xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-05-07 05:52:20 +08:00
Code Issues Packages Projects Releases Wiki Activity

[BugFix] fix CUTLASS MLA full cudagraph (#23200)

Browse Source 
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
This commit is contained in:
Lucas Wilkinson 2025-08-19 18:17:08 -04:00 committed by GitHub
parent 0f4f0191d8
commit 14e2b0730b
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/v1/attention/backends/mla/cutlass_mla.py
View File

@ -21,7 +21,7 @@ logger = init_logger(__name__)
class CutlassMLAMetadataBuilder(MLACommonMetadataBuilder[MLACommonMetadata]): class CutlassMLAMetadataBuilder(MLACommonMetadataBuilder[MLACommonMetadata]):
# enable full CUDA Graph support for decode-only capture # enable full CUDA Graph support for decode-only capture
attn_cudagraph_support: ClassVar[ cudagraph_support: ClassVar[
AttentionCGSupport] = AttentionCGSupport.UNIFORM_SINGLE_TOKEN_DECODE AttentionCGSupport] = AttentionCGSupport.UNIFORM_SINGLE_TOKEN_DECODE