Skip to content

Conversation

@wsmoses
Copy link
Member

@wsmoses wsmoses commented Dec 14, 2025

No description provided.

@wsmoses wsmoses requested review from avik-pal and giordano December 14, 2025 22:39
Reactant.set_default_backend("gpu")
```

```julia [AMD GPUs]
Copy link
Member

Choose a reason for hiding this comment

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

How about putting this together with Nvidia GPUs for the time being? The code to execute is just the same.

Copy link
Member Author

Choose a reason for hiding this comment

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

yeah so one thing I'm debating doing here, is we could have a commented out line set_default_backend("cuda") vs "rocm" which will enforce which one

Copy link
Member

Choose a reason for hiding this comment

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

I like being explicit, it reduces confusion.

Maybe should keep "gpu" as an alias of "cuda" for the rest of v0.2 though?

Copy link
Member Author

Choose a reason for hiding this comment

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

So the current setup at least has gpu essentially as an Alia's for whatever GPU was found on the system, including when metal earlier worked.

I think it's nice to have both so you can either just say get me a GPU or alternatively make sure you get this specifically

Copy link
Member Author

Choose a reason for hiding this comment

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

@giordano what do you think of this?

Copy link
Member

Choose a reason for hiding this comment

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

Looks better (although the code commented out is easy to ignore/miss)

Added comments for setting GPU backends in Reactant usage examples.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants