-
Notifications
You must be signed in to change notification settings - Fork 26.3k
ci: Add check-tpu composite action for TPU detection #170269
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/170269
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 0153132 with merge base 8ebbfb1 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
|
Starting merge as part of PR stack under #170267 |
1 similar comment
|
Starting merge as part of PR stack under #170267 |
|
Starting merge as part of PR stack under #170270 |
|
Starting merge as part of PR stack under #170267 |
These are needed in order to ensure that TPUs can work correctly in our DinD setup. Signed-off-by: Eli Uriegas <[email protected]> Pull Request resolved: #170270 Approved by: https://github.com/oulgen ghstack dependencies: #170272, #170269
has_profile check wasn't working which explains the ROCm issues with oidc. Also removes the cross-account login. I did a sweep and I don't think we maintain an ECR on the lf account so this isn't actually needed Signed-off-by: Eli Uriegas <[email protected]> Pull Request resolved: #170267 Approved by: https://github.com/huydhn, https://github.com/zxiiro ghstack dependencies: #170272, #170269, #170270
Add a composite action that detects whether the current runner has a TPU connected. Uses tpu-info library to query local TPU chips via PCI device scanning. Outputs `has_tpu` as true/false for use in workflow conditionals. Requires setup-linux to have been run first (for uv availability). Signed-off-by: Eli Uriegas <[email protected]> Pull Request resolved: pytorch#170269 Approved by: https://github.com/oulgen ghstack dependencies: pytorch#170272
These are needed in order to ensure that TPUs can work correctly in our DinD setup. Signed-off-by: Eli Uriegas <[email protected]> Pull Request resolved: pytorch#170270 Approved by: https://github.com/oulgen ghstack dependencies: pytorch#170272, pytorch#170269
has_profile check wasn't working which explains the ROCm issues with oidc. Also removes the cross-account login. I did a sweep and I don't think we maintain an ECR on the lf account so this isn't actually needed Signed-off-by: Eli Uriegas <[email protected]> Pull Request resolved: pytorch#170267 Approved by: https://github.com/huydhn, https://github.com/zxiiro ghstack dependencies: pytorch#170272, pytorch#170269, pytorch#170270
Stack from ghstack (oldest at bottom):
Add a composite action that detects whether the current runner has a TPU
connected. Uses tpu-info library to query local TPU chips via PCI device
scanning. Outputs
has_tpuas true/false for use in workflow conditionals.Requires setup-linux to have been run first (for uv availability).
Signed-off-by: Eli Uriegas [email protected]