Skip to content

[Bugfix] Fix potential EAGLE spec decode segfault during graph capture #32818

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
LucasWilkinson merged 1 commit into from
Jan 22, 2026
+8 −4
Merged

[Bugfix] Fix potential EAGLE spec decode segfault during graph capture #32818

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
12 changes: 8 additions & 4 deletions vllm/v1/spec_decode/eagle.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1222,10 +1222,14 @@ def dummy_run(
num_tokens_dp_padded, num_tokens_across_dp = self._pad_batch_across_dp(
num_tokens_unpadded=num_tokens, num_tokens_padded=num_tokens
)
cudagraph_runtime_mode, batch_desc = self.cudagraph_dispatcher.dispatch(
num_tokens_dp_padded
)
num_input_tokens = batch_desc.num_tokens
if use_cudagraphs:
cudagraph_runtime_mode, batch_desc = (
self.cudagraph_dispatcher.dispatch(num_tokens_dp_padded)
)
num_input_tokens = batch_desc.num_tokens
else:
cudagraph_runtime_mode = CUDAGraphMode.NONE
num_input_tokens = num_tokens_dp_padded
if num_tokens_across_dp is not None:
num_tokens_across_dp[self.dp_rank] = num_input_tokens

Expand Down
Loading