introduce definitely_contiguous and use it for reshape and tensor meta data computation. #153432
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
[ghstack-poisoned]
This was referenced May 12, 2025
8000
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/153432
Note: Links to docs will display an error until the docs builds have been completed. ✅ You can merge normally! (3 Unrelated Failures)As of commit f11560c with merge base b8452e5 ( BROKEN TRUNK - The following jobs failed but were present on the merge base:👉 Rebase onto the `viable/strict` branch to avoid these failures
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
…tensor meta data computation. " when a tensor has unbacked symbols it can be general enough to represent both contiguous and non contiguous tensors. in that case we cant really evaluate is_contiguous. In many places in the code base, we check for is_contiguous to take a fast path. but the general path usually works for both contiguous and not contiguous in that case we probably want to use is_known _contiguous API. This is appleid for reshape in this PR and also to tensor meta data computation, the meta data now will have an attribute that says that its contiguous when its always contiguous. We would store that only if is_known _contiguous now. cc ezyang SherlockNoMad EikanWang jgong5 wenzhe-nrv [ghstack-poisoned]
…tensor meta data computation. " when a tensor has unbacked symbols it can be general enough to represent both contiguous and non contiguous tensors. in that case we cant really evaluate is_contiguous. In many places in the code base, we check for is_contiguous to take a fast path. but the general path usually works for both contiguous and not contiguous in that case we probably want to use is_known _contiguous API. This is appleid for reshape in this PR and also to tensor meta data computation, the meta data now will have an attribute that says that its contiguous when its always contiguous. We would store that only if is_known _contiguous now. cc ezyang SherlockNoMad EikanWang jgong5 wenzhe-nrv [ghstack-poisoned]
…tensor meta data computation. " when a tensor has unbacked symbols it can be general enough to represent both contiguous and non contiguous tensors. in that case we cant really evaluate is_contiguous. In many places in the code base, we check for is_contiguous to take a fast path. but the general path usually works for both contiguous and not contiguous in that case we probably want to use is_known _contiguous API. This is appleid for reshape in this PR and also to tensor meta data computation, the meta data now will have an attribute that says that its contiguous when its always contiguous. We would store that only if is_known _contiguous now. cc ezyang SherlockNoMad EikanWang jgong5 wenzhe-nrv [ghstack-poisoned]
…tensor meta data computation. " when a tensor has unbacked symbols it can be general enough to represent both contiguous and non contiguous tensors. in that case we cant really evaluate is_contiguous. In many places in the code base, we check for is_contiguous to take a fast path. but the general path usually works for both contiguous and not contiguous in that case we probably want to use is_known _contiguous API. This is appleid for reshape in this PR and also to tensor meta data computation, the meta data now will have an attribute that says that its contiguous when its always contiguous. We would store that only if is_known _contiguous now. cc ezyang SherlockNoMad EikanWang jgong5 wenzhe-nrv [ghstack-poisoned]
…tensor meta data computation. " when a tensor has unbacked symbols it can be general enough to represent both contiguous and non contiguous tensors. in that case we cant really evaluate is_contiguous. In many places in the code base, we check for is_contiguous to take a fast path. but the general path usually works for both contiguous and not contiguous in that case we probably want to use is_known _contiguous API. This is appleid for reshape in this PR and also to tensor meta data computation, the meta data now will have an attribute that says that its contiguous when its always contiguous. We would store that only if is_known _contiguous now. cc ezyang SherlockNoMad EikanWang jgong5 wenzhe-nrv [ghstack-poisoned]
…tensor meta data computation. " when a tensor has unbacked symbols it can be general enough to represent both contiguous and non contiguous tensors. in that case we cant really evaluate is_contiguous. In many places in the code base, we check for is_contiguous to take a fast path. but the general path usually works for both contiguous and not contiguous in that case we probably want to use is_known _contiguous API. This is appleid for reshape in this PR and also to tensor meta data computation, the meta data now will have an attribute that says that its contiguous when its always contiguous. We would store that only if is_known _contiguous now. cc ezyang SherlockNoMad EikanWang jgong5 wenzhe-nrv [ghstack-poisoned]
Contributor
|
@pytorchbot revert -m "Looks like it broke flex attention tests, see https://hud.pytorch.org/hud/pytorch/pytorch/main/1?per_page=50&name_filter=g6.4xlarge&mergeEphemeralLF=true" -c nosignal |
Collaborator
|
@pytorchbot successfully started a revert job. Check the current status here. |
Collaborator
|
@laithsakka your PR has been successfully reverted. |
pytorchmergebot
added a commit
that referenced
this pull request
May 27, 2025
…nsor meta data computation. (#153432)" This reverts commit 5c6d7ca. Reverted #153432 on behalf of https://github.com/malfet due to Looks like it broke flex attention tests, see https://hud.pytorch.org/hud/pytorch/pytorch/main/1?per_page=50&name_filter=g6.4xlarge&mergeEphemeralLF=true ([comment](#153432 (comment)))
… tensor meta data computation. " when a tensor has unbacked symbols it can be general enough to represent both contiguous and non contiguous tensors. in that case we cant really evaluate is_contiguous. In many places in the code base, we check for is_contiguous to take a fast path. but the general path usually works for both contiguous and not contiguous in that case we probably want to use definitely _contiguous API. This is appleid for reshape in this PR and also to tensor meta data computation, the meta data now will have an attribute that says that its contiguous when its always contiguous. We would store that only if definitely _contiguous is true now. cc ezyang SherlockNoMad EikanWang jgong5 wenzhe-nrv [ghstack-poisoned]
… tensor meta data computation. " when a tensor has unbacked symbols it can be general enough to represent both contiguous and non contiguous tensors. in that case we cant really evaluate is_contiguous. In many places in the code base, we check for is_contiguous to take a fast path. but the general path usually works for both contiguous and not contiguous in that case we probably want to use definitely _contiguous API. This is appleid for reshape in this PR and also to tensor meta data computation, the meta data now will have an attribute that says that its contiguous when its always contiguous. We would store that only if definitely _contiguous is true now. cc ezyang SherlockNoMad EikanWang jgong5 wenzhe-nrv voznesenskym penguinwu Guobing-Chen XiaobingSuper zhuhaozhe blzheng jiayisunx ipiszy chenyang78 kadeng muchulee8 amjames chauhang aakhundov [ghstack-poisoned]
… tensor meta data computation. " when a tensor has unbacked symbols it can be general enough to represent both contiguous and non contiguous tensors. in that case we cant really evaluate is_contiguous. In many places in the code base, we check for is_contiguous to take a fast path. but the general path usually works for both contiguous and not contiguous in that case we probably want to use definitely _contiguous API. This is appleid for reshape in this PR and also to tensor meta data computation, the meta data now will have an attribute that says that its contiguous when its always contiguous. We would store that only if definitely _contiguous is true now. cc ezyang SherlockNoMad EikanWang jgong5 wenzhe-nrv voznesenskym penguinwu Guobing-Chen XiaobingSuper zhuhaozhe blzheng jiayisunx ipiszy chenyang78 kadeng muchulee8 amjames chauhang aakhundov [ghstack-poisoned]
Contributor
Author
|
rebase and fix inductor out of bounds bug |
pianpwk
reviewed
May 27, 2025
|
|
||
| if a.is_contiguous(): | ||
| if definitely_contiguous(a): | ||
| return a.as_strided(shape, utils.make_contiguous_strides_for(shape)) |
Contributor
There was a problem hiding this comment.
also apply this to
pytorch/torch/_refs/__init__.py
Line 3981 in b8452e5
Contributor
Author
There was a problem hiding this comment.
in the next PR we should avoid hitting that completely thats why i did not mark it https://github.com/pytorch/pytorch/pulls/laithsakka
… tensor meta data computation. " when a tensor has unbacked symbols it can be general enough to represent both contiguous and non contiguous tensors. in that case we cant really evaluate is_contiguous. In many places in the code base, we check for is_contiguous to take a fast path. but the general path usually works for both contiguous and not contiguous in that case we probably want to use definitely _contiguous API. This is appleid for reshape in this PR and also to tensor meta data computation, the meta data now will have an attribute that says that its contiguous when its always contiguous. We would store that only if definitely _contiguous is true now. cc ezyang SherlockNoMad EikanWang jgong5 wenzhe-nrv voznesenskym penguinwu Guobing-Chen XiaobingSuper zhuhaozhe blzheng jiayisunx ipiszy chenyang78 kadeng muchulee8 amjames chauhang aakhundov [ghstack-poisoned]
Collaborator
|
Starting merge as part of PR stack under #153441 |
pytorchmergebot
pushed a commit
that referenced
this pull request
May 28, 2025
…153441) * This verifies that the check short circuit is not material. #153431 ``` import torch from torch.export import Dim, export class MyModel(torch.nn.Module): def forward(self, x, ranks): first_k = ranks.max().item() torch._check_is_size(first_k) narrow = x.narrow(dim = 1, start = 0, length = first_k) lt = narrow < narrow.size(1) return lt inps = ( torch.randn((8, 16), device="cuda"), torch.arange(8, device="cuda", dtype=torch.int8) ) spec = { "x": (Dim.AUTO, Dim.AUTO), "ranks": (Dim.AUTO,), } traced = export(MyModel(), inps, dynamic_shapes=spec, strict=True).run_decompositions({}) ``` Pull Request resolved: #153441 Approved by: https://github.com/jansel ghstack dependencies: #153432
Contributor
|
@laithsakka this change: expected_stride = expected_stride * sym_max(x, 1)breaks the |
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):
when a tensor has unbacked symbols it can be general enough to represent both contiguous and non contiguous tensors.
in that case we cant really evaluate is_contiguous. In many places in the code base, we check for is_contiguous to take a fast path. but the general path usually works for both contiguous and not contiguous in that case we probably want
to use definitely _contiguous API.
This is appleid for reshape in this PR and also to tensor meta data computation, the meta data now will have an attribute that says that its contiguous when its always contiguous. We would store that only if definitely _contiguous is true now.
cc @ezyang @SherlockNoMad @EikanWang @jgong5 @wenzhe-nrv @voznesenskym @penguinwu @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @jiayisunx @ipiszy @chenyang78 @kadeng @muchulee8 @amjames @chauhang @aakhundov