-
Notifications
You must be signed in to change notification settings - Fork 26.3k
[MPS] Add Unique and unique_consecutive ops. #88532
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/88532
Note: Links to docs will display an error until the docs builds have been completed. ❌ 1 FailuresAs of commit 14944cc: FLAKY - The following jobs failed but were likely due to flakiness present on master:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
|
|
|
@malfet , rebased . Please give it a review. |
|
/easycla |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks good to me.
Major areas of concern:
- Large tensors (add overflow checks when computing number of elements)
- For the CPU fallbacks, aren't results needed to be copied back to MPS?
Nits:
- Formatting:
if ()block should always be surrounded by{and don't one needs spaces after objc argument name? - std::array vs NSArray (this way one can avoid
[NSNull null])
|
/easycla |
|
@pytorchbot merge -g |
Merge startedYour change will be merged once all checks on your PR pass since you used the green (-g) flag (ETA: 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
|
I'm building this branch myself, and I get this warning an awful lot: on my macos 13.2 M1. I'm building the usual recommended pytorch way: |
Merge failedReason: 2 additional jobs have failed, first few of them are: trunk ,trunk / linux-focal-rocm5.3-py3.8 / test (default, 1, 2, linux.rocm.gpu) Details for Dev Infra teamRaised by workflow job |
|
@pytorchbot merge -f "Lint+MPS is 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 |
Merge failedReason: Command Details for Dev Infra teamRaised by workflow job |
Add check for macos 13.0
|
@pytorchbot merge -g |
Merge startedYour change will be merged once all checks on your PR pass since you used the green (-g) flag (ETA: 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
Merge failedReason: 2 additional jobs have failed, first few of them are: trunk ,trunk / linux-focal-rocm5.3-py3.8 / test (default, 2, 2, linux.rocm.gpu) Details for Dev Infra teamRaised by workflow job |
|
@pytorchbot merge -f "Lint+MPS is 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 |
|
@jkoudys this change should get into the next pytorch nightly binary - could you please try that and let me know if you see any issues? |
|
Is this merged on master? It looks like it is, but I am unable to use the op in yolov5 and pytorch 1.13.1, and I have tested on the nightly too. |
|
@tj-safely nightly builds should have them. Can you try doing a force reinstall to make sure it got installed properly? |
|
@DenisVieriu97 I have tried a few different methods for installing a matching torchvision package to no success. Right now I have this: And get this error message: When I add torchvision to your command or try with only torchvision, it attempts to install many different versions of torchvision and does not install any of them. Finally, I have this torchvision version installed by running this command: which I found from this link: Thank you for your help and further guidance is appreciated. |
|
@tj-safely try uninstalling all previous packages of torch/torchvision/torchaudio, then do a clean install. then: |
|
I am still unable to install torchvision this way. To reproduce:
Everything works successfully until the last command: |
|
@tj-safely could you please try installing only torchvision and torch? |
This worked. For some reason mps is sometimes slower than cpu on yolov5, but I do not get exceptions anymore. Thank you so much for you help! |
And use convenient methods TODO was added by an accidental copy-n-paste of code from #82315 into #88532 Pull Request resolved: #118902 Approved by: https://github.com/kit1980
And use convenient methods TODO was added by an accidental copy-n-paste of code from #82315 into #88532 Pull Request resolved: #118902 Approved by: https://github.com/kit1980
Add check for macos 13.0
Fixes #88487