From 90bbe0a5adc0b9636e84ed3ad058d5853158f4bb Mon Sep 17 00:00:00 2001 From: Wentao Ye <44945378+yewentao256@users.noreply.github.com> Date: Tue, 19 Aug 2025 02:24:16 -0400 Subject: [PATCH] [Log] Warning Once for Cutlass MLA (#23137) Signed-off-by: yewentao256 --- vllm/v1/attention/backends/mla/cutlass_mla.py | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/vllm/v1/attention/backends/mla/cutlass_mla.py b/vllm/v1/attention/backends/mla/cutlass_mla.py index 6e1e5d6533dab..6937ce10ac152 100644 --- a/vllm/v1/attention/backends/mla/cutlass_mla.py +++ b/vllm/v1/attention/backends/mla/cutlass_mla.py @@ -115,7 +115,7 @@ class CutlassMLAImpl(MLACommonImpl[MLACommonMetadata]): self._use_old_cutlass_mla = False force_old_cutlass = os.environ.get("FORCE_OLD_CUTLASS_MLA", None) 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 # TODO: Currently, num_kv_splits is limited to 16 to avoid hanging @@ -123,8 +123,8 @@ class CutlassMLAImpl(MLACommonImpl[MLACommonMetadata]): # FORCE_NUM_KV_SPLITS=1 force_num_kv_splits = os.environ.get("FORCE_NUM_KV_SPLITS", None) if force_num_kv_splits: - logger.warning("Forcing num_kv_splits to %d", - int(force_num_kv_splits)) + logger.warning_once("Forcing num_kv_splits to %d", + int(force_num_kv_splits)) self._num_kv_splits = int(force_num_kv_splits) else: self._num_kv_splits = -1 # => Auto-detect