Skip to content
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

enable windows cuda unit tests #4421

Merged
merged 37 commits into from
Sep 29, 2021
Merged

enable windows cuda unit tests #4421

merged 37 commits into from
Sep 29, 2021

Conversation

mszhanyi
Copy link
Contributor

@mszhanyi mszhanyi commented Sep 16, 2021

Only for unittest_windows_GPU
add nvjpeg in cuda_install
add cuda_install step in unittest_windows_gpu workflow
add cuda driver update step in unittest_windows_gpu workflow

CUDA Runtime , CUDA Driver and CUDA Environments are all necessary to run cuda tests.
It's hard to submit individual PRs.

cc @seemethere

@mszhanyi mszhanyi requested review from malfet, seemethere and NicolasHug and removed request for malfet September 16, 2021 03:07
@mszhanyi mszhanyi marked this pull request as draft September 16, 2021 08:03
@mszhanyi mszhanyi changed the title enable cuda for unittest_windows_GPU enable windows cuda unit tests Sep 18, 2021
@mszhanyi
Copy link
Contributor Author

mszhanyi commented Sep 18, 2021

@NicolasHug
I have little knowledge on vision. Could you please help to fix these test cases.
https://app.circleci.com/pipelines/github/pytorch/vision/10575/workflows/9eabc1d2-2659-413d-9580-ac3014693742/jobs/803604

@mszhanyi mszhanyi marked this pull request as ready for review September 18, 2021 07:09
@mszhanyi
Copy link
Contributor Author

mszhanyi commented Sep 22, 2021

@NicolasHug @fmassa
I updated the workflow cuda version to 11.1 and all tests can pass.
Does the pyd depend on cuda version?
cc @malfet
image

@fmassa
Copy link
Member

fmassa commented Sep 23, 2021

@mszhanyi if the pyd file in Windows is the equivalent of the .so in Linux, then yes, now image.so is dependent on the CUDA version as we have added support for nvjpeg, which comes with cudatoolkit.

@mszhanyi mszhanyi marked this pull request as draft September 26, 2021 07:27
@mszhanyi
Copy link
Contributor Author

mszhanyi commented Sep 27, 2021

All tests passed by rerunning but Workflow: build & Workflow: unittest status not updated

@mszhanyi mszhanyi marked this pull request as ready for review September 27, 2021 08:19
.circleci/unittest/windows/scripts/set_cuda_envs.sh Outdated Show resolved Hide resolved
.circleci/unittest/windows/scripts/set_cuda_envs.sh Outdated Show resolved Hide resolved
@malfet malfet merged commit 932ca5a into pytorch:main Sep 29, 2021
facebook-github-bot pushed a commit that referenced this pull request Sep 30, 2021
Summary:
* enable cuda for unittest

* add one check

* rename

* minor fix

* add more check

* add cuda envs

* add more cuda envs

* add driver update

* update vs integration

* get cuda version

* merge

* add envs in install.sh

* reduce some change and add comments

* minor improve

* fix typo

* exit if driver update failed

* fix lint

* fix lint

* Avoid catching exception to show error message

* try 11.1

* try 11.1

* "try 11.1"

* set the downloaded gpu driver dlls only for 10.2

* check torch.cuda.is_available()

* add display.driver

* Revert "Avoid catching exception to show error message"

This reverts commit 1ce58c5.

* Update .circleci/unittest/windows/scripts/set_cuda_envs.sh

* Update .circleci/unittest/windows/scripts/set_cuda_envs.sh

* Update .circleci/unittest/windows/scripts/set_cuda_envs.sh

Reviewed By: datumbox

Differential Revision: D31268020

fbshipit-source-id: a940a56ac02a5e440f4cbf8fc4b199e429c9002d

Co-authored-by: Nicolas Hug <nicolashug@fb.com>
Co-authored-by: Vasilis Vryniotis <datumbox@users.noreply.github.com>
Co-authored-by: Nikita Shulga <nikita.shulga@gmail.com>
Co-authored-by: Nikita Shulga <nshulga@fb.com>
malfet added a commit that referenced this pull request Oct 22, 2021
GPU tests have special step to install the driver, CUDA install step shouldl just install the runtime
never

Fixes regression introduced by #4421
malfet added a commit that referenced this pull request Oct 22, 2021
GPU tests have special step to install the driver, CUDA install step shouldl just install the runtime
never

Fixes regression introduced by #4421
facebook-github-bot pushed a commit that referenced this pull request Oct 26, 2021
Summary:
GPU tests have special step to install the driver, CUDA install step shouldl just install the runtime
never

Fixes regression introduced by #4421

Reviewed By: NicolasHug

Differential Revision: D31916340

fbshipit-source-id: 700b522fcfe58942603f8eb336811c585c28330e
cyyever pushed a commit to cyyever/vision that referenced this pull request Nov 16, 2021
* enable cuda for unittest

* add one check

* rename

* minor fix

* add more check

* add cuda envs

* add more cuda envs

* add driver update

* update vs integration

* get cuda version

* merge

* add envs in install.sh

* reduce some change and add comments

* minor improve

* fix typo

* exit if driver update failed

* fix lint

* fix lint

* Avoid catching exception to show error message

* try 11.1

* try 11.1

* "try 11.1"

* set the downloaded gpu driver dlls only for 10.2

* check torch.cuda.is_available()

* add display.driver

* Revert "Avoid catching exception to show error message"

This reverts commit 1ce58c5.

* Update .circleci/unittest/windows/scripts/set_cuda_envs.sh

* Update .circleci/unittest/windows/scripts/set_cuda_envs.sh

* Update .circleci/unittest/windows/scripts/set_cuda_envs.sh

Co-authored-by: Nicolas Hug <nicolashug@fb.com>
Co-authored-by: Vasilis Vryniotis <datumbox@users.noreply.github.com>
Co-authored-by: Nikita Shulga <nikita.shulga@gmail.com>
Co-authored-by: Nikita Shulga <nshulga@fb.com>
cyyever pushed a commit to cyyever/vision that referenced this pull request Nov 16, 2021
GPU tests have special step to install the driver, CUDA install step shouldl just install the runtime
never

Fixes regression introduced by pytorch#4421
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

6 participants