THIS PAGE ISN'T FINALIZED !

CUDA (Compute Unified Device Architecture) is a parallel computing architecture developed by Nvidia for graphics processing.

This document provides instructions to install/remove Cuda 4.2 on Ubuntu 12.04.

Installation

Prerequisites

It is necessary to have a Nvidia proprietary driver up to 295.xx (which can be installed automatically or manually). For Ubuntu versions under 12.04, you will have to install Nvidia driver manually.

Be sure your GPU is compatible with Cuda. Refer to this page. Then, install required packages:

sudo apt-get install libxi-dev libxmu-dev freeglut3-dev build-essential binutils-gold

Downloads

You will have to download two programs: "CUDA Toolkit" and "GPU Computing SDK".

Let's download and save them on the Desktop.

32bits:

cd ~/Desktop
wget http://developer.download.nvidia.com/compute/cuda/4_2/rel/toolkit/cudatoolkit_4.2.9_linux_32_ubuntu11.04.run
wget http://developer.download.nvidia.com/compute/cuda/4_2/rel/sdk/gpucomputingsdk_4.2.9_linux.run

64bits:

cd ~/Desktop
wget http://developer.download.nvidia.com/compute/cuda/4_2/rel/toolkit/cudatoolkit_4.2.9_linux_64_ubuntu11.04.run
wget http://developer.download.nvidia.com/compute/cuda/4_2/rel/sdk/gpucomputingsdk_4.2.9_linux.run

Installation of CUDA Toolkit

Go to "Desktop", add execution permissions of the cudatoolkit downloaded file, and execute it with root permissions:

cd ~/Desktop
chmod +x cudatoolkit_4.2.9_linux_*
sudo ./cudatoolkit_4.2.9_linux_*

When it will ask you:

Enter install path (default /usr/local/cuda, '/cuda' will be appended):

type /opt.

Installation of GPU Computing SDK

As before, go to "Desktop", add execution permissions of the gpucomputingsdk downloaded file, and execute it without root permissions:

cd ~/Desktop
chmod +x gpucomputingsdk_4.2.9_linux.run
./gpucomputingsdk_4.2.9_linux.run

When it will ask you:

Enter install path (default ~/NVIDIA_GPU_Computing_SDK):

press [enter] (to use default path), and type /opt when it will ask you:

Enter CUDA install path (default /usr/local/cuda):

Preparation

Update variables, and this on every boot:

export PATH=$PATH:/opt/cuda/bin
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/cuda/lib:/opt/cuda/lib64
echo 'export PATH=$PATH:/opt/cuda/bin' >> ~/.bash_profile
echo 'export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/cuda/lib:/opt/cuda/lib64' >> ~/.bash_profile

Compilation of SDK

Need to be tested !

We finally compile:

cd ~/NVIDIA_GPU_Computing_SDK/C
LINKFLAGS=-L/usr/lib/nvidia-current/ make cuda-install=/opt/cuda

NOTE: In case you are using the latest (post-release updates; version current-updates) instead of the recommended (version current) Nvidia drivers, then change the LINKFLAGS to read:

cd ~/NVIDIA_GPU_Computing_SDK/C
LINKFLAGS=-L/usr/lib/nvidia-current-updates/ make cuda-install=/opt/cuda

You can check for which version you are using by running the "Additional Drivers" program (System Settings->Hardware->Additional Drivers) and check which version is toggled ON.

Test

Now, let's test if Cuda is working great. Type:

~/NVIDIA_GPU_Computing_SDK/C/bin/linux/release/./fluidsGL

As you can see on the following picture, it works !

If you are using optimus with bumblebee, run the above with optirun as below:

optirun ~/NVIDIA_GPU_Computing_SDK/C/bin/linux/release/./fluidsGL

Will come soon...

Cleanup

Now Cuda is installed, the installers files are useless, you can remove them:

cd ~/Desktop
rm cudatoolkit_4.2.9_linux_*
rm gpucomputingsdk_4.2.9_linux.run

Uninstallation

If you want to totally remove Cuda, juste delete the /opt/cuda and ~/NVIDIA_GPU_Computing_SDK folders:

rm -r ~/NVIDIA_GPU_Computing_SDK
sudo rm -r /opt/cuda

and remove the export PATH=$PATH:/opt/cuda/bin and export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/cuda/lib:/opt/cuda/lib64 lines of the ~/.bash_profile file.