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
OpenCL GPU support issue #19557
Comments
I would recommend reaching out to codeplay. OpenCL support for TF is still in early stages, and there may be some additional constraints to use it. |
OpenCL support is integrating into 1.8: |
@Kenop - Is this still an issue ? Feel free to close if this doesn't exist with the latest tensorflow version. |
Just for reference, this branch with TensorFlow 1.8 was only experimental. |
Closing as duplicate of #22 |
System information
Describe the problem
I built TF 1.8 from source with 'y' for OpenCL SYCL and ComputeCPP support in build configuration ('n' for other options). Build was complete successfully, but GPU device seems unavailable in TF, only CPU:0 is listed by
tensorflow.python.client.device_lib.device_lib.list_local_devices()
and
tensorflow.contrib.eager.list_devices()
Also,
tensorflow.matmul(a, b)
is placed on CPU:0, log_device_placement says: "Device mapping: no known devices."bazel test -c opt --config=sycl --test_output=all //tensorflow/python/kernel_tests:basic_gpu_test
fails with errors from SYCL.
Bazel test, clinfo and computecpp_info logs are listed below.
Requirements for OpenCL GPU support seem to be met, but it doesn't work. Am I missing something?
Source code / logs
clinfo
/usr/local/computecpp/bin/computecpp_info
bazel test -c opt --config=sycl --test_output=all //tensorflow/python/kernel_tests:basic_gpu_test
The text was updated successfully, but these errors were encountered: