Skip to content
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.

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

Auto detect and warn against performance pitfalls #699

Closed
avik-pal opened this issue Jun 14, 2024 · 0 comments · Fixed by #727
Closed

Auto detect and warn against performance pitfalls #699

avik-pal opened this issue Jun 14, 2024 · 0 comments · Fixed by #727
Labels
enhancement New feature or request performance
Milestone

Comments

@avik-pal
Copy link
Member

avik-pal commented Jun 14, 2024

Implementation

Set up a preferences-based system for doing the following:

  1. default: Do nothing, i.e., the current state
  2. warn: If we detect type mismatches, we display a warning and tell users to use Experimental.@debug_mode to figure out the exact location of the model. (I don't mind having this as the default but this will be breaking if we choose to do so)
  3. autoconvert: My least favorite version (Flux's default), but few people do seem to like this. We convert the types to the most performant version.
  4. error: Throw an error if eltype mismatch is detected.

Documentation

  1. Document the preferences for eltype detection
  2. Create a Page showcasing the various options, even from GPUArrays verse
  3. Options from DispatchDoctor once that is merged into LuxCore
@avik-pal avik-pal added performance enhancement New feature or request labels Jun 14, 2024
@avik-pal avik-pal added this to the v0.6 milestone Jun 28, 2024
@avik-pal avik-pal linked a pull request Jul 1, 2024 that will close this issue
8 tasks
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
enhancement New feature or request performance
Projects
None yet
Development

Successfully merging a pull request may close this issue.

1 participant