How to turn off mlir during tensorflow2.13 compilation？

[s1mplezpc]

Issue type

Support

Have you reproduced the bug with TensorFlow Nightly?

No

Source

source

TensorFlow version

r2.13-rocm-edhanced-57

Custom code

Yes

OS platform and distribution

ubuntu 18.04

Mobile device

No response

Python version

3.9

Bazel version

No response

GCC/compiler version

No response

CUDA/cuDNN version

No response

GPU model and memory

No response

Current behavior?

I want to turn off mlir during the tensorflow compilation process

Standalone code to reproduce the issue

/tensorflow-upstream/tensorflow/core/kernels/mlir_generated/BUILD:1776:19: Generating kernel '//tensorflow/core/kernels/mlir_generated:sub_gpu_sub_kernels_gpu_i64_i64_kernel_generator' failed: (Aborted): tf_to_kernel failed: error executing command

Relevant log output

No response

[harkgill-amd]

Hi @s1mplezpc, apologies for the lack of response.

Are you still seeing MLIR related build failures with ROCm 6.2.4 and a newer branch such as r2.18-rocm-enhanced? If so, please fill out the remaining fields in the issue report so we can further investigate the build issues on our side.

There have been a couple fixes implemented since the opening of this ticket that address similar MILIR issues. It’s possible that one of these fixes has resolved the error you encountered, making it necessary to update with a rebase.