Incorrect Reduction Arguments in third_party/xla/xla/experiments/triton_autotuning/matmul_lib.py

[K-Wu]

Issue type

Bug

Have you reproduced the bug with TensorFlow Nightly?

Yes

Source

source

TensorFlow version

latest

Custom code

No

OS platform and distribution

No response

Mobile device

No response

Python version

No response

Bazel version

No response

GCC/compiler version

No response

CUDA/cuDNN version

No response

GPU model and memory

No response

Current behavior?

In https://github.com/tensorflow/tensorflow/blob/master/third_party/xla/xla/experiments/triton_autotuning/matmul_lib.py#L301-L309, should the row_size be int(dims.M)*int(dims.N) instead?

Standalone code to reproduce the issue

No need.

Relevant log output

No response

[tilakrayal]

@K-Wu, Could you please share a reproducible code or the use-case that supports your statement so that the issue can be easily understood? Thank you!

[K-Wu]

Hi @tilakrayal ,

I don't have a reproducible code as this is not the core part of tensorflow. Besides, it should be fairly easy to tell by reading this argument. Could you please let the owner (maybe @gflegar) briefly check it?

Thank you.

Best Regards, Kun

[tilakrayal]

@K-Wu, I have raised the PR for the same changes. Could you please take a look at the same PR. https://github.com/tensorflow/tensorflow/pull/62885

Thank you!

[K-Wu]

Hi @tilakrayal, that works. Thank you!

[tilakrayal]

@K-Wu, As per the comment from the Developer in the PR raised, the requested change should happen in the https://github.com/openxla/xla repository. Could you please try to check with the respective repo as the file belongs to the xla api. Thank you!

[github-actions[bot]]

This issue is stale because it has been open for 7 days with no activity. It will be closed if no further activity occurs. Thank you.

[github-actions[bot]]

This issue was closed because it has been inactive for 7 days since being marked as stale. Please reopen if you'd like to work on this further.

[google-ml-butler[bot]]

Are you satisfied with the resolution of your issue? Yes No