-
Notifications
You must be signed in to change notification settings - Fork 22.1k
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
Fix DLPack CUDA stream convention #67618
Conversation
CI Flow Status⚛️ CI FlowRuleset - Version:
You can add a comment to the PR and tag @pytorchbot with the following commands: # ciflow rerun, "ciflow/default" will always be added automatically
@pytorchbot ciflow rerun
# ciflow rerun with additional labels "-l <ciflow/label_name>", which is equivalent to adding these labels manually and trigger the rerun
@pytorchbot ciflow rerun -l ciflow/scheduled -l ciflow/slow For more information, please take a look at the CI Flow Wiki. |
🔗 Helpful links
💊 CI failures summary and remediationsAs of commit c4494d0 (more details on the Dr. CI page): 💚 💚 Looks good so far! There are no failures yet. 💚 💚 This comment was automatically generated by Dr. CI (expand for details).Please report bugs/suggestions to the (internal) Dr. CI Users group. |
f1eabae
to
f42295b
Compare
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.
Thanks, Emilio! Look like we missed stream_ptr
here?
06a8985
to
8114b16
Compare
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.
This LGTM, thanks @emcastillo!
Hey @emcastillo! Thanks for this fix and your patience (I was moving from the West Coast to the East Coast)! It looks like the ROCm failure is real, however:
Any idea what's going on there? One option would be to file a follow-up issue and skip the test on ROCm for now (with a link to the issue). |
Stream numbering for CUDA and ROCm is different, it's I have a vague memory about us talking about this before, and there being an issue with detecting whether we're running on ROCm, but I can't find it back. |
Thanks @mruberry, ler me take a closer look and fix it! |
b72a634
to
c4494d0
Compare
@mruberry all tests passed :) |
@mruberry has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator. |
Apparently for the array API, cuda default stream and per thread stream should be 1 and 2 instead of 0 and 1:
https://data-apis.org/array-api/latest/API_specification/array_object.html?dlpack-self-stream-none#dlpack-self-stream-none.
This caused a problem in the interop with CuPy cupy/cupy#5970 (comment).
cc @rgommers @leofang @mruberry