xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-05-30 06:17:05 +08:00
Code Issues Packages Projects Releases Wiki Activity

[BugFix] skip combo kernel on cpu (#29129)

Browse Source 
Signed-off-by: Boyuan Feng <boyuan@meta.com>
This commit is contained in:
Boyuan Feng 2025-11-20 19:50:59 -08:00 committed by GitHub
parent 56e96b37e4
commit 8c25f9cfb6
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 2 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/config/compilation.py
View File

@ -664,6 +664,8 @@ class CompilationConfig:
is_torch_equal_or_newer("2.9.0.dev") is_torch_equal_or_newer("2.9.0.dev")
and "combo_kernels" not in self.inductor_compile_config and "combo_kernels" not in self.inductor_compile_config
and "benchmark_combo_kernel" not in self.inductor_compile_config and "benchmark_combo_kernel" not in self.inductor_compile_config
# (fixme @boyuan) combo kernel does not support cpu yet.
and not current_platform.is_cpu()
): ):
# use horizontal fusion, which is useful for fusing qk-norm and # use horizontal fusion, which is useful for fusing qk-norm and
# qk-rope when query and key have different shapes. # qk-rope when query and key have different shapes.