biondizzle

0 Followers · 0 Following

• Joined on 2025-12-10

Repositories

25

Projects Packages Public Activity Starred Repositories

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 05:12:31 +00:00

a0363e8911 Fix CuTeDSL scoping: hoist P store vars out of if block

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 05:10:48 +00:00

86bf5771c1 Fix O rescale: use Stage C proven correction_rescale pattern

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 05:09:26 +00:00

e204aa7a4c Fix tOrP0 indexing: 3-dim slice (None,None,kb) not 4-dim

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 05:08:18 +00:00

dda5afee87 Fix CuTeDSL scoping: unconditionally define tOrP0 and tCrP

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 05:07:32 +00:00

0f715bfaff Fix CuTeDSL variable scoping: define tOrP0 and tCrP in both branches

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 05:06:47 +00:00

70db626550 Fix p_tmem_s: use ComposedLayout from make_smem_layout_a, pass as kernel arg

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 05:04:53 +00:00

09cac38a67 Consolidate FMHA stages A/B/C into unified kernel module with SMEM-P stub

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 03:56:57 +00:00

6f834ae8b5 WIP: make_tiled_copy_C for P→SMEM

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 03:54:50 +00:00

8114a225d1 fix: cpasync.CopyOp for reg→SMEM

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 03:52:49 +00:00

0dbdc4f865 fix: CopyAtomUniversalOp

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 03:51:59 +00:00

05173c1992 WIP: tiled copy for P→SMEM (zero fill)

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 03:51:02 +00:00

5a9c299f64 fix: cute.copy(dst, src) order

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 03:50:11 +00:00

398f5cf631 fix: BFloat16 not Float32 for bf16 reg

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 03:49:06 +00:00

9bc7fc9361 WIP: P→SMEM write stub (zero fill, proper mapping TODO)

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 03:47:55 +00:00

ed35a8a4ba fix: partition_A not partition_S

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 03:47:10 +00:00

48432522b8 fix: make_smem_layout_epi not make_epilogue_smem_layout

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 03:46:02 +00:00

07f319d1f3 WIP: SMEM P path for PV (compiles but P write not implemented)

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 03:42:55 +00:00

1be005296c debug: hd=64 with CUDA_LAUNCH_BLOCKING

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 03:42:10 +00:00

482928f142 D1: P store as BF16 using PV A-fragment layout (tOrP0)

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 03:40:11 +00:00

f266c3dae2 D1: align P store and PV A-fragment layouts via tP