Skip to content

Conversation

@SS-JIA
Copy link
Contributor

@SS-JIA SS-JIA commented Jul 3, 2025

Stack from ghstack (oldest at bottom):

Summary:

Changes

  • In partitioner, check if buffer storage can be used only after it has been determined that no valid texture layouts are available

Context

Currently, the logic in the vulkan partitioner is incorrect.

  1. First, it checks what texture layouts may be used to represent the tensors involved in the computation
  2. If no texture layouts are available, it checks if buffer support is available and the tensors are small enough to be within Vulkan buffer limits
  3. Then, it checks if all valid texture layouts are supported by the op.

This introduces a bug in situations where 3 fails, but 2 would pass. However, 2 is not checked due to the way the logic is structured.

The fix is to switch the order of 2 and 3.

Test Plan:

Test Plan

Manual verification + CI

[ghstack-poisoned]
SS-JIA pushed a commit that referenced this pull request Jul 3, 2025
Summary:
## Changes

* In partitioner, check if buffer storage can be used only after it has been determined that no valid texture layouts are available

## Context

Currently, the logic in the vulkan partitioner is incorrect.

1. First, it checks what texture layouts may be used to represent the tensors involved in the computation
2. If no texture layouts are available, it checks if buffer support is available and the tensors are small enough to be within Vulkan buffer limits
3. Then, it checks if all valid texture layouts are supported by the op.

This introduces a bug in situations where 3 fails, but 2 would pass. However, 2 is not checked due to the way the logic is structured.

The fix is to switch the order of 2 and 3.


Test Plan:
## Test Plan

Manual verification + CI

ghstack-source-id: 9ccc526
Pull Request resolved: #12194
@pytorch-bot
Copy link

pytorch-bot bot commented Jul 3, 2025

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/12194

Note: Links to docs will display an error until the docs builds have been completed.

This comment was automatically generated by Dr. CI and updates every 15 minutes.

@facebook-github-bot facebook-github-bot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Jul 3, 2025
@github-actions
Copy link

github-actions bot commented Jul 3, 2025

This PR needs a release notes: label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

@SS-JIA SS-JIA closed this Jul 3, 2025
@SS-JIA SS-JIA had a problem deploying to cherry-pick-bot July 3, 2025 14:14 — with GitHub Actions Failure
@SS-JIA SS-JIA deleted the gh/SS-JIA/251/head branch October 15, 2025 17:59
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants