xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-04-06 05:37:03 +08:00
Code Issues Packages Projects Releases Wiki Activity

add missing kernels for cuda dispatch

Browse Source 
Signed-off-by: vllmellm <vllm.ellm@embeddedllm.com>
This commit is contained in:
vllmellm 2025-11-04 12:23:38 +00:00
parent f10171cb3d
commit fb72ec8218
1 changed files with 11 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
vllm/model_executor/layers/quantization/kernels/scaled_mm/__init__.py
View File

@ -17,6 +17,10 @@ from vllm.model_executor.layers.quantization.kernels.scaled_mm.cutlass import (
CutlassFP8ScaledMMLinearKernel,
CutlassScaledMMLinearKernel,
)
from vllm.model_executor.layers.quantization.kernels.scaled_mm.flashinfer import (
FlashInferScaledMMLinearKernel
)
from vllm.model_executor.layers.quantization.kernels.scaled_mm.pytorch import (
ChannelWiseTorchScaledMMLinearKernel,
PerTensorTorchScaledMMLinearKernel,
@ -54,7 +58,13 @@ _POSSIBLE_INT8_KERNELS: dict[PlatformEnum, list[type[Int8ScaledMMLinearKernel]]]
# in priority/performance order (when available)
_POSSIBLE_FP8_KERNELS: dict[PlatformEnum, list[type[FP8ScaledMMLinearKernel]]] = {
PlatformEnum.CUDA: [CutlassFP8ScaledMMLinearKernel],
PlatformEnum.CUDA: [
FlashInferScaledMMLinearKernel,
CutlassFP8ScaledMMLinearKernel,
PerTensorTorchScaledMMLinearKernel,
RowWiseTorchScaledMMLinearKernel,
ChannelWiseTorchScaledMMLinearKernel,
],
PlatformEnum.ROCM: [
ROCmScaledMMLinearKernel,
PerTensorTorchScaledMMLinearKernel,