Skip to content

[Bugfix] Fix DeepGEMM after #29546 #30267

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
yeqcharlotte merged 4 commits into from
Dec 9, 2025
Merged

[Bugfix] Fix DeepGEMM after #29546 #30267

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
823709d
update
zhewenl Dec 8, 2025
2926f6f
update
zhewenl Dec 8, 2025
2edfcfc
update
zhewenl Dec 8, 2025
b5245db
update
zhewenl Dec 8, 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
16 changes: 10 additions & 6 deletions vllm/model_executor/layers/quantization/utils/fp8_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,6 +30,7 @@
from vllm.platforms import current_platform
from vllm.triton_utils import tl, triton
from vllm.utils.deep_gemm import (
DeepGemmQuantScaleFMT,
fp8_gemm_nt,
is_deep_gemm_e8m0_used,
is_deep_gemm_supported,
Expand Down Expand Up @@ -268,12 +269,15 @@ def _run_deepgemm(
weight: torch.Tensor,
weight_scale: torch.Tensor,
) -> torch.Tensor:
assert self.deepgemm_input_quant_op is not None
q_input, input_scale = per_token_group_quant_fp8_packed_for_deepgemm(
input_2d,
group_size=self.act_quant_group_shape.col,
use_ue8m0=True,
)
if DeepGemmQuantScaleFMT.from_oracle() == DeepGemmQuantScaleFMT.UE8M0:
q_input, input_scale = per_token_group_quant_fp8_packed_for_deepgemm(
input_2d,
group_size=self.act_quant_group_shape.col,
use_ue8m0=True,
)
else:
assert self.deepgemm_input_quant_op is not None
q_input, input_scale = self.deepgemm_input_quant_op(input_2d)
output = torch.empty(
(q_input.shape[0], weight.shape[0]),
dtype=torch.bfloat16,
Expand Down
1 change: 1 addition & 0 deletions vllm/utils/deep_gemm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -399,6 +399,7 @@ def should_use_deepgemm_for_fp8_linear_for_nk(

__all__ = [
"calc_diff",
"DeepGemmQuantScaleFMT",
"fp8_gemm_nt",
"m_grouped_fp8_gemm_nt_contiguous",
"fp8_m_grouped_gemm_nt_masked",
Expand Down