# installs essential packages through apt-get and pip
bash aws_setup.sh
bash install_cuda_8.sh
- Make an account on NVIDIA and download cuDNN.
- After the survey, select "cuDNN v6.0, for CUDA 8.0".
- Download "cuDNN v6.0 Library for Linux" (should be a tar file).
- Transfer (you can use
scp
) the.tgz
file from your local computer to your EC2 instance. - Install cuDNN from your ec2 instance with
bash install_cuDNN.sh
.
pip install tensorflow-gpu
bash install_monitoring_tools.sh
git clone https://github.com/BVLC/caffe.git
cd caffe
cp Makefile.config.example Makefile.config
Manual Step: edit Makefile.config
to set USE_CUDNN := 1
.
bash caffe_dependencies.sh
make all
make pycaffe
make test
make runtest
Change the path name and add the following line to your .bashrc
file:
export PYTHONPATH="${PYTHONPATH}:/path/to/caffe/python"
Source your .bashrc
file.
source ~/.bashrc