[dcp][hf] Improve HF consolidation algorithm #158648
Closed
+191
−350
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
Before we had a bunch of if-else cases based on sharding strategy to decide how to save the tensor with different logic for different strategies. This can be consolidated into one function that uses an algorithm to handle all cases by finding the max possible contiguous bytes that can be written Differential Revision: [D78489438](https://our.internmc.facebook.com/intern/diff/D78489438/) [ghstack-poisoned]
ankitageorge
added a commit
that referenced
this pull request
Jul 18, 2025
Before we had a bunch of if-else cases based on sharding strategy to decide how to save the tensor with different logic for different strategies. This can be consolidated into one function that uses an algorithm to handle all cases by finding the max possible contiguous bytes that can be written Differential Revision: [D78489438](https://our.internmc.facebook.com/intern/diff/D78489438/) ghstack-source-id: 297076053 Pull Request resolved: #158648
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/158648
Note: Links to docs will display an error until the docs builds have been completed. ❌ 4 New FailuresAs of commit 2c6ae1f with merge base f077c24 ( NEW FAILURES - The following jobs have failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
|
This pull request was exported from Phabricator. Differential Revision: D78489438 |
Before we had a bunch of if-else cases based on sharding strategy to decide how to save the tensor with different logic for different strategies. This can be consolidated into one function that uses an algorithm to handle all cases by finding the max possible contiguous bytes that can be written Differential Revision: [D78489438](https://our.internmc.facebook.com/intern/diff/D78489438/) cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta [ghstack-poisoned]
ankitageorge
added a commit
that referenced
this pull request
Jul 18, 2025
Pull Request resolved: #158648 Before we had a bunch of if-else cases based on sharding strategy to decide how to save the tensor with different logic for different strategies. This can be consolidated into one function that uses an algorithm to handle all cases by finding the max possible contiguous bytes that can be written ghstack-source-id: 297088500 @exported-using-ghexport Differential Revision: [D78489438](https://our.internmc.facebook.com/intern/diff/D78489438/)
|
This pull request was exported from Phabricator. Differential Revision: D78489438 |
Before we had a bunch of if-else cases based on sharding strategy to decide how to save the tensor with different logic for different strategies. This can be consolidated into one function that uses an algorithm to handle all cases by finding the max possible contiguous bytes that can be written Differential Revision: [D78489438](https://our.internmc.facebook.com/intern/diff/D78489438/) cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta [ghstack-poisoned]
This was referenced Jul 29, 2025
|
This pull request was exported from Phabricator. Differential Revision: D78489438 |
saumishr
reviewed
Aug 8, 2025
| and finding natural boundaries where contiguity breaks. | ||
| """ | ||
| # Start with elements remaining in the last dimension | ||
| max_contiguous = sub_tensor_shape[-1] - indices[-1] |
There was a problem hiding this comment.
Add some validations on the input lists here and log the appropriate errors to avoid index out of bounds errors later which become hard to debug.
| sub_tensor_shape[0] == tensor_shape[0] | ||
| and sub_tensor_shape[1] < tensor_shape[1] | ||
| ) | ||
| def _calculate_max_contiguous_elements( |
There was a problem hiding this comment.
This method needs robust UT.
Before we had a bunch of if-else cases based on sharding strategy to decide how to save the tensor with different logic for different strategies. This can be consolidated into one function that uses an algorithm to handle all cases by finding the max possible contiguous bytes that can be written Differential Revision: [D78489438](https://our.internmc.facebook.com/intern/diff/D78489438/) cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta [ghstack-poisoned]
ankitageorge
added a commit
that referenced
this pull request
Aug 8, 2025
Pull Request resolved: #158648 Before we had a bunch of if-else cases based on sharding strategy to decide how to save the tensor with different logic for different strategies. This can be consolidated into one function that uses an algorithm to handle all cases by finding the max possible contiguous bytes that can be written ghstack-source-id: 301722789 @exported-using-ghexport Differential Revision: [D78489438](https://our.internmc.facebook.com/intern/diff/D78489438/)
|
This pull request was exported from Phabricator. Differential Revision: D78489438 |
Before we had a bunch of if-else cases based on sharding strategy to decide how to save the tensor with different logic for different strategies. This can be consolidated into one function that uses an algorithm to handle all cases by finding the max possible contiguous bytes that can be written Differential Revision: [D78489438](https://our.internmc.facebook.com/intern/diff/D78489438/) cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta [ghstack-poisoned]
ankitageorge
added a commit
that referenced
this pull request
Aug 8, 2025
Pull Request resolved: #158648 Before we had a bunch of if-else cases based on sharding strategy to decide how to save the tensor with different logic for different strategies. This can be consolidated into one function that uses an algorithm to handle all cases by finding the max possible contiguous bytes that can be written ghstack-source-id: 301740422 @exported-using-ghexport Differential Revision: [D78489438](https://our.internmc.facebook.com/intern/diff/D78489438/)
|
This pull request was exported from Phabricator. Differential Revision: D78489438 |
saumishr
approved these changes
Aug 8, 2025
|
@pytorchmergebot merge -i |
Merge startedYour change will be merged while ignoring the following 4 checks: pull / linux-jammy-py3.13-clang12 / test (dynamo_wrapped, 2, 3, lf.linux.2xlarge), pull / linux-jammy-py3.13-clang12 / test (dynamo_wrapped, 1, 3, lf.linux.2xlarge), pull / linux-jammy-py3.9-clang12 / test (dynamo_wrapped, 1, 3, lf.linux.2xlarge), pull / linux-jammy-py3.9-clang12 / test (dynamo_wrapped, 2, 3, lf.linux.2xlarge) Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
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.
Stack from ghstack (oldest at bottom):
Before we had a bunch of if-else cases based on sharding strategy to decide how to save the tensor with different logic for different strategies. This can be consolidated into one function that uses an algorithm to handle all cases by finding the max possible contiguous bytes that can be written
Differential Revision: D78489438
cc @H-Huang @awgu @wanchaol @fegin @fduwjj @wz337 @wconstab @d4l3k @pragupta