Installing the deep learning frameworks

Find instructions for installing the deep learning framework on an IBM® Power® System server.

Setting up the software repository

The PowerAI deep learning packages are distributed in a tar.gz file that contains an rpm and a readme file. The tar.gz file must be extracted on the local system. Installing the rpm creates an installation repository on the local system.

Install the repository package by running the following command:

sudo rpm -ihv mldl-repo-*.rpm

Installing all frameworks at the same time

All the Deep Learning frameworks can be installed at the same time by using the power-mldl meta-package:

sudo yum install power-mldl

Installing the Python 3 versions of the frameworks

The Python 3 versions of the frameworks can be installed at the same time by using the power-mldl-py3 meta-package.

sudo yum install power-mldl-py3

Installing frameworks individually

You can install the deep learning frameworks individually. The framework packages include the following versions:

Table 1. Framework packages
Package	Description	Version
`caffe-bvlc`	Berkeley Vision and Learning Center (BVLC) upstream Caffe	1.0.0
`caffe-ibm`	IBM Optimized version of BVLC Caffe	1.0.0
`pytorch`	PyTorch	1.0rc1
`tensorflow`	TensorFlow	1.12.0
`tensorboard`	Web Applications for inspecting TensorFlow run and graphs	1.12.0
`tensorflow-probability`	Library for probabilistic reasoning and statistical analysis in TensorFlow	0.5.0

The Python 3 version of each framework appends -py3 to the package name.

Note: PowerAI 1.5.4 introduces caffe-bvlc-py3 and caffe-ibm-py3, providing new Python 3 versions of Caffe in PowerAI.

Table 2. Python 3 versions
Package	Description	Version
`caffe-bvlc-py3`	Berkeley Vision and Learning Center (BVLC) upstream Caffe	1.0.0
`caffe-ibm-py3`	IBM Optimized version of BVLC Caffe	1.0.0
`pytorch-py3`	PyTorch	v1.0rc1
`tensorflow-py3`	TensorFlow	1.12.0
`tensorboard-py3`	Web Applications for inspecting TensorFlow runs and graphs	1.12.0
`tensorflow-probability-py3`	Library for probabilistic reasoning and statistical analysis in TensorFlow	0.5.0

To install each of these frameworks, run the following command:

sudo yum install <framework>

Installing IBM PowerAI distributed deep learning (DDL) packages

To enable distributing model training across a cluster of Power Systems™ servers, install PowerAI distributed deep learning. DDL includes IBM Spectrum™ MPI for communication between systems.

Install the PowerAI distributed deep learning packages by using the following command:

sudo yum install power-ddl

Note: DDL is an optional component. Other PowerAI components can be installed and used without installing DDL.

To use InfiniBand for DDL communications, install the latest Mellanox OFED driver. See the Download tab. If you are running PowerAI from within an Ubuntu container, Mellanox OFED 4.3 must be used

Installing IBM PowerAI Snap ML packages

Install the PowerAI Snap ML packages by using the following command:

sudo yum install power-snapml

Note: Snap ML is an optional component. Other PowerAI components can be installed and used without installing Snap ML.

Accepting the PowerAI license agreement

Read the license agreement and accept the terms and conditions before you use any of the frameworks.

sudo /opt/DL/license/bin/accept-powerai-license.sh

After you finish reading the license agreement, future installs can be automated to silently accept the license agreement.

sudo IBM_POWERAI_LICENSE_ACCEPT=yes /opt/DL/license/bin/accept-powerai-license.sh