Enable cache line locking support in SSD kernel #2949
Closed
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
✅ Deploy Preview for pytorch-fbgemm-docs ready!
To edit notification comments on pull requests, go to your Netlify site configuration. |
|
This pull request was exported from Phabricator. Differential Revision: D60812956 |
|
This pull request was exported from Phabricator. Differential Revision: D60812956 |
sryap
added a commit
to sryap/FBGEMM
that referenced
this pull request
Aug 8, 2024
Summary: X-link: facebookresearch/FBGEMM#51 Pull Request resolved: pytorch#2949 This diff enables cache line locking in `ssd_cache_actions_insert_kernel`. We updated the cache slot cost computation (within a cache set) to facilitate the cache line locking logic. Before this diff, the cache slots are ranked based on the their costs (i.e., their timestamps which can be retrieved from the cache state). The cache slots that contain the lowest costs (lowest timestamps) will be evicted first. However, when cache line locking is enabled, the cache slot that has the lowest timestamp cannot be used if it is locked. Therefore, we assign unavailable cache slots (i.e., being locked or being occupied by another index in the same batch) with the highest cost (that is the current timestamp). The slot will not be evicted if its cost is the same as the current timestamp. Differential Revision: D60812956
|
This pull request was exported from Phabricator. Differential Revision: D60812956 |
sryap
added a commit
to sryap/FBGEMM
that referenced
this pull request
Aug 8, 2024
Summary: X-link: facebookresearch/FBGEMM#51 Pull Request resolved: pytorch#2949 This diff enables cache line locking in `ssd_cache_actions_insert_kernel`. We updated the cache slot cost computation (within a cache set) to facilitate the cache line locking logic. Before this diff, the cache slots are ranked based on the their costs (i.e., their timestamps which can be retrieved from the cache state). The cache slots that contain the lowest costs (lowest timestamps) will be evicted first. However, when cache line locking is enabled, the cache slot that has the lowest timestamp cannot be used if it is locked. Therefore, we assign unavailable cache slots (i.e., being locked or being occupied by another index in the same batch) with the highest cost (that is the current timestamp). The slot will not be evicted if its cost is the same as the current timestamp. Differential Revision: D60812956
Differential Revision: D60413116
Summary: X-link: facebookresearch/FBGEMM#51 Pull Request resolved: pytorch#2949 This diff enables cache line locking in `ssd_cache_actions_insert_kernel`. We updated the cache slot cost computation (within a cache set) to facilitate the cache line locking logic. Before this diff, the cache slots are ranked based on the their costs (i.e., their timestamps which can be retrieved from the cache state). The cache slots that contain the lowest costs (lowest timestamps) will be evicted first. However, when cache line locking is enabled, the cache slot that has the lowest timestamp cannot be used if it is locked. Therefore, we assign unavailable cache slots (i.e., being locked or being occupied by another index in the same batch) with the highest cost (that is the current timestamp). The slot will not be evicted if its cost is the same as the current timestamp. Reviewed By: ehsanardestani Differential Revision: D60812956
|
This pull request was exported from Phabricator. Differential Revision: D60812956 |
sryap
force-pushed
the
export-D60812956
branch
from
253d276 to
e0fb4d4
Compare
|
This pull request has been merged in 19990d8. |
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.
Summary:
This diff enables cache line locking in
ssd_cache_actions_insert_kernel.We updated the cache slot cost computation (within a cache set) to
facilitate the cache line locking logic. Before this diff, the cache
slots are ranked based on the their costs (i.e., their timestamps
which can be retrieved from the cache state). The cache slots that
contain the lowest costs (lowest timestamps) will be evicted first.
However, when cache line locking is enabled, the cache slot that has
the lowest timestamp cannot be used if it is locked. Therefore, we
assign unavailable cache slots (i.e., being locked or being occupied
by another index in the same batch) with the highest cost (that is the
current timestamp). The slot will not be evicted if its cost is the
same as the current timestamp.
Differential Revision: D60812956