[Refactor] Make FP8 Linear Ops use kernel abstraction (#27814)

Signed-off-by: vllmellm <vllm.ellm@embeddedllm.com>
This commit is contained in:
vllmellm
2026-01-20 14:48:20 +08:00
committed by GitHub
parent e9c83cdc51
commit 148117ea2e
30 changed files with 1464 additions and 1035 deletions
+1 -1
View File
@@ -372,7 +372,7 @@ def _rocm_aiter_gemm_a8w8_impl(
# gemm_a8w8_CK(a, b, scale_a, scale_b, bias) expects
# a to be [M, K]
# b to be [N, K]
# CutlassScaledMMLinearKernel prepare weight `w_q` in [K, N] format
# CutlassInt8ScaledMMLinearKernel prepare weight `w_q` in [K, N] format
return gemm_a8w8_CK(A, B, As, Bs, bias, output_dtype)