Replace hardcoded buffer sizes with cuMemGetAllocationGranularity #82
Conversation
This file contains hidden or 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
Summary: Previously, the code used SFINAE templates to detect and use c10::CachingAllocator::kLargeBuffer and kSmallBuffer, with hardcoded fallbacks of 20MB and 2MB respectively. This approach had issues with different PyTorch versions that had the buffer sizes defined in different locations, making it unreliable across PyTorch versions. This diff simplifies the implementation by directly querying the allocation granularity from the CUDA driver using cuMemGetAllocationGranularity. This ensures we use the correct, device-specific granularity for memory registration chunks, making the code more robust and portable across different GPU architectures and PyTorch versions. Differential Revision: D88688463
meta-cla
bot
added
the
CLA Signed
|
@pavanbalaji has exported this pull request. If you are a Meta employee, you can view the originating Diff in D88688463. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.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.
Summary:
Previously, the code used SFINAE templates to detect and use
c10::CachingAllocator::kLargeBuffer and kSmallBuffer, with hardcoded
fallbacks of 20MB and 2MB respectively. This approach had issues with
different PyTorch versions that had the buffer sizes defined in
different locations, making it unreliable across PyTorch versions.
This diff simplifies the implementation by directly querying the
allocation granularity from the CUDA driver using
cuMemGetAllocationGranularity. This ensures we use the correct,
device-specific granularity for memory registration chunks, making the
code more robust and portable across different GPU architectures and
PyTorch versions.
Differential Revision: D88688463