mirror of
https://github.com/storytold/storyteller-ml.git
synced 2026-10-09 00:09:55 +00:00
Justin John
136 lines
4.1 KiB
Bash
136 lines
4.1 KiB
Bash
#!/bin/bash
|
|
|
|
#Script last updated 25-03-2022
|
|
|
|
#Script that seems to be able to build bazel/tf2 on Ubuntu 20.04
|
|
#I really deeply dislike the bazel build system which is bloated and obfuscated for no reason, just Google "NIH syndrome"
|
|
#However Tensorflow2 is a great NN framework
|
|
# See this video "How To Make Package Managers Cry" -> https://www.youtube.com/watch?v=NSemlYagjIU#t=19m0s
|
|
|
|
echo "Check CUDA"
|
|
ls /usr/local/cuda/extras/CUPTI/lib64
|
|
ls /usr/local/cuda
|
|
ls /usr/local/cuda/lib64/ | grep libcudnn.so
|
|
nvcc -V
|
|
|
|
|
|
VERSION="2.8"
|
|
|
|
#Get number of bytes in RAM
|
|
RAM=`free | grep Mem | tr -s ' '| cut -f2 -d ' '`
|
|
|
|
|
|
BazelPleaseSlowDown="--local_resources 2048,.5,1.0"
|
|
|
|
if [ "$RAM" -gt "11286464" ]; then
|
|
echo "More than 12GB";
|
|
BazelPleaseSlowDown=" "
|
|
fi
|
|
if [ "$RAM" -gt "15297500" ]; then
|
|
echo "More than 16GB";
|
|
BazelPleaseSlowDown=" "
|
|
fi
|
|
if [ "$RAM" -gt "31861780" ]; then
|
|
echo "More than 32GB";
|
|
BazelPleaseSlowDown=" "
|
|
fi
|
|
|
|
|
|
#Tensorflow is a great Neural network library that unfortunately is coupled to the terrible Bazel build system
|
|
#This is a download and build script for Ubuntu 20.04, that should work building release 2.8
|
|
|
|
sudo apt-get install python3-dev python3-pip python3-venv python3-tk
|
|
|
|
pip3 install -U --user pip numpy wheel packaging
|
|
pip3 install -U --user keras_preprocessing --no-deps
|
|
|
|
cd ~/Documents
|
|
mkdir 3dParty
|
|
cd 3dParty
|
|
|
|
#wget https://github.com/bazelbuild/bazel/releases/download/4.2.1/bazel-4.2.1-installer-linux-x86_64.sh
|
|
#chmod +x bazel-4.2.1-installer-linux-x86_64.sh
|
|
#./bazel-4.2.1-installer-linux-x86_64.sh --user
|
|
|
|
#r2.8
|
|
mkdir -p "$HOME/.bazel/bin" && cd "$HOME/.bazel/bin" && curl -fLO https://releases.bazel.build/4.2.1/release/bazel-4.2.1-linux-x86_64 && chmod +x bazel-4.2.1-linux-x86_64
|
|
|
|
#Create shared directory
|
|
if [ -f ~/.bashrc ]
|
|
then
|
|
if cat ~/.bashrc | grep -q "BAZEL_CANCER"
|
|
then
|
|
echo "Bazel includes seem to be set-up.."
|
|
else
|
|
USER=`whoami`
|
|
echo "#BAZEL_CANCER" >> ~/.bashrc
|
|
echo "source $HOME/.bazel/bin/bazel-complete.bash" >> ~/.bashrc
|
|
echo "export PATH=\"\$PATH:\$HOME/bin\"" >> ~/.bashrc
|
|
source ~/.bashrc
|
|
fi
|
|
fi
|
|
|
|
if [ ! -d tensorflow ]
|
|
then
|
|
git clone https://github.com/tensorflow/tensorflow.git
|
|
fi
|
|
|
|
cd tensorflow
|
|
git pull
|
|
git checkout r$VERSION
|
|
|
|
|
|
#Make sure to check your target CPU and when asked used the correct -march= / -mtune=
|
|
# for example for an old intel i7 -march=nehalem is used..
|
|
#https://gcc.gnu.org/onlinedocs/gcc/x86-Options.html
|
|
|
|
|
|
echo "Answers to configure questions :"
|
|
echo "/usr/bin/python3"
|
|
echo "/usr/lib/python3/dist-packages"
|
|
echo "cuda Y"
|
|
echo "tensorrt Y"
|
|
echo "CUDA 11.2"
|
|
echo "CuDNN 8"
|
|
echo "TensorRT 8"
|
|
echo "/usr/local/cuda/,/usr/local/cuda/include/,/usr/local/cuda/bin/,/usr/local/cuda/lib64/,/usr/local/cuda/lib64/,/usr/local/tensorrt-8.2.3/,/usr/local/tensorrt-8.2.3/include/,/usr/local/tensorrt-8.2.3/lib/,"
|
|
echo "Compute capability 6.1 ( for GTX 1050 + cards )"
|
|
|
|
#Attempt to inform the configure script on how to find the CUDA stuff..
|
|
export CUDNN_INSTALL_PATH=/usr/local/cuda/,/usr/local/cuda/include/,/usr/local/cuda/bin/,/usr/local/cuda/lib64/,/usr/local/cuda/lib64/,/usr/local/tensorrt-8.2.3/,/usr/local/tensorrt-8.2.3/include/,/usr/local/tensorrt-8.2.3/lib/
|
|
|
|
|
|
./configure
|
|
|
|
bazel clean --expunge
|
|
|
|
|
|
|
|
#You should use CUDA 11.2 and cudnn-11.2-linux-x64-v8.1.1.33 and TensorRT 8.2.3
|
|
|
|
bazel build --config=opt --config=cuda --config=mkl --config=monolithic $BazelPleaseSlowDown //tensorflow/tools/pip_package:build_pip_package
|
|
./bazel-bin/tensorflow/tools/pip_package/build_pip_package ~/Documents/3dParty/
|
|
#To install
|
|
#pip3 --user install ~/Documents/3dParty/tensorflow-2.4.0-cp36-cp36m-linux_x86_64.whl
|
|
|
|
|
|
|
|
bazel build --config opt --config=cuda --config=monolithic //tensorflow/tools/lib_package:libtensorflow
|
|
cp bazel-bin/tensorflow/tools/lib_package/libtensorflow.tar.gz ~/Documents/3dParty/libtensorflow-r$VERSION.tar.gz
|
|
|
|
|
|
#Build tensorflow lite
|
|
#https://www.tensorflow.org/lite/guide/build_cmake
|
|
mkdir tflite_buld
|
|
cd tflite_buld/
|
|
cmake ../tensorflow/lite
|
|
cmake --build . -j
|
|
|
|
|
|
echo "Please visit ~/Documents/3dParty/ to collect your tensorflow python3 wheel, and C Library.."
|
|
echo "Will now use : python -c 'import tensorflow as tf;' to test your tensorflow"
|
|
|
|
python -c 'import tensorflow as tf;'
|
|
|
|
exit 0
|