The Power Machine Learning and Deep Learning Reference Architecture (Release 2)

look_w

Installing the system
Download and install the pre-requisites

Ubuntu 14.04, available for POWER at
NVIDIA CUDA 7.5 from
 Select Operating System: Linux
 Select Architecture: ppc64le
 Select Distribution Ubuntu
 Select Version 14.04
 Select the Installer Type that best fits your needs
 Follow the Linux installation instructions in the CUDA Quick Start Guide linked from the
download page, including the steps describing how to set up the CUDA development
environment by updating PATH and LD_LIBRARY_PATH.
. NVIDIA cuDNN version 3 from
Registration in NVIDIA's Accelerated Computing Developer Program is required.
4. Unpack the cuDNN libraries as follows:
$ cd /usr/local
$ sudo tar --no-same-owner -xzvf cudnn-7.0-linux-ppc64le-v3.0-prod.tgz
5. Add the IBM Machine Learning / Deep Learning repository:
$ sudo apt-get install software-properties-common
$ sudo apt-add-repository ppa:ibmpackages/mldl
6. Update the package cache to include packages in the MLDL repository
$ sudo apt-get update
7. If the system is not connected to the internet, two more steps are required
a. On an internet-connected machine, follow the steps above then download a copy
of the repository to be installed on non-connected machines:
$ apt-get download mldl-repo-local
b. Transfer the downloaded file mldl-repo-local*.deb to the non-connected
machine(s), and install the repository package locally and update the package
cache:
$ sudo dpkg -i mldl-repo-local*.deb
$ sudo apt-get update
Installing the MLDL Distro
Install all supplied frameworks at once with the following repository command:
$ sudo apt-get install power-mldl
Alternatively, install frameworks individually as follows.

Install the IBM Theano package:	$ sudo apt-get install theano
Install the IBM Caffe package:	$ sudo apt-get install caffe
Install the IBM Torch package:	$ su