AMD CPU mxnet performance benchmarks

[azai91]

For bugs or installation issues, please provide the following information.
Is it possible for someone at AMD to run benchmarks on AMD CPUs? To run benchmarks you just need to run the following:

python example/image-classification/benchmark_score.py

Can this benchmark be run on the default mxnet package and the pip mxnet-mkl package?

There is a thread to make the mxnet-mkl package the default mxnet package so we need to check that the performance changes for AMD users.
https://lists.apache.org/thread.html/ca3c858a712a66f4b3443dfb395d84e8fdf2c86e297bcd28a3a36ebd@%3Cdev.mxnet.apache.org%3E

Environment info

Operating System:
Linux

Compiler:

Package used (Python/R/Scala/Julia):
Python

MXNet version:
1.3.0 (Just pip mxnet and pip mxnet-mkl which will get 1.3.0)

Or if installed from source:

MXNet commit hash (git rev-parse HEAD):

If you are using python package, please provide
1.3.0 (Just pip mxnet and pip mxnet-mkl which will get 1.3.0)

Python version and distribution:

If you are using R package, please provide

R sessionInfo():

Error Message:

Please paste the full error message, including stack trace.

Minimum reproducible example

if you are using your own code, please provide a short script that reproduces the error.

Steps to reproduce

or if you are running standard examples, please provide the commands you have run that lead to the error.

1. pip install mxnet
2. Git clone repo to get benchmark script
3. On line 73 of the example/image-classification/benchmark_score.py remove gpus from the list. (Change the devs list to devs = [mx.cpu()])
4. Run the script python example/image-classification/benchmark_score.py
5. Repeat this with pip install mxnet-mkl

What have you tried to solve it?