biondizzle

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel

2026-05-28 05:09:51 +00:00

db17d8db9a fix: cvta.to.shared PTX for SMEM address

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel

2026-05-28 05:09:30 +00:00

e12a81ae36 fix: include cstdint

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel

2026-05-28 05:09:09 +00:00

0c73a024ba fix: guard CUTLASS includes with __CUDA_ARCH__ for host compilation

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel

2026-05-28 05:08:28 +00:00

41e59a2423 FMHA SM100: Add SMEM descriptor construction for tcgen05.mma

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel

2026-05-28 05:06:49 +00:00

3eb432d064 fix: CUTLASS path /root/cutlass

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel

2026-05-28 05:06:14 +00:00

66d9f5c60f fix: --x cu for .cuh compilation

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel

2026-05-28 05:05:56 +00:00

4dcd80ea0d fix: use full nvcc path

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel

2026-05-28 05:05:33 +00:00

fac7275f2b test: nvcc compilation test for FMHA SM100 kernel

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel

2026-05-28 05:04:46 +00:00

230c350c77 FMHA SM100: Raw CUDA C++ decode kernel — initial skeleton

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel

2026-05-28 04:59:03 +00:00

b2d0417a46 NVFP4-1.1: Mark fp4_quant.py as toolchain-blocked, clean up test files

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel

2026-05-28 04:57:48 +00:00

650bcdcccf test: f32 vs i32 GMEM store

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel

2026-05-28 04:56:18 +00:00

cc37ce6dbf test: absolute minimum CuTeDSL int store + float cmp

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel

2026-05-28 04:55:25 +00:00

c4fdfc7789 test: isolate which fp4_quant function causes LLVM ERROR

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel

2026-05-28 04:54:28 +00:00

b3eb46d4ec NVFP4-1.1: Restore threshold RNE approach — inline PTX blocked by toolchain

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel

2026-05-28 04:50:23 +00:00

71ee1485ea test: constraints runner

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel

2026-05-28 04:50:10 +00:00

c55c237fcd test: different constraint strings + bitcast approach

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel

2026-05-28 04:49:04 +00:00

4806e9ba11 test: llvm.inline_asm with Int32._mlir_type matching cvt_i8_bf16 pattern

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel

2026-05-28 04:47:05 +00:00

ade49d964d fix: test_ptx_runner path

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel

2026-05-28 04:46:55 +00:00

dc9596c6bc test: sub-process isolation for each f32→i32 approach

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel

2026-05-28 04:46:07 +00:00

136a89f4e3 test: compare nvvm.inline_ptx approaches + arith.fptosi