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 20:10:40 +00:00

8787f6a989 SMEM-P: add debug_swap_mn flag to test swapped coordinate mapping

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 20:09:49 +00:00

a272ce10e3 SMEM-P: add debug_p_one flag to write constant P=1.0

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 20:08:32 +00:00

440e49b39d auto: pre-test commit

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 20:08:31 +00:00

3df9cb57f8 SMEM-P: implement two-phase softmax with normalization before SMEM write

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 20:07:04 +00:00

dc93168f57 Revert TMEM layout change (caused error)

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 20:06:25 +00:00

8d48d6d543 SMEM-P: try using PV A-operand layout directly for TMEM-P

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 20:05:04 +00:00

6a2f3d2e27 SMEM-P: debug compute offset for known coordinate

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 20:04:30 +00:00

db9d9b09d2 SMEM-P: add crd2idx debug attempt

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 20:03:23 +00:00

8879f0b701 SMEM-P: test pattern based on fragment indices (k,j)

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 20:02:36 +00:00

420bb62330 auto: pre-test commit

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 20:02:33 +00:00

caf722f78f SMEM-P: add shape debug prints

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 20:00:36 +00:00

2f71868fdd SMEM-P: add debug prints for coordinates

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 19:59:55 +00:00

a166a19ae5 SMEM-P: implement CUTLASS LLM fixes - dynamic frg_tile, local coordinate conversion

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 19:53:47 +00:00

afc7fbefc3 auto: pre-test commit

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 19:53:44 +00:00

72226cef44 SMEM-P: try transposed mapping (swap m/n)

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 19:53:27 +00:00

e4a225a138 SMEM-P: current state - working but mapping wrong (cos 0.02)

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 19:52:48 +00:00

00d08eec28 SMEM-P: debug with linear index pattern m*128+n

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 19:52:05 +00:00

111b50b62a SMEM-P: debug with test pattern (k+j)*0.01

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 19:38:13 +00:00

ca1fb7a1ca SMEM-P: fix coordinate access - qk_coord is (m,n) not ((m,n),0,0)

biondizzle pushed to master at biondizzle/nvfp4-megamoe-kernel 2026-05-23 19:37:36 +00:00

096a48b5cb SMEM-P: fix scoping - define tTMEM_LOADcS_frg unconditionally