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

openxla pin update to 20250123 #8621

Open
wants to merge 6 commits into
base: master
Choose a base branch
from
Open

openxla pin update to 20250123 #8621

wants to merge 6 commits into from

Conversation

zpcore
Copy link
Collaborator

@zpcore zpcore commented Jan 24, 2025

Update openxla pin.

@zpcore
Copy link
Collaborator Author

zpcore commented Jan 24, 2025

The comment in WORKSPACE file in outdated. We are not using curl -L https://github.com/openxla/xla/archive/<git hash>.tar.gz | sha256sum to get the xla-hash anymore. xla-hash should be the openxla git commit hash.

@bhavya01 bhavya01 self-requested a review January 24, 2025 17:59
@bhavya01
Copy link
Collaborator

The TPU tests didn't run due to missing symbol error. I thought the CI uses locally built wheels and we shouldn't see this error here. Can you confirm what's the issue?

@zpcore
Copy link
Collaborator Author

zpcore commented Jan 24, 2025

The TPU tests didn't run due to missing symbol error. I thought the CI uses locally built wheels and we shouldn't see this error here. Can you confirm what's the issue?

The TPU CI is using pip install for torch_xla:

Downloading torch_xla-2.5.1-cp310-cp310-manylinux_2_28_x86_64.whl (90.6 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 90.6/90.6 MB 96.9 MB/s eta 0:00:00

I think we suppose to use local build.

@tengyifei tengyifei removed the tpuci label Jan 25, 2025
@zpcore zpcore requested a review from ManfeiBai as a code owner January 27, 2025 05:31
@zpcore
Copy link
Collaborator Author

zpcore commented Jan 27, 2025

The test

 python -m pytest test/test_tpu_paged_attention_kernel.py -v -k "test_paged_attention_without_query_padding279"

Keeps failing after the pin update. I narrow down the version issue:
jax-0.4.39.dev20250113, jaxlib-0.4.39.dev20250113 --> Pass
jax-0.4.39.dev20250114, jaxlib-0.4.39.dev20250114 --> Fail

Hi @vanbasten23, I saw you added the test. I feel like this is probably due to JAX made an update to the pallas kernel. Just FYI in case you see the issue in the future.

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.

3 participants