Skip to content

[Feature] Reduce the cost of torchair #4756

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
jianzs wants to merge 1 commit into from
Closed

[Feature] Reduce the cost of torchair #4756

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
1 change: 1 addition & 0 deletions vllm_ascend/torchair/torchair_model_runner.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -255,6 +255,7 @@ def _compile_torchair_graph(self, torchair_graph_batch_sizes) -> None:
# Trigger torchair graph capture for specific shapes.
# Capture the large shapes first so that the smaller shapes
# can reuse the memory pool allocated for the large shapes.
torch._dynamo.config.inline_inbuilt_nn_modules = False
for idx, num_tokens in enumerate(reversed(torchair_graph_batch_sizes)):
for _ in range(self.vllm_config.compilation_config.
cudagraph_num_of_warmups):
Expand Down
Loading