no matching function for call to 'MakeOneDnnStream'

[waarmond]

Issue type

Build/Install

Have you reproduced the bug with TensorFlow Nightly?

Yes

Source

source

TensorFlow version

2.16.1+ git

Custom code

No

OS platform and distribution

ArchLinux

Mobile device

No response

Python version

3.10

Bazel version

6.1.0

GCC/compiler version

No response

CUDA/cuDNN version

No response

GPU model and memory

No response

Current behavior?

ERROR: [.. bazel cache ..]/local_xla/xla/service/cpu/BUILD:1600:11: Compiling xla/service/cpu/onednn_matmul.cc failed: (Exit 1): clang-17 failed: error executing command (from target @local_xla//xla/service/cpu:onednn_matmul) /usr/bin/clang-17 -U_FORTIFY_SOURCE -fstack-protector -Wall -Wthread-safety -Wself-assign -Wunused-but-set-parameter -Wno-free-nonheap-object -fcolor-diagnostics -fno-omit-frame-pointer -g0 -O2 ... (remaining 220 arguments skipped)
external/local_xla/xla/service/cpu/onednn_matmul.cc:186:24: error: no matching function for call to 'MakeOneDnnStream'
  186 |   auto onednn_stream = MakeOneDnnStream(cpu_engine, thread_pool.get());
      |                        ^~~~~~~~~~~~~~~~
external/local_xla/xla/service/cpu/onednn_matmul.cc:146:14: note: candidate function not viable: no known conversion from 'pointer' (aka 'tsl::OneDnnThreadPool *') to 'dnnl::threadpool_interop::threadpool_iface *' for 2nd argument
  146 | dnnl::stream MakeOneDnnStream(
      |              ^
  147 |     const dnnl::engine& cpu_engine,
  148 |     dnnl::threadpool_interop::threadpool_iface* thread_pool) {
      |     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
external/local_xla/xla/service/cpu/onednn_matmul.cc:322:24: error: no matching function for call to 'MakeOneDnnStream'
  322 |   auto onednn_stream = MakeOneDnnStream(cpu_engine, thread_pool.get());
      |                        ^~~~~~~~~~~~~~~~
external/local_xla/xla/service/cpu/onednn_matmul.cc:146:14: note: candidate function not viable: no known conversion from 'pointer' (aka 'tsl::OneDnnThreadPool *') to 'dnnl::threadpool_interop::threadpool_iface *' for 2nd argument
  146 | dnnl::stream MakeOneDnnStream(
      |              ^
  147 |     const dnnl::engine& cpu_engine,
  148 |     dnnl::threadpool_interop::threadpool_iface* thread_pool) {
      |     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2 errors generated.

Not fixed by #64421 ENABLE_ONEDNN_V3

I've updated from onednn 3.1.1 to 3.4 to no avail.

Standalone code to reproduce the issue

same as in 63298

bazel build --repo_env=TF_PYTHON_VERSION=3.10 //tensorflow/tools/pip_package:wheel --repo_env=WHEEL_NAME=tf_nightly

Relevant log output

No response

[sushreebarsa]

@waarmond Could you please try with the following command and let us know;

bazel build --repo_env=TF_PYTHON_VERSION=3.10 //tensorflow/tools/pip_package:build_pip_package --repo_env=WHEEL_NAME=tf_nightly

Thank you!

[waarmond]

Duplicate of #63360

[google-ml-butler]

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