Skip to content

perf(moe): dequant nvfp4 prefill experts by bit placement and load activations contiguously - #578

Closed
YevheniiKotyrlo wants to merge 4 commits into
FlashML-org:mainfrom
YevheniiKotyrlo:perf-nvfp4-prefill-gemm
Closed

perf(moe): dequant nvfp4 prefill experts by bit placement and load activations contiguously#578
YevheniiKotyrlo wants to merge 4 commits into
FlashML-org:mainfrom
YevheniiKotyrlo:perf-nvfp4-prefill-gemm

test(kernels): compile the NVFP4 GEMMs at a ragged K in the cross-arc…

c0714c0
Select commit
Loading
Failed to load commit list.
Sign in for the full log view

Annotations

1 warning and 1 notice
label
succeeded Oct 13, 2026 in 5s