Fix search path for torch allocator in editable installs and ensure CUDA support is available #1498
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
vyasr
changed the title
harrism
approved these changes
Mar 6, 2024
|
/merge |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description
The current logic for finding the allocator .so file presumes that torch.py is in the same directory tree as the compiled allocator. While this is true when the package is installed, it is not true for editable installs of the package where py files are left in the source tree while compiled modules are placed into a build tree (both are found by indirection).
The current allocator logic also assumes that if pytorch is available it is a version compiled with CUDA support. This PR also adds a check to verify that we don't try to set the allocator on a CPU-only build of torch.
Checklist