Skip to content

[Model]Force use triton compressed_tensor_moe instead of cutlass #22345

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

Closed
access2rohit wants to merge 4 commits into from
Closed

[Model]Force use triton compressed_tensor_moe instead of cutlass #22345

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
a64dc82
add env variable to force use triton compressed_tensor_moe instead of…
access2rohit Aug 6, 2025
c412bdd
Update vllm/envs.py
access2rohit Aug 6, 2025
3aa5a40
fix lint error line too long
access2rohit Aug 11, 2025
11f21bd
fix yapf error
access2rohit Aug 12, 2025
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
11 changes: 11 additions & 0 deletions vllm/envs.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -481,6 +481,17 @@ def get_vllm_port() -> Optional[int]:
lambda: bool(int(os.environ["VLLM_USE_FLASHINFER_SAMPLER"]))
if "VLLM_USE_FLASHINFER_SAMPLER" in os.environ else None,

# If set, vllm will force use Triton compressed tensor MOE kernel;
# otherwise it will use the cutlass based one.
"VLLM_TRITON_COMPRESSED_TENSORS_MOE_KERNEL":
lambda: bool(
int(os.getenv("VLLM_TRITON_COMPRESSED_TENSORS_MOE_KERNEL", "0"))),

# If set, vllm will force flashinfer to use tensor cores;
# otherwise will use heuristic based on model architecture.
"VLLM_FLASHINFER_FORCE_TENSOR_CORES":
lambda: bool(int(os.getenv("VLLM_FLASHINFER_FORCE_TENSOR_CORES", "0"))),

# Pipeline stage partition strategy
"VLLM_PP_LAYER_PARTITION":
lambda: os.getenv("VLLM_PP_LAYER_PARTITION", None),
Expand Down
6 changes: 4 additions & 2 deletions vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors_moe.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -523,8 +523,10 @@ def __init__(
# cutlass path
self.is_fp8_w8a8_sm100 = quant_config._is_fp8_w8a8_sm100(
self.weight_quant, self.input_quant)
self.use_cutlass = (quant_config._is_fp8_w8a8_sm90(
self.weight_quant, self.input_quant) or self.is_fp8_w8a8_sm100)
self.use_cutlass = ((quant_config._is_fp8_w8a8_sm90(
self.weight_quant, self.input_quant) or self.is_fp8_w8a8_sm100) and
not envs.VLLM_TRITON_COMPRESSED_TENSORS_MOE_KERNEL)
self.fused_experts = None # type: ignore[assignment]
self.disable_expert_map = False

def create_weights(self, layer: torch.nn.Module, num_experts: int,
Expand Down