Skip to content

GitLab

Unverified Commit 90bbe0a5 authored by Wentao Ye's avatar Wentao Ye Committed by GitHub
Browse files

[Log] Warning Once for Cutlass MLA (#23137) 


Signed-off-by: default avataryewentao256 <zhyanwentao@126.com>
parent e75f3422
Hide whitespace changes
Inline Side-by-side
Showing with 3 additions and 3 deletions
vllm/v1/attention/backends/mla/cutlass_mla.py
View file @ 90bbe0a5
...@@ -115,7 +115,7 @@ class CutlassMLAImpl(MLACommonImpl[MLACommonMetadata]): ...@@ -115,7 +115,7 @@ class CutlassMLAImpl(MLACommonImpl[MLACommonMetadata]):
self._use_old_cutlass_mla = False self._use_old_cutlass_mla = False
force_old_cutlass = os.environ.get("FORCE_OLD_CUTLASS_MLA", None) force_old_cutlass = os.environ.get("FORCE_OLD_CUTLASS_MLA", None)
if force_old_cutlass: if force_old_cutlass:
logger.warning("Forcing old cutlass mla kernel") logger.warning_once("Forcing old cutlass mla kernel")
self._use_old_cutlass_mla = True self._use_old_cutlass_mla = True
# TODO: Currently, num_kv_splits is limited to 16 to avoid hanging # TODO: Currently, num_kv_splits is limited to 16 to avoid hanging
...@@ -123,8 +123,8 @@ class CutlassMLAImpl(MLACommonImpl[MLACommonMetadata]): ...@@ -123,8 +123,8 @@ class CutlassMLAImpl(MLACommonImpl[MLACommonMetadata]):
# FORCE_NUM_KV_SPLITS=1 # FORCE_NUM_KV_SPLITS=1
force_num_kv_splits = os.environ.get("FORCE_NUM_KV_SPLITS", None) force_num_kv_splits = os.environ.get("FORCE_NUM_KV_SPLITS", None)
if force_num_kv_splits: if force_num_kv_splits:
logger.warning("Forcing num_kv_splits to %d", logger.warning_once("Forcing num_kv_splits to %d",
int(force_num_kv_splits)) int(force_num_kv_splits))
self._num_kv_splits = int(force_num_kv_splits) self._num_kv_splits = int(force_num_kv_splits)
else: else:
self._num_kv_splits = -1 # => Auto-detect self._num_kv_splits = -1 # => Auto-detect
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment