xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-10 08:45:00 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Stability fix] turn off HMA allocator when connector is set (#27592)

Browse Source 
Signed-off-by: KuntaiDu <kuntai@uchicago.edu>
Signed-off-by: Kuntai Du <kuntai@uchicago.edu>
This commit is contained in:
Kuntai Du 2025-10-27 18:32:23 -07:00 committed by GitHub
parent a8d2e326ec
commit 255e34ca50
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 14 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
vllm/config/vllm.py
View File

@ -597,6 +597,20 @@ class VllmConfig:
if not current_platform.support_hybrid_kv_cache():
# Hybrid KV cache manager is not supported on non-GPU platforms.
self.scheduler_config.disable_hybrid_kv_cache_manager = True
if self.kv_transfer_config is not None:
# NOTE(Kuntai): turn HMA off for connector for now.
# TODO(Kuntai): have a more elegent solution to check and
# turn off HMA for connector that does not support HMA.
logger.warning(
"Turning off hybrid kv cache manager because "
"`--kv-transfer-config` is set. This will reduce the "
"performance of vLLM on LLMs with sliding window attention "
"or Mamba attention. If you are a developer of kv connector"
", please consider supporting hybrid kv cache manager for "
"your connector by making sure your connector is a subclass"
" of `SupportsHMA` defined in kv_connector/v1/base.py."
)
self.scheduler_config.disable_hybrid_kv_cache_manager = True
if self.kv_events_config is not None:
# Hybrid KV cache manager is not compatible with KV events.
self.scheduler_config.disable_hybrid_kv_cache_manager = True