How to run onednn convolution test on CPU

[akote123]

I wanted to run xla/service/cpu/tests/onednn_convolution_test.cc in cpu ,what is the command I have to use to run this example and also to verify the onednn path can I use ONEDNN_VERBOSE=1?

[penpornk]

cc: @TensorFlow-MKL

Hi @akote123,

oneDNN convolutions are currently only enabled in XLA:CPU's classic runtime. You use the classic runtime by setting the environment variable XLA_FLAGS=--xla_cpu_use_thunk_runtime=0. We are working on adding support to the current runtime as well. We'll update here when it's done.