biondizzle/nvfp4-megamoe-kernel
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
512 Commits 1 Branch 17 Tags
50e9b5da81477e000e4770e94de9d175204f5ff0
Commit Graph

3 Commits

Author SHA1 Message Date
biondizzle
f07643791e Fix hidden_size: shared expert uses 7168, not HC_DIM 28672 2026-05-18 20:10:32 +00:00
biondizzle
c1aa4af123 Shared expert: dedicated CuTeDSL runner with proper scale assembly 
- CuTeDSLSharedExpertRunner: num_groups=1 GEMM, no scatter/routing
- _assemble_scales_single_group: pad to 128 rows + Blackwell swizzle
- All buffers pre-allocated for cudagraph compatibility
- Updated test to use dedicated runner instead of MoE runner hack
 2026-05-18 20:08:34 +00:00
biondizzle
e8b289e30d WIP: CuTeDSL shared expert kernel 
Dedicated runner (shared_expert_pipeline.py) and test (test_shared_expert.py).
Tried reusing MoE runner with 1 expert — fails because MoE runner assumes
hidden_size != HC_DIM for scatter. Need dedicated runner with correct
scale assembly. Will continue tomorrow.
 2026-05-18 20:02:19 +00:00