Skip to content

[fix] fix dsv3_router_gemm filter #7750

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
merged 1 commit into from
Jul 3, 2025
Merged

[fix] fix dsv3_router_gemm filter #7750

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
3 changes: 2 additions & 1 deletion python/sglang/srt/models/deepseek_v2.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -233,7 +233,8 @@ def forward(self, hidden_states):
)

if (
hidden_states.shape[0] < 4
_is_cuda
and hidden_states.shape[0] < 4
and hidden_states.shape[1] == 7168
and self.weight.shape[0] == 256
and _device_sm >= 90
Comment on lines 235 to 240
Copy link
Contributor

@gemini-code-assist gemini-code-assist bot Jul 3, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

This change correctly restricts a CUDA-specific optimization path, preventing potential crashes on other hardware like AMD GPUs. To ensure long-term stability and prevent regressions, it would be beneficial to add a unit test for this logic. A test could simulate a non-CUDA environment (e.g., by mocking _is_cuda to False and _device_sm to a value >= 90) and verify that this specialized code path is not executed.

Expand Down
Loading