biondizzle/nvfp4-megamoe-kernel
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
1,930 Commits 1 Branch 17 Tags
ca661d32e8e94de838d550da52e934dd7f5c6312
Commit Graph

2 Commits

Author SHA1 Message Date
biondizzle
5290c91c35 fix quantize_nvfp4 kernel: use proven single-thread-per-CTA pattern from deinterleave_quantize.cu 
The warp shuffle approach failed because __shfl_down_sync with 16 threads
has undefined behavior for the odd nibble. Use the same pattern as the
working deinterleave_quantize.cu: 1 CTA per 16-element block, 16 threads
per CTA, each thread reads all 16 elements sequentially and computes
amax + quantize + pack.
 2026-05-25 16:21:44 +00:00
biondizzle
c2e3d15633 NVFP4-1.1 integration: GPU-only quantize kernel + MoE pipeline wiring 
- Add quantize_nvfp4.cu: BF16→FP4 GPU kernel (no CPU sync, warp shuffle amax)
- Add quantize_nvfp4_gpu() bridge in ops/quantize.py
- Fix deinterleave_quantize kernel path (dsv4/ops/kernels → dsv4/kernels/cuda)
- Wire GPU quantize into Nvfp4MoE._run_impl():
  - L1 input: quantize_nvfp4_gpu (replaces quantize_activation_nvfp4)
  - Fused SwiGLU L2: deinterleave_quantize_nvfp4_cuda (single kernel)
  - Non-fused L2: quantize_nvfp4_gpu
- Add test_nvfp4_gpu_quantize.py for both kernels
 2026-05-25 16:19:07 +00:00