Note: This section is only required for installing
the Conda package. If you are using the Docker image, you can skip this section.
- GPU-related Software
- For Ubuntu 18.04, install the following software:
- CUDA 11.0
- cuDNN 8.0.4
- CUDA driver >=450.80.02
- NVIDIA GPU Drivers
- Install the GPU driver using the
apt-getcommand or directly install the CUDA package with the driver. For example:sudo apt-get install nvidia-driver-450 - CUDA Toolkit
- Get the CUDA package associated with the Ubuntu version from https://developer.nvidia.com/cuda-toolkit-archive and directly install the NVIDIA CUDA runfile package.
- cuDNN SDK
- Get the cuDNN from https://developer.nvidia.com/cudnn
and append its installation directory to the $LD_LIBRARY_PATH environmental
variable.
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cudnn-8.0.4/lib64 - CUPTI
- CUPTI is required by vai_p_tensorflow and is installed with CUDA. Add the
CUPTI directory to the $LD_LIBRARY_PATH
environment variable. For
example:
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda/extras/CUPTI/lib64 - NCCL
- NCCL is required by vai_p_caffe. Download NCCL from here and install
it.
sudo dpkg -i nccl-repo-ubuntu1804-2.6.4-ga-cuda10.0_1-1_amd64.deb sudo apt update sudo apt install libnccl2=2.6.4-1+cuda10.0 libnccl-dev=2.6.4-1+cuda10.0