[ExecuTorch] Avoid some extra heap allocations for TensorImplPtr #5711
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
With the previous approach, we put the sizes, dim_order, and strides vectors onto the heap, make_shared was forced to separately allocate the control block for the TensorImpl because it used a custom deleter, and then of course the TensorImpl itself is on the heap (total of 5 allocations). Now, we combine the vectors (but not their pointed-to data), deleter, control block, and TensorImpl into a single allocation, saving 4 allocations. Differential Revision: [D63493196](https://our.internmc.facebook.com/intern/diff/D63493196/) [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/5711
Note: Links to docs will display an error until the docs builds have been completed. ❌ 28 New FailuresAs of commit 5d14295 with merge base d2ba238 ( 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: D63493196 |
facebook-github-bot
added
the
CLA Signed
This was referenced Sep 26, 2024
swolchok
added a commit
that referenced
this pull request
Sep 26, 2024
With the previous approach, we put the sizes, dim_order, and strides vectors onto the heap, make_shared was forced to separately allocate the control block for the TensorImpl because it used a custom deleter, and then of course the TensorImpl itself is on the heap (total of 5 allocations). Now, we combine the vectors (but not their pointed-to data), deleter, control block, and TensorImpl into a single allocation, saving 4 allocations. Differential Revision: [D63493196](https://our.internmc.facebook.com/intern/diff/D63493196/) ghstack-source-id: 244960941 Pull Request resolved: #5711
…mplPtr" With the previous approach, we put the sizes, dim_order, and strides vectors onto the heap, make_shared was forced to separately allocate the control block for the TensorImpl because it used a custom deleter, and then of course the TensorImpl itself is on the heap (total of 5 allocations). Now, we combine the vectors (but not their pointed-to data), deleter, control block, and TensorImpl into a single allocation, saving 4 allocations. Differential Revision: [D63493196](https://our.internmc.facebook.com/intern/diff/D63493196/) [ghstack-poisoned]
|
This pull request was exported from Phabricator. Differential Revision: D63493196 |
swolchok
added a commit
that referenced
this pull request
Sep 26, 2024
Pull Request resolved: #5711 With the previous approach, we put the sizes, dim_order, and strides vectors onto the heap, make_shared was forced to separately allocate the control block for the TensorImpl because it used a custom deleter, and then of course the TensorImpl itself is on the heap (total of 5 allocations). Now, we combine the vectors (but not their pointed-to data), deleter, control block, and TensorImpl into a single allocation, saving 4 allocations. ghstack-source-id: 244967824 @exported-using-ghexport Differential Revision: [D63493196](https://our.internmc.facebook.com/intern/diff/D63493196/)
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):
With the previous approach, we put the sizes, dim_order, and
strides vectors onto the heap, make_shared was forced to separately
allocate the control block for the TensorImpl because it used a custom
deleter, and then of course the TensorImpl itself is on the heap
(total of 5 allocations). Now, we combine the vectors (but not their
pointed-to data), deleter, control block, and TensorImpl into a single
allocation, saving 4 allocations.
Differential Revision: D63493196