Update ROCm base docker images to focal (ubuntu20.04)#79596
Update ROCm base docker images to focal (ubuntu20.04)#79596jithunnair-amd wants to merge 2 commits intopytorch:masterfrom
Conversation
🔗 Helpful links
❌ 2 New Failures, 1 Flaky FailuresAs of commit 8383e0d (more details on the Dr. CI page): Expand to see more
🕵️ 2 new failures recognized by patternsThe following CI failures do not appear to be due to upstream breakages
|
|
@seemethere I think you will need to create new docker tags in the registry? Please create |
I'll add the images. |
|
Created pytorch-linux-focal-rocm5.0-py3.7 and pytorch-linux-focal-rocm5.1-py3.7 |
|
@kit1980 @malfet @seemethere I'm not sure why the |
As a temporary workaround to unblock you, try changing something inside |
|
@pytorchbot rebase |
|
@pytorchbot successfully started a rebase job. Check the current status here |
|
Rebase failed due to Command Raised by https://github.com/pytorch/pytorch/actions/runs/2618842266 |
|
@jithunnair-amd please manually rebase (looks like there is a conflict of sorts) |
6430227 to
8383e0d
Compare
|
@malfet The ROCm docker build jobs succeeded. Merging this PR. Will file another PR to move the ROCm CI jobs to use focal images. |
|
@pytorchbot merge |
|
@pytorchbot successfully started a merge job. Check the current status here |
|
Merge failed due to Refusing to merge as mandatory check(s) pull failed for rule OSS CI |
|
@jithunnair-amd are you not concerned with the ROCm failure? |
|
@pytorchbot merge -f |
|
@pytorchbot successfully started a merge job. Check the current status here |
|
Hey @jithunnair-amd. |
There probably should have been a single PR to build rocm focal images and also switch CI to use them. There is now an inconsistency in the hard-coded image names in .circleci/docker/build.sh compared to the expected image names in the various workflows. The build.sh script is attempting to build a missing bionic image and the ROCM_VERSION variable is getting set incorrectly as "5.1" instead of "5.1.1". |
|
@malfet can you revert? |
|
@pytorchbot revert -m "Jeff asked for it" -c nosignal |
Sure, though I believe you should be able to issue revert command as well (please try next time and ping me if this is not the case) |
|
@jeffdaily Indeed I was, and was trying to figure out whether we had the upgrade for the docker images and the CI jobs in one PR last time. Didn't expect it would get landed :) |
|
@pytorchbot successfully started a revert job. Check the current status here |
|
@jithunnair-amd your PR has been successfully reverted. |
This reverts commit 8302bbe. Reverted #79596 on behalf of https://github.com/malfet due to Jeff asked for it
…81031) Re-attempting after original PR #79596 was reverted due to causing ROCm build failures Pull Request resolved: #81031 Approved by: https://github.com/jeffdaily, https://github.com/malfet
…81031) (#81031) Summary: Re-attempting after original PR #79596 was reverted due to causing ROCm build failures Pull Request resolved: #81031 Approved by: https://github.com/jeffdaily, https://github.com/malfet Test Plan: contbuild & OSS CI, see https://hud.pytorch.org/commit/pytorch/pytorch/8a5d9843ff5d5dd865fc922853a15b3e7e459fdb Reviewed By: mehtanirav Differential Revision: D37719967 Pulled By: mehtanirav fbshipit-source-id: 8be30b4fecb0dc2911661f6a5259e147f1726286
…#80015) CI doesn't have any MI25s anymore. Should improve docker and Pytorch build times in CI for ROCm. Will take out of Draft mode after #79596 is merged Pull Request resolved: #80015 Approved by: https://github.com/jeffdaily, https://github.com/malfet
…#80015) CI doesn't have any MI25s anymore. Should improve docker and Pytorch build times in CI for ROCm. Will take out of Draft mode after #79596 is merged Pull Request resolved: #80015 Approved by: https://github.com/jeffdaily, https://github.com/malfet
No description provided.