VultureProject / vulture-tensorflow

Instruction for building Tensorflow package for FreeBSD

Guidelines for building tensorflow C++ libraries on FreeBSD 11.X and 12.X

Following guidelines have been tested on FreeBSD 11.1 and FreeBSD 11.2 Do not hesitate to report any problem or improvements

Prerequisites

FreeBSD packages dependencies

In addition to typical packages required to build binaries, you will need clang and bazel :

pkg install clang
pkg install bazel

Google Protobuf

You need a specific version of Protobuf to compile tensorflow. This depends of the tensorflow version.

As for TF 1.10, Protobuf v3.6.1 is required.

Download https://github.com/protocolbuffers/protobuf/releases/download/v3.6.1/protobuf-all-3.6.1.tar.gz
tar xvf protobuf-all-3.6.1.tar.gz
cd protobuf-3.6.1
./configure
make
make install

TensorFlow compilation

Download sources

git clone --recursive https://github.com/tensorflow/tensorflow.git

Switch to the required version

cd tensorflow
git checkout -b r1.10
bazel fetch @grpc//:gpr_base

Configure required features

./configure

Apply patches for FreeBSD

This is not needed as we won't compile Python
echo > tensorflow/requirements.txt

#Adapt paths to your bazel tmp directory
patch ./tensorflow/core/platform/env.cc < ../patches/env.cc.patch
patch ./tensorflow/core/platform/posix/posix_file_system.cc < ../patches/posix_file_system.cc.patch
sed -i '' -E 's/if cpu_value in \("Linux"\)\:/if cpu_value in ("Linux","FreeBSD"):/g' third_party/gpus/rocm_configure.bzl

Compile

You need to adapt --copt (mavx, sse4.2...) to your processor bazel build -c opt --copt=-mavx --copt=-msse4.2 --cxxopt=-std=c++11 --host_linkopt=-lexecinfo --config=noaws --config=nogcp --config=nohdfs --config=noignite --config=nokafka --define=using_rocm=false //tensorflow:libtensorflow_cc.so //tensorflow:libtensorflow_framework.so

Tensorflow libraries are available here:

• ./bazel-bin/tensorflow/libtensorflow_framework.so
• ./bazel-bin/tensorflow/libtensorflow_cc.so