Implementations shall run on Centos 7.2 (64-bit) with CUDA version 7.5.  cuDNN v5 for CUDA 7.5 (May 12, 2016 release) will be installed on NIST machines for implementations using cuDNN.  Please follow the installation instructions below.

1. Download and install CentOS 7.2 - http://nigos.nist.gov:8080/evaluations/CentOS-7-x86_64-Everything-1511.iso.  A complete list of RPMs installed on NIST machines is available at http://nigos.nist.gov:8080/janus/environment/installed_rpms.txt.  Note: If you need a different version of a library than what NIST has installed, please provide it to NIST with your submission.
2. Download and install the EPEL repository - http://nigos.nist.gov:8080/janus/environment/epel-release-latest-7.noarch.rpm.
3. Download cuda-repo-rhel7-7-5-local-7.5-18.x86_64.rpm from https://developer.nvidia.com/cuda-downloads.
4. Install CUDA following these instructions - http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#redhat-installation.
5. Reboot your machine after CUDA installation.
6. (Optional) Download and install cuDNN v5 for CUDA 7.5, May 12, 2016 release (cudnn-7.5-linux-x64-v5.0-ga.tgz) from https://developer.nvidia.com/cudnn.

NOTE: Any additional library dependencies must be provided to NIST in a localized folder hierarchy:

<top level submission folder>/
|--- lib/
|      |-- all participant-supplied libraries
|--- config/
|      |-- all configuration and developer-defined data
|--- doc/ 
|      |-- any participant-provided documentation regarding the submission
|___ src/
       |__ all source code