @@ ], "pid_type": "flat" }, - "intermediate_14336_numtokens_152": { - "block_sizes": [ - 64, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_160": { - "block_sizes": [ - 32, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_160": { - "block_sizes": [ - 128, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_160": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_160": { - "block_sizes": [ - 64, - 4 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "first", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_160": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_160": { - "block_sizes": [ - 128, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "tensor_descriptor", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_168": { - "block_sizes": [ - 128, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_168": { - "block_sizes": [ - 32, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_168": { - "block_sizes": [ - 64, - 32 - ], - "loop_orders": [ - [ - 1, - 0 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "last" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, "intermediate_8192_numtokens_168": { "block_sizes": [ - 64, - 8 + 128, + 128 ], "loop_orders": [ [ @@ -20615,7 +20621,7 @@ "intermediate_11008_numtokens_168": { "block_sizes": [ 64, - 4 + 256 ], "loop_orders": [ [ @@ -20647,7 +20653,7 @@ "", "" ], - "num_warps": 16, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -20659,8 +20665,8 @@ }, "intermediate_14336_numtokens_168": { "block_sizes": [ - 32, - 512 + 64, + 32 ], "loop_orders": [ [ @@ -20672,7 +20678,7 @@ true ], "l2_groupings": [ - 2 + 64 ], "range_unroll_factors": [ 0 @@ -20688,68 +20694,23 @@ null ], "load_eviction_policies": [ - "", - "", - "" + "first", + "last", + "first" ], "num_warps": 2, - "num_stages": 1, + "num_stages": 6, "indexing": [ + "tensor_descriptor", "pointer", - "pointer", - "pointer", - "tensor_descriptor" + "tensor_descriptor", + "pointer" ], "pid_type": "flat" }, "intermediate_2048_numtokens_176": { "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 1, - 0 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_176": { - "block_sizes": [ - 32, + 128, 32 ], "loop_orders": [ @@ -20782,7 +20743,7 @@ "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -20792,10 +20753,10 @@ ], "pid_type": "flat" }, - "intermediate_4096_numtokens_176": { + "intermediate_2880_numtokens_176": { "block_sizes": [ - 4, - 8 + 16, + 256 ], "loop_orders": [ [ @@ -20822,15 +20783,60 @@ "range_flattens": [ null ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_176": { + "block_sizes": [ + 128, + 4 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], "load_eviction_policies": [ "", "", "first" ], - "num_warps": 8, - "num_stages": 1, + "num_warps": 4, + "num_stages": 6, "indexing": [ - "pointer", + "tensor_descriptor", "pointer", "pointer", "pointer" @@ -20839,9 +20845,54 @@ }, "intermediate_8192_numtokens_176": { "block_sizes": [ - 8, + 1, + 8192 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ 16 ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "first" + ], + "num_warps": 16, + "num_stages": 5, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_176": { + "block_sizes": [ + 64, + 256 + ], "loop_orders": [ [ 0, @@ -20872,52 +20923,7 @@ "", "" ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_176": { - "block_sizes": [ - 8, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -20929,8 +20935,8 @@ }, "intermediate_14336_numtokens_176": { "block_sizes": [ - 8, - 32 + 128, + 256 ], "loop_orders": [ [ @@ -20962,7 +20968,7 @@ "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -20974,53 +20980,8 @@ }, "intermediate_2048_numtokens_184": { "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_184": { - "block_sizes": [ - 8, - 16 + 2, + 256 ], "loop_orders": [ [ @@ -21032,7 +20993,7 @@ true ], "l2_groupings": [ - 1 + 2 ], "range_unroll_factors": [ 0 @@ -21053,19 +21014,64 @@ "" ], "num_warps": 16, - "num_stages": 2, + "num_stages": 6, "indexing": [ "pointer", "pointer", "pointer", - "tensor_descriptor" + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_184": { + "block_sizes": [ + 64, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" ], "pid_type": "flat" }, "intermediate_4096_numtokens_184": { "block_sizes": [ - 32, - 8 + 128, + 64 ], "loop_orders": [ [ @@ -21109,7 +21115,7 @@ }, "intermediate_8192_numtokens_184": { "block_sizes": [ - 8, + 64, 64 ], "loop_orders": [ @@ -21122,7 +21128,52 @@ true ], "l2_groupings": [ - 2 + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 32, + "num_stages": 8, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_184": { + "block_sizes": [ + 64, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 ], "range_unroll_factors": [ 0 @@ -21152,203 +21203,23 @@ ], "pid_type": "flat" }, - "intermediate_11008_numtokens_184": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, "intermediate_14336_numtokens_184": { "block_sizes": [ - 16, + 64, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ 64 ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "last" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_192": { - "block_sizes": [ - 32, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_192": { - "block_sizes": [ - 8, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_192": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 4 - ], "range_unroll_factors": [ 0 ], @@ -21365,21 +21236,21 @@ "load_eviction_policies": [ "first", "", - "" + "last" ], - "num_warps": 16, - "num_stages": 2, + "num_warps": 8, + "num_stages": 3, "indexing": [ + "tensor_descriptor", "pointer", "pointer", - "pointer", - "pointer" + "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_8192_numtokens_192": { + "intermediate_2048_numtokens_192": { "block_sizes": [ - 4, + 128, 32 ], "loop_orders": [ @@ -21392,7 +21263,7 @@ true ], "l2_groupings": [ - 2 + 1 ], "range_unroll_factors": [ 0 @@ -21422,9 +21293,144 @@ ], "pid_type": "flat" }, - "intermediate_11008_numtokens_192": { + "intermediate_2880_numtokens_192": { + "block_sizes": [ + 64, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_192": { + "block_sizes": [ + 8, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "first", + "first" + ], + "num_warps": 16, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_192": { "block_sizes": [ 32, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "" + ], + "num_warps": 32, + "num_stages": 1, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_192": { + "block_sizes": [ + 16, 256 ], "loop_orders": [ @@ -21455,22 +21461,22 @@ "load_eviction_policies": [ "", "", - "" + "first" ], - "num_warps": 2, - "num_stages": 2, + "num_warps": 32, + "num_stages": 7, "indexing": [ + "tensor_descriptor", "pointer", - "pointer", - "pointer", - "pointer" + "tensor_descriptor", + "tensor_descriptor" ], "pid_type": "flat" }, "intermediate_14336_numtokens_192": { "block_sizes": [ - 8, - 16 + 128, + 256 ], "loop_orders": [ [ @@ -21502,11 +21508,11 @@ "", "" ], - "num_warps": 4, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", - "tensor_descriptor", + "pointer", "pointer", "pointer" ], @@ -21514,7 +21520,7 @@ }, "intermediate_2048_numtokens_200": { "block_sizes": [ - 32, + 128, 32 ], "loop_orders": [ @@ -21527,7 +21533,7 @@ true ], "l2_groupings": [ - 2 + 1 ], "range_unroll_factors": [ 0 @@ -21547,7 +21553,7 @@ "", "" ], - "num_warps": 1, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -21559,8 +21565,8 @@ }, "intermediate_2880_numtokens_200": { "block_sizes": [ - 32, - 32 + 8, + 512 ], "loop_orders": [ [ @@ -21589,10 +21595,10 @@ ], "load_eviction_policies": [ "", - "last", + "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -21604,8 +21610,8 @@ }, "intermediate_4096_numtokens_200": { "block_sizes": [ - 64, - 32 + 4, + 512 ], "loop_orders": [ [ @@ -21617,7 +21623,7 @@ true ], "l2_groupings": [ - 1 + 16 ], "range_unroll_factors": [ 0 @@ -21633,24 +21639,24 @@ null ], "load_eviction_policies": [ + "last", "", - "", - "" + "first" ], - "num_warps": 2, + "num_warps": 1, "num_stages": 1, "indexing": [ "pointer", "pointer", "pointer", - "pointer" + "tensor_descriptor" ], "pid_type": "flat" }, "intermediate_8192_numtokens_200": { "block_sizes": [ - 32, - 32 + 128, + 128 ], "loop_orders": [ [ @@ -21682,7 +21688,7 @@ "", "" ], - "num_warps": 1, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -21694,17 +21700,17 @@ }, "intermediate_11008_numtokens_200": { "block_sizes": [ - 8, - 32 + 1, + 1024 ], "loop_orders": [ [ - 0, - 1 + 1, + 0 ] ], "flatten_loops": [ - true + false ], "l2_groupings": [ 1 @@ -21723,15 +21729,15 @@ null ], "load_eviction_policies": [ - "", - "", - "" + "first", + "first", + "first" ], - "num_warps": 1, - "num_stages": 1, + "num_warps": 32, + "num_stages": 3, "indexing": [ "pointer", - "pointer", + "tensor_descriptor", "pointer", "pointer" ], @@ -21740,7 +21746,7 @@ "intermediate_14336_numtokens_200": { "block_sizes": [ 16, - 8 + 128 ], "loop_orders": [ [ @@ -21752,7 +21758,7 @@ true ], "l2_groupings": [ - 1 + 64 ], "range_unroll_factors": [ 0 @@ -21772,154 +21778,19 @@ "", "first" ], - "num_warps": 4, - "num_stages": 1, + "num_warps": 32, + "num_stages": 6, "indexing": [ "pointer", "pointer", - "pointer", + "tensor_descriptor", "pointer" ], "pid_type": "flat" }, "intermediate_2048_numtokens_208": { "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_208": { - "block_sizes": [ - 64, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_208": { - "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_208": { - "block_sizes": [ - 256, + 128, 32 ], "loop_orders": [ @@ -21952,20 +21823,20 @@ "", "" ], - "num_warps": 16, - "num_stages": 2, + "num_warps": 8, + "num_stages": 1, "indexing": [ - "tensor_descriptor", - "tensor_descriptor", + "pointer", + "pointer", "pointer", "pointer" ], "pid_type": "flat" }, - "intermediate_11008_numtokens_208": { + "intermediate_2880_numtokens_208": { "block_sizes": [ - 64, - 64 + 256, + 16 ], "loop_orders": [ [ @@ -21977,7 +21848,7 @@ true ], "l2_groupings": [ - 1 + 8 ], "range_unroll_factors": [ 0 @@ -21993,24 +21864,24 @@ null ], "load_eviction_policies": [ - "last", - "last", - "" + "first", + "", + "last" ], - "num_warps": 8, - "num_stages": 1, + "num_warps": 32, + "num_stages": 8, "indexing": [ "pointer", "pointer", - "pointer", + "tensor_descriptor", "pointer" ], "pid_type": "flat" }, - "intermediate_14336_numtokens_208": { + "intermediate_4096_numtokens_208": { "block_sizes": [ - 16, - 128 + 256, + 32 ], "loop_orders": [ [ @@ -22037,12 +21908,147 @@ "range_flattens": [ null ], + "load_eviction_policies": [ + "last", + "first", + "last" + ], + "num_warps": 4, + "num_stages": 2, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_208": { + "block_sizes": [ + 128, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], "load_eviction_policies": [ "", "", "" ], - "num_warps": 32, + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_208": { + "block_sizes": [ + 32, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "first", + "first" + ], + "num_warps": 8, + "num_stages": 5, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_208": { + "block_sizes": [ + 128, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -22064,10 +22070,100 @@ ] ], "flatten_loops": [ - false + true ], "l2_groupings": [ - 2 + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "last" + ], + "num_warps": 8, + "num_stages": 4, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_216": { + "block_sizes": [ + 4, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "first", + "first" + ], + "num_warps": 8, + "num_stages": 7, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_216": { + "block_sizes": [ + 128, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 ], "range_unroll_factors": [ 0 @@ -22087,7 +22183,7 @@ "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -22097,10 +22193,10 @@ ], "pid_type": "flat" }, - "intermediate_2880_numtokens_216": { + "intermediate_8192_numtokens_216": { "block_sizes": [ - 16, - 128 + 1, + 8192 ], "loop_orders": [ [ @@ -22127,22 +22223,292 @@ "range_flattens": [ null ], + "load_eviction_policies": [ + "", + "", + "last" + ], + "num_warps": 2, + "num_stages": 7, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_216": { + "block_sizes": [ + 1, + 16384 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "last" + ], + "num_warps": 4, + "num_stages": 4, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_216": { + "block_sizes": [ + 128, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_224": { + "block_sizes": [ + 32, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "first", + "first" + ], + "num_warps": 16, + "num_stages": 5, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_224": { + "block_sizes": [ + 4, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_224": { + "block_sizes": [ + 128, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_224": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], "load_eviction_policies": [ "", "last", - "" + "first" ], "num_warps": 32, "num_stages": 1, "indexing": [ - "pointer", + "tensor_descriptor", "pointer", "pointer", "pointer" ], "pid_type": "flat" }, - "intermediate_4096_numtokens_216": { + "intermediate_11008_numtokens_224": { "block_sizes": [ 32, 32 @@ -22157,143 +22523,8 @@ true ], "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_216": { - "block_sizes": [ - 16, 32 ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_216": { - "block_sizes": [ - 32, - 4 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "first" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_216": { - "block_sizes": [ - 64, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], "range_unroll_factors": [ 0 ], @@ -22312,20 +22543,20 @@ "", "last" ], - "num_warps": 32, - "num_stages": 1, + "num_warps": 2, + "num_stages": 3, "indexing": [ "pointer", "pointer", - "pointer", + "tensor_descriptor", "pointer" ], "pid_type": "flat" }, - "intermediate_2048_numtokens_224": { + "intermediate_14336_numtokens_224": { "block_sizes": [ - 32, - 16 + 1, + 2048 ], "loop_orders": [ [ @@ -22337,7 +22568,7 @@ false ], "l2_groupings": [ - 1 + 32 ], "range_unroll_factors": [ 0 @@ -22354,126 +22585,36 @@ ], "load_eviction_policies": [ "", - "last", + "", "" ], - "num_warps": 2, - "num_stages": 1, + "num_warps": 8, + "num_stages": 8, "indexing": [ "pointer", + "tensor_descriptor", "pointer", - "pointer", - "pointer" + "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_2880_numtokens_224": { + "intermediate_2048_numtokens_232": { "block_sizes": [ - 64, + 1, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ 64 ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_224": { - "block_sizes": [ - 64, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 4 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_224": { - "block_sizes": [ - 16, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], "range_unroll_factors": [ 0 ], @@ -22494,51 +22635,6 @@ ], "num_warps": 16, "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_224": { - "block_sizes": [ - 256, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 1, "indexing": [ "tensor_descriptor", "pointer", @@ -22547,9 +22643,9 @@ ], "pid_type": "flat" }, - "intermediate_14336_numtokens_224": { + "intermediate_2880_numtokens_232": { "block_sizes": [ - 32, + 256, 8 ], "loop_orders": [ @@ -22562,7 +22658,7 @@ true ], "l2_groupings": [ - 1 + 4 ], "range_unroll_factors": [ 0 @@ -22579,10 +22675,10 @@ ], "load_eviction_policies": [ "last", - "", - "first" + "first", + "last" ], - "num_warps": 4, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -22592,100 +22688,10 @@ ], "pid_type": "flat" }, - "intermediate_2048_numtokens_232": { - "block_sizes": [ - 16, - 8 - ], - "loop_orders": [ - [ - 1, - 0 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 4 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_232": { - "block_sizes": [ - 64, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, "intermediate_4096_numtokens_232": { "block_sizes": [ - 16, - 4 + 128, + 64 ], "loop_orders": [ [ @@ -22729,20 +22735,20 @@ }, "intermediate_8192_numtokens_232": { "block_sizes": [ - 32, - 32 + 256, + 8 ], "loop_orders": [ [ - 1, - 0 + 0, + 1 ] ], "flatten_loops": [ - false + true ], "l2_groupings": [ - 1 + 2 ], "range_unroll_factors": [ 0 @@ -22760,9 +22766,9 @@ "load_eviction_policies": [ "", "", - "" + "first" ], - "num_warps": 4, + "num_warps": 16, "num_stages": 1, "indexing": [ "pointer", @@ -22774,20 +22780,20 @@ }, "intermediate_11008_numtokens_232": { "block_sizes": [ - 16, - 32 + 4, + 1024 ], "loop_orders": [ [ - 0, - 1 + 1, + 0 ] ], "flatten_loops": [ - true + false ], "l2_groupings": [ - 1 + 32 ], "range_unroll_factors": [ 0 @@ -22803,24 +22809,24 @@ null ], "load_eviction_policies": [ - "", + "first", "last", - "" + "first" ], - "num_warps": 2, - "num_stages": 2, + "num_warps": 1, + "num_stages": 8, "indexing": [ "pointer", + "tensor_descriptor", "pointer", - "pointer", - "pointer" + "tensor_descriptor" ], "pid_type": "flat" }, "intermediate_14336_numtokens_232": { "block_sizes": [ - 32, - 8 + 8, + 4096 ], "loop_orders": [ [ @@ -22852,7 +22858,7 @@ "", "" ], - "num_warps": 2, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -22864,142 +22870,7 @@ }, "intermediate_2048_numtokens_240": { "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_240": { - "block_sizes": [ - 8, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_240": { - "block_sizes": [ - 16, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "tensor_descriptor", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_240": { - "block_sizes": [ - 32, + 64, 8 ], "loop_orders": [ @@ -23011,231 +22882,6 @@ "flatten_loops": [ true ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_240": { - "block_sizes": [ - 8, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_240": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_248": { - "block_sizes": [ - 16, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_248": { - "block_sizes": [ - 16, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_248": { - "block_sizes": [ - 256, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], "l2_groupings": [ 2 ], @@ -23255,9 +22901,54 @@ "load_eviction_policies": [ "first", "", + "last" + ], + "num_warps": 4, + "num_stages": 5, + "indexing": [ + "pointer", + "pointer", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_240": { + "block_sizes": [ + 4, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", "" ], - "num_warps": 16, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -23267,9 +22958,189 @@ ], "pid_type": "flat" }, - "intermediate_8192_numtokens_248": { + "intermediate_4096_numtokens_240": { "block_sizes": [ - 64, + 4, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "" + ], + "num_warps": 8, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_240": { + "block_sizes": [ + 1, + 1024 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "first", + "last" + ], + "num_warps": 4, + "num_stages": 8, + "indexing": [ + "pointer", + "pointer", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_240": { + "block_sizes": [ + 4, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "first" + ], + "num_warps": 32, + "num_stages": 7, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_240": { + "block_sizes": [ + 1, + 8192 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "last" + ], + "num_warps": 32, + "num_stages": 2, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_248": { + "block_sizes": [ + 128, 32 ], "loop_orders": [ @@ -23300,10 +23171,10 @@ "load_eviction_policies": [ "", "", - "first" + "" ], - "num_warps": 4, - "num_stages": 2, + "num_warps": 8, + "num_stages": 1, "indexing": [ "pointer", "pointer", @@ -23312,10 +23183,10 @@ ], "pid_type": "flat" }, - "intermediate_11008_numtokens_248": { + "intermediate_2880_numtokens_248": { "block_sizes": [ - 64, - 4 + 4, + 2048 ], "loop_orders": [ [ @@ -23343,11 +23214,146 @@ null ], "load_eviction_policies": [ - "last", + "", "", "" ], - "num_warps": 1, + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_248": { + "block_sizes": [ + 128, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_248": { + "block_sizes": [ + 256, + 16 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 2, + "num_stages": 4, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_248": { + "block_sizes": [ + 4, + 8192 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -23359,8 +23365,8 @@ }, "intermediate_14336_numtokens_248": { "block_sizes": [ - 64, - 256 + 8, + 4096 ], "loop_orders": [ [ @@ -23403,456 +23409,6 @@ "pid_type": "flat" }, "intermediate_2048_numtokens_272": { - "block_sizes": [ - 128, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_272": { - "block_sizes": [ - 8, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_272": { - "block_sizes": [ - 128, - 32 - ], - "loop_orders": [ - [ - 1, - 0 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_272": { - "block_sizes": [ - 128, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_272": { - "block_sizes": [ - 16, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_272": { - "block_sizes": [ - 64, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_288": { - "block_sizes": [ - 4, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "last", - "", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_288": { - "block_sizes": [ - 8, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "tensor_descriptor", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_288": { - "block_sizes": [ - 64, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_288": { - "block_sizes": [ - 128, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_288": { "block_sizes": [ 256, 32 @@ -23866,51 +23422,6 @@ "flatten_loops": [ true ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_288": { - "block_sizes": [ - 16, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], "l2_groupings": [ 1 ], @@ -23932,7 +23443,7 @@ "", "" ], - "num_warps": 1, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -23942,9 +23453,9 @@ ], "pid_type": "flat" }, - "intermediate_2048_numtokens_304": { + "intermediate_2880_numtokens_272": { "block_sizes": [ - 8, + 128, 64 ], "loop_orders": [ @@ -23977,96 +23488,6 @@ "", "" ], - "num_warps": 32, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_304": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_304": { - "block_sizes": [ - 128, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], "num_warps": 8, "num_stages": 1, "indexing": [ @@ -24077,22 +23498,22 @@ ], "pid_type": "flat" }, - "intermediate_8192_numtokens_304": { + "intermediate_4096_numtokens_272": { "block_sizes": [ - 8, - 64 + 1, + 4096 ], "loop_orders": [ [ - 0, - 1 + 1, + 0 ] ], "flatten_loops": [ - true + false ], "l2_groupings": [ - 1 + 64 ], "range_unroll_factors": [ 0 @@ -24110,157 +23531,22 @@ "load_eviction_policies": [ "", "last", - "" + "first" ], - "num_warps": 16, + "num_warps": 32, "num_stages": 1, "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_304": { - "block_sizes": [ - 64, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_304": { - "block_sizes": [ - 64, - 4 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_320": { - "block_sizes": [ - 128, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", + "tensor_descriptor", "pointer", "tensor_descriptor", - "pointer" + "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_2880_numtokens_320": { + "intermediate_8192_numtokens_272": { "block_sizes": [ - 64, - 32 + 8, + 512 ], "loop_orders": [ [ @@ -24272,7 +23558,97 @@ true ], "l2_groupings": [ - 1 + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "", + "last" + ], + "num_warps": 2, + "num_stages": 6, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_272": { + "block_sizes": [ + 8, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "", + "last" + ], + "num_warps": 4, + "num_stages": 8, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_272": { + "block_sizes": [ + 512, + 16 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 32 ], "range_unroll_factors": [ 0 @@ -24290,19 +23666,109 @@ "load_eviction_policies": [ "", "", - "" + "last" ], - "num_warps": 4, - "num_stages": 1, + "num_warps": 8, + "num_stages": 8, "indexing": [ "pointer", "pointer", "pointer", - "pointer" + "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_4096_numtokens_320": { + "intermediate_2048_numtokens_288": { + "block_sizes": [ + 64, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "last" + ], + "num_warps": 32, + "num_stages": 4, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_288": { + "block_sizes": [ + 8, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "last" + ], + "num_warps": 2, + "num_stages": 1, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_288": { "block_sizes": [ 512, 4 @@ -24317,7 +23783,97 @@ true ], "l2_groupings": [ - 1 + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "first" + ], + "num_warps": 1, + "num_stages": 2, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_288": { + "block_sizes": [ + 1, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "first", + "" + ], + "num_warps": 2, + "num_stages": 3, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_288": { + "block_sizes": [ + 1, + 8192 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 2 ], "range_unroll_factors": [ 0 @@ -24335,10 +23891,10 @@ "load_eviction_policies": [ "last", "", - "" + "last" ], - "num_warps": 1, - "num_stages": 2, + "num_warps": 8, + "num_stages": 3, "indexing": [ "pointer", "pointer", @@ -24347,10 +23903,55 @@ ], "pid_type": "flat" }, - "intermediate_8192_numtokens_320": { + "intermediate_14336_numtokens_288": { "block_sizes": [ - 64, - 128 + 1, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "last" + ], + "num_warps": 1, + "num_stages": 5, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_304": { + "block_sizes": [ + 256, + 32 ], "loop_orders": [ [ @@ -24392,11 +23993,103 @@ ], "pid_type": "flat" }, - "intermediate_11008_numtokens_320": { + "intermediate_2880_numtokens_304": { "block_sizes": [ - 32, + 1, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 2 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 2 + ], + "range_multi_buffers": [ + false + ], + "range_flattens": [ + true + ], + "load_eviction_policies": [ + "last", + "", + "last" + ], + "num_warps": 16, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "persistent_blocked", + "num_sm_multiplier": 2, + "maxnreg": 64 + }, + "intermediate_4096_numtokens_304": { + "block_sizes": [ + 16, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ 32 ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "last" + ], + "num_warps": 16, + "num_stages": 2, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_304": { + "block_sizes": [ + 1, + 1024 + ], "loop_orders": [ [ 0, @@ -24424,23 +24117,23 @@ ], "load_eviction_policies": [ "", - "", - "" + "last", + "last" ], - "num_warps": 8, - "num_stages": 1, + "num_warps": 32, + "num_stages": 4, "indexing": [ - "tensor_descriptor", "tensor_descriptor", "pointer", - "pointer" + "tensor_descriptor", + "tensor_descriptor" ], "pid_type": "flat" }, - "intermediate_14336_numtokens_320": { + "intermediate_11008_numtokens_304": { "block_sizes": [ 128, - 16 + 256 ], "loop_orders": [ [ @@ -24472,8 +24165,323 @@ "", "" ], - "num_warps": 32, + "num_warps": 8, "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_304": { + "block_sizes": [ + 4, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "" + ], + "num_warps": 16, + "num_stages": 6, + "indexing": [ + "pointer", + "pointer", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_320": { + "block_sizes": [ + 1, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "" + ], + "num_warps": 2, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_320": { + "block_sizes": [ + 128, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_320": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "last" + ], + "num_warps": 8, + "num_stages": 7, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_320": { + "block_sizes": [ + 1, + 8192 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "" + ], + "num_warps": 16, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_320": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "first", + "" + ], + "num_warps": 2, + "num_stages": 3, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_320": { + "block_sizes": [ + 8, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "first" + ], + "num_warps": 32, + "num_stages": 3, "indexing": [ "pointer", "tensor_descriptor", @@ -24484,7 +24492,7 @@ }, "intermediate_2048_numtokens_336": { "block_sizes": [ - 2, + 256, 32 ], "loop_orders": [ @@ -24522,3029 +24530,14 @@ "indexing": [ "pointer", "pointer", - "tensor_descriptor", + "pointer", "pointer" ], "pid_type": "flat" }, "intermediate_2880_numtokens_336": { - "block_sizes": [ - 8, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_336": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_336": { - "block_sizes": [ - 64, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "last" - ], - "num_warps": 8, - "num_stages": 2, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_336": { - "block_sizes": [ - 8, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "last", - "", - "" - ], - "num_warps": 2, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_336": { - "block_sizes": [ - 8, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_352": { - "block_sizes": [ - 32, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_352": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_352": { "block_sizes": [ 16, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "last" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_352": { - "block_sizes": [ - 64, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_352": { - "block_sizes": [ - 8, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_352": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "last" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_368": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_368": { - "block_sizes": [ - 8, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_368": { - "block_sizes": [ - 8, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_368": { - "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_368": { - "block_sizes": [ - 32, - 4 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_368": { - "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "first", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_384": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_384": { - "block_sizes": [ - 64, - 256 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "first" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_384": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "last" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_384": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_384": { - "block_sizes": [ - 8, - 256 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_384": { - "block_sizes": [ - 32, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_400": { - "block_sizes": [ - 64, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "first" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_400": { - "block_sizes": [ - 8, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_400": { - "block_sizes": [ - 64, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_400": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_400": { - "block_sizes": [ - 256, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "last" - ], - "num_warps": 32, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_400": { - "block_sizes": [ - 8, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_416": { - "block_sizes": [ - 128, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_416": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "first" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_416": { - "block_sizes": [ - 64, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "first", - "", - "last" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_416": { - "block_sizes": [ - 128, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_416": { - "block_sizes": [ - 64, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "first", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_416": { - "block_sizes": [ - 32, - 256 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_432": { - "block_sizes": [ - 16, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_432": { - "block_sizes": [ - 32, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_432": { - "block_sizes": [ - 16, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 4, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_432": { - "block_sizes": [ - 16, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 8, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_432": { - "block_sizes": [ - 16, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "tensor_descriptor", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_432": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 2, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_448": { - "block_sizes": [ - 4, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_448": { - "block_sizes": [ - 8, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_448": { - "block_sizes": [ - 4, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_448": { - "block_sizes": [ - 32, - 256 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "last", - "", - "" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_448": { - "block_sizes": [ - 16, - 256 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_448": { - "block_sizes": [ - 16, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "last", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_464": { - "block_sizes": [ - 32, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "last" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_464": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_464": { - "block_sizes": [ - 16, - 64 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_464": { - "block_sizes": [ - 8, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_464": { - "block_sizes": [ - 128, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "first" - ], - "num_warps": 16, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_464": { - "block_sizes": [ - 128, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "first", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_480": { - "block_sizes": [ - 4, - 16 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "last", - "", - "" - ], - "num_warps": 2, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_480": { - "block_sizes": [ - 4, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "first", - "first" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_480": { - "block_sizes": [ - 8, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "last", - "", - "" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_480": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_480": { - "block_sizes": [ - 64, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "last", - "", - "last" - ], - "num_warps": 1, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_480": { - "block_sizes": [ - 16, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 4 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "last", - "", - "" - ], - "num_warps": 8, - "num_stages": 2, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_496": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - false - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_496": { - "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "first", - "" - ], - "num_warps": 8, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_496": { - "block_sizes": [ - 16, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 4 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 2, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "tensor_descriptor", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_8192_numtokens_496": { - "block_sizes": [ - 32, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "tensor_descriptor" - ], - "pid_type": "flat" - }, - "intermediate_11008_numtokens_496": { - "block_sizes": [ - 32, - 128 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 4, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_14336_numtokens_496": { - "block_sizes": [ - 256, - 8 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 1, - "num_stages": 2, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2048_numtokens_512": { - "block_sizes": [ - 32, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 2 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "last" - ], - "num_warps": 32, - "num_stages": 1, - "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_2880_numtokens_512": { - "block_sizes": [ - 16, - 32 - ], - "loop_orders": [ - [ - 0, - 1 - ] - ], - "flatten_loops": [ - true - ], - "l2_groupings": [ - 1 - ], - "range_unroll_factors": [ - 0 - ], - "range_warp_specializes": [], - "range_num_stages": [ - 0 - ], - "range_multi_buffers": [ - null - ], - "range_flattens": [ - null - ], - "load_eviction_policies": [ - "", - "", - "" - ], - "num_warps": 16, - "num_stages": 1, - "indexing": [ - "pointer", - "tensor_descriptor", - "pointer", - "pointer" - ], - "pid_type": "flat" - }, - "intermediate_4096_numtokens_512": { - "block_sizes": [ - 128, 512 ], "loop_orders": [ @@ -27577,8 +24570,8 @@ "", "" ], - "num_warps": 16, - "num_stages": 2, + "num_warps": 8, + "num_stages": 1, "indexing": [ "pointer", "pointer", @@ -27587,9 +24580,54 @@ ], "pid_type": "flat" }, - "intermediate_8192_numtokens_512": { + "intermediate_4096_numtokens_336": { "block_sizes": [ - 32, + 16, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "", + "first" + ], + "num_warps": 2, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_336": { + "block_sizes": [ + 256, 128 ], "loop_orders": [ @@ -27598,6 +24636,276 @@ 1 ] ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_336": { + "block_sizes": [ + 4, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "" + ], + "num_warps": 4, + "num_stages": 7, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_336": { + "block_sizes": [ + 256, + 8 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "last" + ], + "num_warps": 16, + "num_stages": 8, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_352": { + "block_sizes": [ + 512, + 1 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "first" + ], + "num_warps": 1, + "num_stages": 4, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_352": { + "block_sizes": [ + 4, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "" + ], + "num_warps": 32, + "num_stages": 7, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_352": { + "block_sizes": [ + 512, + 4 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 16, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_352": { + "block_sizes": [ + 1, + 8192 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], "flatten_loops": [ false ], @@ -27617,12 +24925,102 @@ "range_flattens": [ null ], + "load_eviction_policies": [ + "", + "last", + "first" + ], + "num_warps": 16, + "num_stages": 2, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_352": { + "block_sizes": [ + 16, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "" + ], + "num_warps": 16, + "num_stages": 2, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_352": { + "block_sizes": [ + 32, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], "load_eviction_policies": [ "", "", "" ], - "num_warps": 32, + "num_warps": 8, "num_stages": 1, "indexing": [ "pointer", @@ -27632,9 +25030,1764 @@ ], "pid_type": "flat" }, - "intermediate_11008_numtokens_512": { + "intermediate_2048_numtokens_368": { + "block_sizes": [ + 4, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "first", + "first" + ], + "num_warps": 8, + "num_stages": 4, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_368": { + "block_sizes": [ + 128, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "first", + "" + ], + "num_warps": 4, + "num_stages": 4, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_368": { + "block_sizes": [ + 64, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 32, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_368": { + "block_sizes": [ + 2, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "last" + ], + "num_warps": 1, + "num_stages": 4, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_368": { + "block_sizes": [ + 128, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_368": { "block_sizes": [ 32, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_384": { + "block_sizes": [ + 256, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_384": { + "block_sizes": [ + 512, + 2 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "first", + "last" + ], + "num_warps": 8, + "num_stages": 3, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_384": { + "block_sizes": [ + 4, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "last" + ], + "num_warps": 8, + "num_stages": 5, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_384": { + "block_sizes": [ + 128, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "first" + ], + "num_warps": 4, + "num_stages": 2, + "indexing": [ + "pointer", + "pointer", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_384": { + "block_sizes": [ + 1, + 8192 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "first" + ], + "num_warps": 4, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_384": { + "block_sizes": [ + 128, + 16 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "first" + ], + "num_warps": 32, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_400": { + "block_sizes": [ + 1, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 4, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_400": { + "block_sizes": [ + 16, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_400": { + "block_sizes": [ + 1, + 1024 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "first" + ], + "num_warps": 1, + "num_stages": 1, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_400": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "", + "last" + ], + "num_warps": 8, + "num_stages": 4, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_400": { + "block_sizes": [ + 2, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "last" + ], + "num_warps": 4, + "num_stages": 3, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_400": { + "block_sizes": [ + 4, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "first" + ], + "num_warps": 8, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_416": { + "block_sizes": [ + 256, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_416": { + "block_sizes": [ + 32, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_416": { + "block_sizes": [ + 512, + 8 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "" + ], + "num_warps": 8, + "num_stages": 7, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_416": { + "block_sizes": [ + 1, + 2048 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "first" + ], + "num_warps": 8, + "num_stages": 8, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_416": { + "block_sizes": [ + 256, + 8 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "", + "last" + ], + "num_warps": 4, + "num_stages": 7, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_416": { + "block_sizes": [ + 128, + 16 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "first", + "first" + ], + "num_warps": 16, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_432": { + "block_sizes": [ + 256, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_432": { + "block_sizes": [ + 8, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_432": { + "block_sizes": [ + 64, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_432": { + "block_sizes": [ + 256, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "" + ], + "num_warps": 32, + "num_stages": 5, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_432": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "first", + "first" + ], + "num_warps": 1, + "num_stages": 8, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_432": { + "block_sizes": [ + 512, + 4 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "last" + ], + "num_warps": 1, + "num_stages": 7, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_448": { + "block_sizes": [ + 256, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_448": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "" + ], + "num_warps": 2, + "num_stages": 6, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_448": { + "block_sizes": [ + 8, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "last" + ], + "num_warps": 16, + "num_stages": 2, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_448": { + "block_sizes": [ + 128, + 8 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "first", + "last" + ], + "num_warps": 32, + "num_stages": 3, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_448": { + "block_sizes": [ + 1, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 4, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_448": { + "block_sizes": [ + 64, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 16 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 32, + "num_stages": 8, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_464": { + "block_sizes": [ + 256, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_464": { + "block_sizes": [ + 8, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_464": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 1, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_464": { + "block_sizes": [ + 256, 128 ], "loop_orders": [ @@ -27667,8 +26820,278 @@ "", "" ], - "num_warps": 1, + "num_warps": 8, "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_464": { + "block_sizes": [ + 1, + 16384 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 32, + "num_stages": 6, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_464": { + "block_sizes": [ + 64, + 512 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", + "" + ], + "num_warps": 32, + "num_stages": 7, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_480": { + "block_sizes": [ + 16, + 32 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "first", + "" + ], + "num_warps": 16, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_480": { + "block_sizes": [ + 128, + 16 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "", + "" + ], + "num_warps": 8, + "num_stages": 5, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_480": { + "block_sizes": [ + 64, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 8 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "", + "first" + ], + "num_warps": 2, + "num_stages": 1, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_480": { + "block_sizes": [ + 1, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 2 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "first", + "" + ], + "num_warps": 1, + "num_stages": 2, "indexing": [ "tensor_descriptor", "pointer", @@ -27677,9 +27100,459 @@ ], "pid_type": "flat" }, - "intermediate_14336_numtokens_512": { + "intermediate_11008_numtokens_480": { "block_sizes": [ - 16, + 1, + 1024 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 4, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_480": { + "block_sizes": [ + 1, + 16384 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "last", + "first" + ], + "num_warps": 32, + "num_stages": 3, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_496": { + "block_sizes": [ + 1, + 2048 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "last" + ], + "num_warps": 16, + "num_stages": 7, + "indexing": [ + "pointer", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_496": { + "block_sizes": [ + 8, + 256 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 4, + "num_stages": 8, + "indexing": [ + "pointer", + "pointer", + "tensor_descriptor", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_496": { + "block_sizes": [ + 256, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_496": { + "block_sizes": [ + 256, + 128 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_496": { + "block_sizes": [ + 1, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 4 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "first", + "last", + "last" + ], + "num_warps": 8, + "num_stages": 4, + "indexing": [ + "tensor_descriptor", + "pointer", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_496": { + "block_sizes": [ + 4, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "first" + ], + "num_warps": 4, + "num_stages": 4, + "indexing": [ + "pointer", + "tensor_descriptor", + "tensor_descriptor", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2048_numtokens_512": { + "block_sizes": [ + 512, + 16 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_2880_numtokens_512": { + "block_sizes": [ + 8, + 2048 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "" + ], + "num_warps": 8, + "num_stages": 1, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_4096_numtokens_512": { + "block_sizes": [ + 8, 128 ], "loop_orders": [ @@ -27697,6 +27570,97 @@ "range_unroll_factors": [ 0 ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "last", + "last", + "last" + ], + "num_warps": 16, + "num_stages": 2, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_8192_numtokens_512": { + "block_sizes": [ + 1, + 2048 + ], + "loop_orders": [ + [ + 1, + 0 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 64 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "", + "last" + ], + "num_warps": 4, + "num_stages": 4, + "indexing": [ + "pointer", + "pointer", + "pointer", + "pointer" + ], + "pid_type": "flat" + }, + "intermediate_11008_numtokens_512": { + "block_sizes": [ + 1, + 4096 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + false + ], + "l2_groupings": [ + 1 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], "range_num_stages": [ 0 ], @@ -27709,18 +27673,62 @@ "load_eviction_policies": [ "first", "", + "first" + ], + "num_warps": 16, + "num_stages": 7, + "indexing": [ + "tensor_descriptor", + "tensor_descriptor", + "pointer", + "tensor_descriptor" + ], + "pid_type": "flat" + }, + "intermediate_14336_numtokens_512": { + "block_sizes": [ + 128, + 64 + ], + "loop_orders": [ + [ + 0, + 1 + ] + ], + "flatten_loops": [ + true + ], + "l2_groupings": [ + 32 + ], + "range_unroll_factors": [ + 0 + ], + "range_warp_specializes": [], + "range_num_stages": [ + 0 + ], + "range_multi_buffers": [ + null + ], + "range_flattens": [ + null + ], + "load_eviction_policies": [ + "", + "first", "" ], - "num_warps": 1, - "num_stages": 2, + "num_warps": 2, + "num_stages": 7, "indexing": [ - "pointer", - "pointer", - "pointer", - "pointer" + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor", + "tensor_descriptor" ], - "pid_type": "flat", - "range_warp_specializes": [] + "pid_type": "flat" } } }