[MPS] Fix type casting copy with storage offset #95573
Conversation
pytorch-bot
bot
added
ciflow/mps
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/95573
Note: Links to docs will display an error until the docs builds have been completed. ⏳ No Failures, 1 PendingAs of commit 9a3d2fc: This comment was automatically generated by Dr. CI and updates every 15 minutes. |
ezyang
added
the
triaged
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
|
@pytorchbot merge |
Merge failedReason: New commits were pushed while merging. Please rerun the merge command. Details for Dev Infra teamRaised by workflow job |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
Merge failedReason: 1 jobs have failed, first few of them are: windows-binary-libtorch-release / libtorch-cpu-shared-with-deps-release-test Details for Dev Infra teamRaised by workflow job |
|
@pytorchbot rebase |
|
@pytorchbot successfully started a rebase job. Check the current status here |
|
Successfully rebased |
pytorchmergebot
force-pushed
the
fix_type_casting_offset_mm
branch
from
2d1a996
to
9a3d2fc
Compare
|
@pytorchbot merge -f "All tests are green." |
Merge startedYour change will be merged immediately since you used the force (-f) flag, bypassing any CI checks (ETA: 1-5 minutes). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
This PR handles the case where the `dst` tensor of type casting has a storage offset by creating a temporary buffer to store results and then copy them back to the dst with the offset added. Fixes #95417 Pull Request resolved: pytorch/pytorch#95573 Approved by: https://github.com/kulinseth
This PR handles the case where the `dst` tensor of type casting has a storage offset by creating a temporary buffer to store results and then copy them back to the dst with the offset added. Fixes #95417 Pull Request resolved: pytorch/pytorch#95573 Approved by: https://github.com/kulinseth
This PR handles the case where the `dst` tensor of type casting has a storage offset by creating a temporary buffer to store results and then copy them back to the dst with the offset added. Fixes #95417 Pull Request resolved: pytorch/pytorch#95573 Approved by: https://github.com/kulinseth
|
this regression existed in 1.13.1 hence not qualified for 2.0.1 release |
|
@atalman sorry if it's a dumb question since I don't really know how release cycles for pytorch works, but doesn't this fit under the category 2: 'Low risk critical fixes for silent correctness'? |
|
If qualified I can send a PR against the release branch. |
This PR handles the case where the `dst` tensor of type casting has a storage offset by creating a temporary buffer to store results and then copy them back to the dst with the offset added. Fixes pytorch#95417 Pull Request resolved: pytorch#95573 Approved by: https://github.com/kulinseth
This reverts commit b7c2a65.
This PR handles the case where the `dst` tensor of type casting has a storage offset by creating a temporary buffer to store results and then copy them back to the dst with the offset added. Fixes pytorch#95417 Pull Request resolved: pytorch#95573 Approved by: https://github.com/kulinseth
This PR handles the case where the
dsttensor of type casting has a storage offset by creating a temporary buffer to store results and then copy them back to the dst with the offset added.Fixes #95417