Skip to content

Respect passed in device overrides in engine args #18808

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
wants to merge 1 commit into from
Closed

Respect passed in device overrides in engine args #18808

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
2 changes: 1 addition & 1 deletion vllm/engine/arg_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -979,7 +979,7 @@ def create_engine_config(
from vllm.platforms import current_platform
current_platform.pre_register_and_update()

device_config = DeviceConfig(device=current_platform.device_type)
device_config = DeviceConfig(device=self.device)
Copy link
Member

Choose a reason for hiding this comment

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

We have already deprecated the --device argument in #18399 after multiple platform plugin authors agreed to it. Could you elaborate on how this is used in your application? cc @youkaichao

Copy link
Collaborator

Choose a reason for hiding this comment

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

Seems we intentionally make "device" arg not honored. Wondering @Adolfo-Karim , do we need to use "device" in EngineArg for something?

Copy link
Collaborator

Choose a reason for hiding this comment

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

Discuss offline with @Adolfo-Karim , it's more like a "CPU" deterministic logic.

Copy link
Member

Choose a reason for hiding this comment

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

"CPU" deterministic logic

what does this mean?

Copy link
Collaborator

Choose a reason for hiding this comment

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

https://github.com/vllm-project/vllm/blob/main/vllm/platforms/__init__.py#L156, here, we check the "CPU" in the package name. We don't actually have package, since we use monolithic / unified codebase.

model_config = self.create_model_config()

# * If VLLM_USE_V1 is unset, we enable V1 for "supported features"
Expand Down