TensorFlow ROCm port: Building From Source

Intro

This instruction provides a starting point for build TensorFlow ROCm port from source. Note: it is recommended to start with a clean Ubuntu 18.04 system

Install ROCm

Follow steps at Basic Installation to install ROCm stack. NOTE: ROCm install instructions recommend a purge and reinstall of ROCm rather than upgrading from previous release. For details of the ROCm instructions, please refer to the ROCm QuickStart Installation Guide.

To build with ROCm3.10, set the following environment variables, and add those environment variables at the end of ~/.bashrc

export ROCM_PATH=/opt/rocm/
export HCC_HOME=$ROCM_PATH/hcc
export HIP_PATH=$ROCM_PATH/hip
export PATH=$HCC_HOME/bin:$HIP_PATH/bin:$PATH
export ROCM_TOOLKIT_PATH=$ROCM_PATH

Install required python packages

Install the following python dependencies:

sudo apt-get update && sudo apt-get install -y \
    python-dev \
    python-pip \
    python-wheel \
    python3-numpy \
    python3-dev \
    python3-wheel \
    python3-mock \
    python3-future \
    python3-pip \
    python3-yaml \
    python3-setuptools && \
    sudo apt-get clean

pip3 install keras_preprocessing setuptools keras_applications jupyter --upgrade
pip3 install numpy==1.18.5

Build Tensorflow 1.15 release

Install bazel

echo "deb [arch=amd64] http://storage.googleapis.com/bazel-apt stable jdk1.8" | sudo tee /etc/apt/sources.list.d/bazel.list
curl https://bazel.build/bazel-release.pub.gpg | sudo apt-key add -
sudo apt-get update && sudo apt-get install -y openjdk-8-jdk openjdk-8-jre unzip wget git
cd ~/ && wget https://github.com/bazelbuild/bazel/releases/download/0.24.1/bazel-0.24.1-installer-linux-x86_64.sh
sudo bash ~/bazel*.sh

Build TensorFlow 1.15 ROCm backend

# Clone tensorflow source code 
cd ~ && git clone -b r1.15-rocm-enhanced https://github.com/ROCmSoftwarePlatform/tensorflow-upstream.git tensorflow

# Python 3: Build and install TensorFlow ROCm port pip3 package
cd ~/tensorflow && ./build_rocm_python3

Build Tensorflow 2.3 release

Install bazel

echo "deb [arch=amd64] http://storage.googleapis.com/bazel-apt stable jdk1.8" | sudo tee /etc/apt/sources.list.d/bazel.list
curl https://bazel.build/bazel-release.pub.gpg | sudo apt-key add -
sudo apt-get update && sudo apt-get install -y openjdk-8-jdk openjdk-8-jre unzip wget git
cd ~ && rm -rf bazel*.sh && wget https://github.com/bazelbuild/bazel/releases/download/3.1.0/bazel-3.1.0-installer-linux-x86_64.sh  && bash bazel*.sh && rm -rf ~/*.sh

Build TensorFlow 2.3 ROCm backend

# Clone tensorflow source code 
cd ~ && git clone -b r2.3-rocm-enhanced https://github.com/ROCmSoftwarePlatform/tensorflow-upstream.git tensorflow

# Build and install TensorFlow ROCm port pip3 package
cd ~/tensorflow && ./build_rocm_python3