Wentao Ye 37ece593c1 [Perf] Padded nvfp4 quant kernel to remove additional copy, 2.4%~5.7% e2e performance improvement (#42774) ...

Signed-off-by: yewentao256 <zhyanwentao@126.com>

2026-05-18 16:38:12 -07:00

..

mixed_precision

[Bugfix][Kernel][ROCm] Fix triton_w4a16 scales mismatch when BLOCK_K > group_size (#39705)

2026-04-13 14:29:45 -04:00

mxfp4

[XPU] [CT] Enable CT W4A4MxFp4 path and add xpu kernel (#38896)

2026-05-13 06:43:00 -07:00

mxfp8

[MXFP8] [XPU] add a new compressed tensor schema and add a xpu mxfp8 gemm kernel (#38707)

2026-04-13 16:59:20 +08:00

nvfp4

[Perf] Padded nvfp4 quant kernel to remove additional copy, 2.4%~5.7% e2e performance improvement (#42774)

2026-05-18 16:38:12 -07:00

scaled_mm

[XPU] fix weight scale shape (#42725)

2026-05-17 16:55:10 +08:00

__init__.py

[Kernel][UX] Add --linear-backend arg for linear kernel selection (#39538)

2026-05-15 17:07:39 -07:00

base.py

[W8A8 Block Linear Refactor][2/N] Remove W8A8Fp8BlockLinearOp and adopt Fp8 block linear kernel selections. (#33892)

2026-04-09 08:50:39 +08:00