test_all.testbin show many long time tests

[BeamOfLight]

Issue summary

Hello!

I have problems with performance. Running the test_all.testbin show many long time tests like these:

[ RUN ] ScaleLayerTest/0.TestForwardScale
[ OK ] ScaleLayerTest/0.TestForwardScale (3 ms)
[ RUN ] ScaleLayerTest/0.TestForwardScaleAxis2
[ OK ] ScaleLayerTest/0.TestForwardScaleAxis2 (4 ms)
[ RUN ] ScaleLayerTest/0.TestGradientEltwise
[ OK ] ScaleLayerTest/0.TestGradientEltwise (2076 ms)
[ RUN ] ScaleLayerTest/0.TestGradientEltwiseWithParam
[ OK ] ScaleLayerTest/0.TestGradientEltwiseWithParam (201799 ms)

How can i fix it?
Do I need to install the driver ( for videocard Radeon Pro WX 7100 ) from http://support.amd.com/en-us/download/workstation?os=Linux+x86_64# or ROCm kernel already contains the required driver?

Steps to reproduce

Freshly compiled hipCaffe with default Makefile.config. Compilation flags = -O3
I am using the latest ROCm from debian packages. Running the test_all.testbin show problem with long time tests.

Your system configuration

Operating system: Ubuntu 16.04.2 LTS with 4.9.0-kfd-compute-rocm-rel-1.6-77
Compiler: GCC v5.4.0, HCC clang 5.0
CUDA version (if applicable): not applicable
CUDNN version (if applicable): not applicable
BLAS: ATLAS
Python or MATLAB version (for pycaffe and matcaffe respectively): not applicable

[parallelo]

Hello @BeamOfLight,

Thanks for giving us a heads-up on this -- definitely much appreciated. We'll take a look.

Best,

Jeff

[parallelo]

FYI - We're able to reproduce this and have started investigating. Thanks again for the heads-up.

Best,

Jeff