Replace 11.1 with 11.2 on CI for Windows #51598
Closed
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
janeyx99
requested review from
peterjc123 and
mszhanyi
and removed request for
peterjc123
💊 CI failures summary and remediationsAs of commit a62140a (more details on the Dr. CI page):
🕵️ 3 new failures recognized by patternsThe following CI failures do not appear to be due to upstream breakages:
|
| Job | Step | Action |
|---|---|---|
 binary_windows_wheel_3_9_cpu_nightly_build |
Build | 🔁 rerun |
| binary_windows_wheel_3_9_cu112_nightly_build |
Build | 🔁 rerun |
| binary_windows_wheel_3_9_cu101_nightly_build |
Build | 🔁 rerun |
1 job timed out:
pytorch_linux_xenial_cuda11_1_cudnn8_py3_gcc7_test
This comment was automatically generated by Dr. CI (expand for details).
Follow this link to opt-out of these comments for your Pull Requests.Please report bugs/suggestions to the (internal) Dr. CI Users group.
janeyx99
force-pushed
the
ci-all/replace-windows-11.1-ci-with-11.2
branch
from
8d4498c to
74a76ee
Compare
peterjc123
reviewed
Feb 3, 2021
The windows tests errors don't appear in commits on master, though. |
janeyx99
force-pushed
the
ci-all/replace-windows-11.1-ci-with-11.2
branch
from
74a76ee to
db380bc
Compare
walterddr
reviewed
Feb 4, 2021
|
@janeyx99 magma for cuda11.2 size is smaller, Is it right? |
Yes, I believe CUDA 11.2 is smaller than CUDA 11.1. |
walterddr
approved these changes
Feb 8, 2021
janeyx99
force-pushed
the
ci-all/replace-windows-11.1-ci-with-11.2
branch
from
db380bc to
357d627
Compare
|
After rebasing and rerunning, I don't think the |
|
I reproduced the failure of |
|
I opened an issue at #51980 for the failing test to track it separately. |
There was a problem hiding this comment.
@janeyx99 has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
janeyx99
force-pushed
the
ci-all/replace-windows-11.1-ci-with-11.2
branch
from
357d627 to
6adc4e0
Compare
janeyx99
force-pushed
the
ci-all/replace-windows-11.1-ci-with-11.2
branch
from
becef8d to
8f88054
Compare
zasdfgbnm
reviewed
Feb 9, 2021
| @@ -12685,6 +12685,7 @@ def test_per_sample_weights(mode, trainable_scale): | |||
| for mode, trainable in itertools.product(modes, trainable_scale): | |||
| test_per_sample_weights(mode, trainable) | |||
|
|
|||
| @unittest.skipIf(IS_WINDOWS, "FIXME: CUDA error: too many resources requested on 11.2") | |||
There was a problem hiding this comment.
Could we also have a issue for this one, and reference the issue number here?
There was a problem hiding this comment.
#52002 yes I made one here (linked in description)
There was a problem hiding this comment.
@janeyx99 has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
facebook-github-bot
pushed a commit
that referenced
this pull request
Mar 2, 2021
Summary: The previous code allowed these tests to run every four hours on certain ci-all branches...which is really bad and resource intensive. This code removes that, but then disallows the 11.2 and 9.2 tests to be run on ci-all branches. To debug CUDA 11.2 or 9.2 tests, one must now manually change the config to allow for them. (Look at #51888 and #51598 for examples of how to do that.) Pull Request resolved: #53069 Reviewed By: H-Huang Differential Revision: D26739738 Pulled By: janeyx99 fbshipit-source-id: 7577b9b2e876bac0e4e868ce2a1f3ffdb6aca597
aocsa
pushed a commit
to Quansight/pytorch
that referenced
this pull request
Mar 15, 2021
…3069) Summary: The previous code allowed these tests to run every four hours on certain ci-all branches...which is really bad and resource intensive. This code removes that, but then disallows the 11.2 and 9.2 tests to be run on ci-all branches. To debug CUDA 11.2 or 9.2 tests, one must now manually change the config to allow for them. (Look at pytorch#51888 and pytorch#51598 for examples of how to do that.) Pull Request resolved: pytorch#53069 Reviewed By: H-Huang Differential Revision: D26739738 Pulled By: janeyx99 fbshipit-source-id: 7577b9b2e876bac0e4e868ce2a1f3ffdb6aca597
|
@zasdfgbnm Did you come across them? Is it specific in cuda11.2 |
|
@mszhanyi No, it's caused by the VS upgrade, namely this commit microsoft/STL#1336 in STL. They forget to make the fallback to C_STD function for functions like |
xsacha
pushed a commit
to xsacha/pytorch
that referenced
this pull request
Mar 31, 2021
Summary: Adding CUDA 11.2 to Windows CI. Disabled tests: The following ran into `CUDA error: misaligned address` for CUDA 11.2: (issue linked below) `test_where_scalar_valid_combination_cuda_complex128` in test_torch.py `test_sgn_complex_cuda` in test_autograd.py The following ran into `CUDA error: too many resources requested for launch` for CUDA 11.2: (pytorch#52002) test_EmbeddingBag_per_sample_weights_and_new_offsets_cuda_int64_float64 test_EmbeddingBag_per_sample_weights_and_offsets_cuda_int64_float64 Pull Request resolved: pytorch#51598 Reviewed By: mrshenli Differential Revision: D26344965 Pulled By: janeyx99 fbshipit-source-id: 3c9a4ed16d748969e96593220ec0a9f33e1ffcef
xsacha
pushed a commit
to xsacha/pytorch
that referenced
this pull request
Mar 31, 2021
…3069) Summary: The previous code allowed these tests to run every four hours on certain ci-all branches...which is really bad and resource intensive. This code removes that, but then disallows the 11.2 and 9.2 tests to be run on ci-all branches. To debug CUDA 11.2 or 9.2 tests, one must now manually change the config to allow for them. (Look at pytorch#51888 and pytorch#51598 for examples of how to do that.) Pull Request resolved: pytorch#53069 Reviewed By: H-Huang Differential Revision: D26739738 Pulled By: janeyx99 fbshipit-source-id: 7577b9b2e876bac0e4e868ce2a1f3ffdb6aca597
github-actions
bot
deleted the
ci-all/replace-windows-11.1-ci-with-11.2
branch
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Adding CUDA 11.2 to Windows CI.
Disabled tests:
The following ran into
CUDA error: misaligned addressfor CUDA 11.2: (issue linked below)test_where_scalar_valid_combination_cuda_complex128in test_torch.pytest_sgn_complex_cudain test_autograd.pyThe following ran into
CUDA error: too many resources requested for launchfor CUDA 11.2: (#52002)test_EmbeddingBag_per_sample_weights_and_new_offsets_cuda_int64_float64
test_EmbeddingBag_per_sample_weights_and_offsets_cuda_int64_float64
The following ran into test assertion failures in test_optim.py (#51992)
test_adadelta
test_adam
test_adamw
test_multi_tensor_optimizers
test_rmsprop