Skip to content

Latest commit

 

History

History
136 lines (118 loc) · 3.72 KB

tensorflow-install-basic.md

File metadata and controls

136 lines (118 loc) · 3.72 KB
Raw

Tensorflow ROCm port: Basic installation

Intro

This instruction provides a starting point for TensorFlow ROCm port (mostly via deb packages). Note: it is recommended to start with a clean Ubuntu 18.04 system

Install ROCm

export ROCM_PATH=/opt/rocm
export DEBIAN_FRONTEND noninteractive
sudo apt update && sudo apt install -y wget software-properties-common

Add the ROCm repository:

wget -qO - http://repo.radeon.com/rocm/apt/debian/rocm.gpg.key | sudo apt-key add -
sudo sh -c 'echo deb [arch=amd64] http://repo.radeon.com/rocm/apt/debian/ xenial main > /etc/apt/sources.list.d/rocm.list'

Tensorflow CSB nigtly build requires ROCm2.8, use the follwoing ROCm repository instead:

wget -qO - http://repo.radeon.com/rocm/apt/debian/rocm.gpg.key | sudo apt-key add -
sudo sh -c 'echo deb [arch=amd64] http://repo.radeon.com/rocm/apt/2.8.0/ xenial main > /etc/apt/sources.list.d/rocm.list'

Install misc pkgs:

sudo apt-get update && sudo apt-get install -y \
  build-essential \
  clang \
  clang-format \
  clang-tidy \
  cmake \
  cmake-qt-gui \
  ssh \
  curl \
  apt-utils \
  pkg-config \
  g++-multilib \
  git \
  libunwind-dev \
  libfftw3-dev \
  libelf-dev \
  libncurses5-dev \
  libpthread-stubs0-dev \
  vim \
  gfortran \
  libboost-program-options-dev \
  libssl-dev \
  libboost-dev \
  libboost-system-dev \
  libboost-filesystem-dev \
  rpm \
  wget && \
  sudo apt-get clean && \
  sudo rm -rf /var/lib/apt/lists/*

Install ROCm pkgs:

sudo apt-get update && \
    sudo apt-get install -y --allow-unauthenticated \
    rocm-dkms rocm-dev rocm-libs rccl \
    rocm-device-libs \
    hsa-ext-rocr-dev hsakmt-roct-dev hsa-rocr-dev \
    rocm-opencl rocm-opencl-dev \
    rocm-utils

Add username to 'video' group and reboot:

sudo adduser $LOGNAME video
sudo reboot

Install required python packages

On Python 2-based systems:

sudo apt-get update && sudo apt-get install -y \
    python-numpy \
    python-dev \
    python-wheel \
    python-mock \
    python-future \
    python-pip \
    python-yaml \
    python-setuptools && \
    sudo apt-get clean && \
    sudo rm -rf /var/lib/apt/lists/*

On Python 3-based systems:

sudo apt-get update && sudo apt-get install -y \
    python3-numpy \
    python3-dev \
    python3-wheel \
    python3-mock \
    python3-future \
    python3-pip \
    python3-yaml \
    python3-setuptools && \
    sudo apt-get clean && \
    sudo rm -rf /var/lib/apt/lists/*

Install Tensorflow Community Supported Builds

Link to the upstream Tensorflow CSB doc: https://github.com/tensorflow/tensorflow#community-supported-builds

We provide nightly tensorflow-rocm whl packages for Python 2.7, 3.5, 3.6 and 3.7 based systems. After downloading the compatible whl package, you can use pip/pip3 to install.

For example, the following commands can be used to download and install the tensorflow-rocm nightly CSB package on an Ubuntu 18.04 system previously configured with ROCm3.7 and Python3.6:

wget http://ml-ci.amd.com:21096/job/tensorflow-rocm-nightly/lastSuccessfulBuild/artifact/pip35_test/whl/tensorflow_rocm-2.0.0-cp35-cp35m-manylinux1_x86_64.whl
pip3 install --user tensorflow_rocm-2.0.0-cp35-cp35m-manylinux1_x86_64.whl

Install TensorFlow ROCm release build

Uninstall any previously-installed tensorflow whl packages:

pip list | grep tensorflow && pip uninstall -y tensorflow

We maintain tensorflow-rocm whl packages on PyPI here. Starting ROCm3.7, ROCm dropped the support of Ubunty16.04 system, hence, Python2.7, Python3.5 based whl packages will not be provided on PyPi.

For Python 3 based systems:

# Pip3 install the whl package from PyPI
pip3 install --user tensorflow-rocm --upgrade