[Fusion] Manual ActivationQuantFusionPass initial application (#51415)

Signed-off-by: mgoin <mgoin64@gmail.com>
Signed-off-by: ElizaWszola <ewszola@redhat.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Co-authored-by: ElizaWszola <ewszola@redhat.com>
This commit is contained in:
Misha Goin
2026-09-03 13:37:31 -04:00
committed by GitHub
co-authored by Claude Opus 4.8 ElizaWszola
parent d4d703caf9
commit b7624069ff
10 changed files with 740 additions and 19 deletions
@@ -25,6 +25,9 @@ from vllm.model_executor.kernels.linear.scaled_mm.cutlass import (
from vllm.model_executor.kernels.linear.scaled_mm.flashinfer import (
FlashInferFP8ScaledMMLinearKernel,
)
from vllm.model_executor.kernels.linear.scaled_mm.pytorch import (
PerTensorTorchFP8ScaledMMLinearKernel,
)
from vllm.model_executor.kernels.linear.scaled_mm.ScaledMMLinearKernel import (
FP8ScaledMMLinearLayerConfig,
Int8ScaledMMLinearKernel,
@@ -46,6 +49,7 @@ SUPPORTING = {
CutlassFP8ScaledMMLinearKernel,
FlashInferFP8ScaledMMLinearKernel,
FlashInferCutlassNvFp4LinearKernel,
PerTensorTorchFP8ScaledMMLinearKernel,
}