xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-01-02 00:34:06 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Refactor] Remove Unused Env VLLM_ENABLE_MOE_ALIGN_BLOCK_SIZE_TRITON (#20334)

Browse Source 
Signed-off-by: yewentao256 <zhyanwentao@126.com>
This commit is contained in:
Wentao Ye 2025-07-01 22:03:43 -04:00 committed by GitHub
parent 7058d7dd5d
commit 9dae7d46bf
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 0 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
vllm/envs.py
View File

@ -104,7 +104,6 @@ if TYPE_CHECKING:
VLLM_SERVER_DEV_MODE: bool = False
VLLM_V1_OUTPUT_PROC_CHUNK_SIZE: int = 128
VLLM_MLA_DISABLE: bool = False
VLLM_ENABLE_MOE_ALIGN_BLOCK_SIZE_TRITON: bool = False
VLLM_RAY_PER_WORKER_GPUS: float = 1.0
VLLM_RAY_BUNDLE_INDICES: str = ""
VLLM_CUDART_SO_PATH: Optional[str] = None
@ -769,12 +768,6 @@ environment_variables: dict[str, Callable[[], Any]] = {
"VLLM_MLA_DISABLE":
lambda: bool(int(os.getenv("VLLM_MLA_DISABLE", "0"))),
# If set, vLLM will use the Triton implementation of moe_align_block_size,
# i.e. moe_align_block_size_triton in fused_moe.py.
"VLLM_ENABLE_MOE_ALIGN_BLOCK_SIZE_TRITON":
lambda: bool(int(os.getenv("VLLM_ENABLE_MOE_ALIGN_BLOCK_SIZE_TRITON", "0"))
),
# Number of GPUs per worker in Ray, if it is set to be a fraction,
# it allows ray to schedule multiple actors on a single GPU,
# so that users can colocate other actors on the same GPUs as vLLM.

1
vllm/model_executor/layers/fused_moe/moe_align_block_size.py
View File

@ -94,7 +94,6 @@ def moe_align_block_size_stage4(
# Triton implementation based on:
# https://github.com/sgl-project/sglang/commit/ba5112ff691d791a9e38c6c71f59324a5fcb49d0
# TODO(wentao): Deprecated this function in the future.
def moe_align_block_size_triton(
topk_ids: torch.Tensor,
num_experts: int,