Skip to content

[Revert] Remove DeepGEMM availability check in DeepseekV32IndexerMetadataBuilder #38076

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
chaunceyjiang merged 1 commit into from
Mar 26, 2026
Merged

[Revert] Remove DeepGEMM availability check in DeepseekV32IndexerMetadataBuilder #38076

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 0 additions & 7 deletions vllm/v1/attention/backends/mla/indexer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,7 +10,6 @@
from vllm.utils.deep_gemm import (
get_paged_mqa_logits_metadata,
has_deep_gemm,
is_deep_gemm_supported,
)
from vllm.utils.math_utils import cdiv
from vllm.utils.platform_utils import num_compute_units
Expand Down Expand Up @@ -216,12 +215,6 @@ def get_cudagraph_support(
vllm_config: VllmConfig,
kv_cache_spec: AttentionSpec,
) -> AttentionCGSupport:
if not is_deep_gemm_supported():
logger.warning_once(
"DeepGEMM is not available. Disabling CUDA graph support "
"for sparse attention indexer. This may reduce performance.",
)
return AttentionCGSupport.NEVER
return AttentionCGSupport.UNIFORM_BATCH

def __init__(self, *args, **kwargs):
Expand Down
Loading