Files
nvfp4-megamoe-kernel/dsv4
biondizzle e07d79868f CUDA graph: Fix _assemble_scales_single_group swizzle size
The pre-allocated buffer is max-sized, but pad_and_swizzle_single
operates on the full buffer dimensions. Fix: pass a correctly-sized
view (buf[:padded_rows, :padded_cols]) so the swizzle produces the
right output size.

Same fix applied to both linear.py and shared_expert.py.
2026-06-03 17:02:34 +00:00
..