[MPS] Fix conv backward pass for channels last #141009
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/141009
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: ✅ You can merge normally! (1 Unrelated Failure)As of commit ed563b4 with merge base a440a01 ( FLAKY - The following job failed but was likely due to flakiness present on trunk:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
pytorch-bot
bot
added
ciflow/mps
Looks like a regression caused by use of strided API. Fixed by ignoring memory layout, as for strided API it shoudl not matter, shoudl it? Fixes #140902
malfet
force-pushed
the
malfet/fix-conv-backward-cl
branch
from
b4c3618 to
d902d9a
Compare
|
@pytorchbot merge -f "Lint + MPS 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). Please use Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
Looks like a regression caused by use of strided API, but adding the test revealed (at least in CI), that on Ventura it worked but returned garbage results, so fixed by removing all the logic about channels last (as it's irrelevant for strided API case and placeholder already turns tensor into a correct one) This also allows one to remove `mem_format_key` and `ns_shape_key` (it was redundant even back then, as `mem_format_key` + `getTensorsStringKey(grad_output_t)` already uniquely identified the operation) Fixes pytorch#140902 Pull Request resolved: pytorch#141009 Approved by: https://github.com/manuelcandales
3 participants
Looks like a regression caused by use of strided API, but adding the test revealed (at least in CI), that on Ventura it worked but returned garbage results, so fixed by removing all the logic about channels last (as it's irrelevant for strided API case and placeholder already turns tensor into a correct one)
This also allows one to remove
mem_format_keyandns_shape_key(it was redundant even back then, asmem_format_key+getTensorsStringKey(grad_output_t)already uniquely identified the operation)Fixes #140902