Add Metal backend type definitions and utilities #15019
Open
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
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/15019
Note: Links to docs will display an error until the docs builds have been completed. ❗ 1 Active SEVsThere are 1 currently active SEVs. If your PR is affected, please view them below: ❌ 12 New Failures, 8 Unrelated FailuresAs of commit 7590e37 with merge base 9560800 ( NEW FAILURES - The following jobs have failed:
FLAKY - The following jobs failed but were likely due to flakiness present on trunk:
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. |
meta-cla
bot
added
the
CLA Signed
This was referenced Oct 10, 2025
manuelcandales
requested review from
larryliu0820 and
mergennachin
and removed request for
cccclai and
shoumikhin
mergennachin
requested changes
Oct 12, 2025
Comment on lines
22
to
23
| // Let's use 2 for MPS | ||
| return 2; |
There was a problem hiding this comment.
Don't you wanna be consistent with c10/core/DeviceType.h?
It is either 11 or 13
Comment on lines
+28
to
+29
| using AOTIRuntimeError = Error; | ||
| using AOTITorchError = Error; |
There was a problem hiding this comment.
Do you need to distinguish both errors? Seems like AOTIRuntimeError is not used
There was a problem hiding this comment.
Yes, AOTInductorModelContainerRun and family return AOTIRuntimeError while aoti_torch_empty_strided and family return AOTITorchError
Comment on lines
49
to
88
| // Utility function to convert sizes pointer to vector | ||
| std::vector<executorch::aten::SizesType> convert_sizes_to_vector( | ||
| int64_t ndim, | ||
| const int64_t* sizes_ptr) { | ||
| std::vector<executorch::aten::SizesType> sizes(ndim); | ||
| for (int i = 0; i < ndim; i++) { | ||
| sizes[i] = static_cast<executorch::aten::SizesType>(sizes_ptr[i]); | ||
| } | ||
| return sizes; | ||
| } | ||
|
|
||
| // Utility function to convert strides pointer to vector or calculate from sizes | ||
| std::vector<executorch::aten::StridesType> convert_strides_to_vector( | ||
| int64_t ndim, | ||
| const int64_t* sizes_ptr, | ||
| const int64_t* strides_ptr) { | ||
| std::vector<executorch::aten::StridesType> strides(ndim); | ||
|
|
||
| if (strides_ptr != nullptr) { | ||
| // Use provided strides. | ||
| for (int64_t i = 0; i < ndim; i++) { | ||
| strides[i] = static_cast<executorch::aten::StridesType>(strides_ptr[i]); | ||
| } | ||
| } else { | ||
| // Calculate strides from sizes. | ||
| if (ndim > 0) { | ||
| strides[ndim - 1] = static_cast<executorch::aten::StridesType>( | ||
| 1); // Last dimension has stride 1 | ||
| for (int64_t i = ndim - 2; i >= 0; i--) { | ||
| if (sizes_ptr[i + 1] == 0) { | ||
| strides[i] = strides[i + 1]; // Copy stride when size is 0 | ||
| } else { | ||
| strides[i] = static_cast<executorch::aten::StridesType>( | ||
| static_cast<int64_t>(strides[i + 1]) * sizes_ptr[i + 1]); | ||
| } | ||
| } | ||
| } | ||
| } | ||
| return strides; | ||
| } |
There was a problem hiding this comment.
Let's move these to a different common location. There's no metal specific logic here. Maybe backends/aoti/utils.h
Comment on lines
59
to
60
| std::vector<executorch::aten::SizesType> sizes, | ||
| std::vector<executorch::aten::StridesType> strides) { |
There was a problem hiding this comment.
this implementation will copy. maybe do this
const std::vector<executorch::aten::SizesType>& sizes,
const std::vector<executorch::aten::StridesType>& strides
mergennachin
approved these changes
Oct 13, 2025
manuelcandales
added a commit
to manuelcandales/executorch-1
that referenced
this pull request
Oct 14, 2025
Implement foundational types and utilities for Metal backend including: - AOTI type aliases (AOTITensorHandle, AOTIRuntimeError, AOTITorchError) - Device type handling functions - Tensor storage size queries - Tensor attribute utilities ghstack-source-id: 7bfa3ae ghstack-comment-id: 3392299883 Pull-Request: pytorch#15019
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
Implement foundational types and utilities for Metal backend including: