biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[3/N][Attention] Move AttentionMetadata-related code from utils.py to backend.py (#32054)

Browse Source 
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
This commit is contained in:
Matthew Bonanni
2026-01-12 12:13:56 -05:00
committed by GitHub
parent 7c0d3c5152
commit 20228cb851
37 changed files with 374 additions and 370 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
docs/design/cuda_graphs.md
View File

@@ -149,7 +149,7 @@ The CUDA Graphs wrapper no longer manages the warm-up logic. The warm-up process
## CUDA Graphs Compatibility of Attention Backends
To signal the CUDA Graphs compatibility of the attention backends, we introduce a new enum type [AttentionCGSupport][vllm.v1.attention.backends.utils.AttentionCGSupport], which is an enum type that tracks the capability of the attention backend to support CUDA Graphs. The value is sorted in the order of the capability, i.e., `ALWAYS`> `UNIFORM_BATCH`> `UNIFORM_SINGLE_TOKEN_DECODE`> `NEVER`.
To signal the CUDA Graphs compatibility of the attention backends, we introduce a new enum type [AttentionCGSupport][vllm.v1.attention.backend.AttentionCGSupport], which is an enum type that tracks the capability of the attention backend to support CUDA Graphs. The value is sorted in the order of the capability, i.e., `ALWAYS`> `UNIFORM_BATCH`> `UNIFORM_SINGLE_TOKEN_DECODE`> `NEVER`.
```python
class AttentionCGSupport(enum.Enum):