CentOS 7 Environment Configuration

1. Check GPU device recognition

  $ yum install pciutils
  $ sudo lspci | grep NVIDIA
  3D controller: NVIDIA Corporation GK210GL [Tesla K80] indicates recognition as K80
  3D controller: NVIDIA Corporation Device 1b38 (rev a1) indicates P40

2. Obtain the cuda network source and configure it

The official source address of NVidia is http://developer.download.nvidia.com/compute/cuda/repos/rhel7/x86_64/

  $ wget http://developer.download.nvidia.com/compute/cuda/repos/rhel7/x86_64/cuda-repo-rhel7-8.0.61-1.x86_64.rpm
  $ rpm -Uvh cuda-repo-rhel7-8.0.61-1.x86_64.rpm

Note: Installing the nvidia driver requires the kernel-devel package, which can be installed as follows:

  $ wget http://vault.centos.org/7.0.1406/updates/x86_64/Packages/kernel-devel-3.10.0-123.4.4.el7.x86_64.rpm
  $ wget http://vault.centos.org/7.0.1406/updates/x86_64/Packages/kernel-headers-3.10.0-123.4.4.el7.x86_64.rpm
  $ rpm -Uvh kernel-devel-3.10.0-123.4.4.el7.x86_64.rpm
  $ rpm -Uvh kernel-headers-3.10.0-123.4.4.el7.x86_64.rpm

3. Install cuda 8.0

$ yum install cuda-8-0

3.1 Check driver status

$ sudo nvidia-smi

4. Test the basic GPU functionality (optional)

4.1 Add LD path


$ export LD_LIBRARY_PATH="/usr/local/cuda-7.5/lib64:/usr/lib64/:$LD_LIBRARY_PATH"

4.2 Install cuda examples

  $ cd /usr/local/cuda/bin
  $ sh cuda-install-samples-8.0.sh ~/cuda-test/
  $ cd ~/cuda-test/NVIDIA_CUDA-8.0_Samples
  $ make
  $ ./bin/x86_64/linux/release/deviceQuery To get device status
  $ ./bin/x86_64/linux/release/bandwidthTest To test the device bandwidth

Note: If lnvcuvid error found during the compiling process, run:


$ find . -type f -execdir sed -i 's/UBUNTU_PKG_NAME = "nvidia-367"/UBUNTU_PKG_NAME = "nvidia-375"/g' '{}' \

In which nvidia-375 is the current installed driver version

5. Install cudnn

Optional, note: Different AI frameworks support different versions of cudnn

5.1 Download cudnn software package

, registration of nvidia account is required for downloading.

Note: Download cuDNN v5.1 Library for Linux for CentOS

5.2 Installation

Example using cudnn5.1, since TensorFlow currently only supports 5.1 $ tar -zxf cudnn-8.0-linux-x64-v5.1.tgz

The path for decompression can be freely chosen, generally under /usr/lib, here assume it as <CUDNN_INSTALL_PATH> $ export LD_LIBRARY_PATH=:$LD_LIBRARY_PATH

FAQ

1. Why does nvidia-smi find the GPU usage is 100%?

This problem is due to the system’s inaccurate reading of the gpu status information. Executing the following command can correct it and allow the system to read the command correctly. # nvidia-smi -pm 1

Driver Installation Guide Ubuntu14.04 Environment Configuration