How to Install CUDA on Ubuntu 18.04

CUDA aims at enabling a dramatic increase in computing performance by harnessing the power of the graphics processing unit (GPU) on your system. In this article, I'll show you how to Install CUDA on Ubuntu 18.04. CUDA is a parallel programming model and computing platform developed by NVIDIA.

You can install CUDA on Ubuntu 18.04 using one of the following methods:

  1. From distribution-independent package (runfile packages)
  2. From distribution-specific packages (RPM and Deb packages)

CUDA Pre-installation Checks

Below are a number of checks that you need to perform before installing CUDA Toolkit and Driver on your Ubuntu system.

Verify the system has CUDA-capable GPU

You need to verify that your GPU can work with CUDA, run the following command to check:

$ lspci | grep -i nvidia
01:00.0 3D controller: NVIDIA Corporation GM206M [GeForce GTX 965M] (rev a1)

If you just installed a driver card, you may need to manually update PCI database for above command to return valid output.

# update-pciids

Verify the system has gcc installed

For development using the CUDA, you need to make sure gcc is installed. You can check if it's installed using the command:

$ gcc --version

If not installed, install it with apt-get as below:

$ sudo apt install gcc-6 g++-6

Verify the system has the correct kernel headers and development packages installed.

The CUDA Driver requires that the kernel headers and development packages for the running version of the kernel be installed at the time of the driver installation, as well whenever the driver is rebuilt. You can install kernel headers and development tools using:

$ sudo apt-get install linux-headers-$(uname -r)

Install NVIDIA Driver

CUDA needs Nvidia driver installed on your machine. Install it on Ubuntu 18.04 using the command:

$ sudo apt install nvidia-384

Once this has been installed, you can proceed to install Nvidia CUDA toolkit.

Download the NVIDIA CUDA Toolkit

Depending on your installation method of choice, you need to download equivalent package.  I prefer installing CUDA from a runfile on Ubuntu 18.04 since it is hard to encounter dependency issues.

As of this writing, the latest release of CUDA is v9.2. Since the package size is above 1GB, I'll use wget command to download it so that I can resume easily if the connection gets broken. The CUDA Toolkit contains the CUDA driver and tools needed to create, build and run a CUDA application as well as libraries, header files, CUDA samples source code, and other resources.

$ cd Dowloads
$ wget -c

Once the package has been downloaded locally, make it executable and install it.

# chmod +x
# ./ --verbose --silent --toolkit --override

You should get output similar to below on complete installation.

= Summary =

Toolkit: Installed in /usr/local/cuda-9.2
Samples: Not Selected

Please make sure that
 - PATH includes /usr/local/cuda-9.2/bin
 - LD_LIBRARY_PATH includes /usr/local/cuda-9.2/lib64, or, add /usr/local/cuda-9.2/lib64 to /etc/ and run ldconfig as root

Modify your .bashrc file to include Cuda bin in its path:

export PATH="$PATH:/usr/local/cuda-9.2/bin"

Ensure CUDA library path is present.

# echo "/usr/local/cuda-9.2/lib64" >> /etc/
# ldconfig

Check for available patches and download the .run file then install it.

# wget
# chmod +x
# ./ --silent --accept-eula

Welcome to the CUDA Patcher.
Installation complete!
Installation directory: /usr/local/cuda-9.2

Create symlinks to GCC6 in the CUDA bin folder:

# ln -s /usr/bin/gcc-6 /usr/local/cuda-9.2/bin/gcc
# ln -s /usr/bin/g++-6 /usr/local/cuda-9.2/bin/g++

Read also :

You should now have CUDA toolkit ready. Start creating your projects and enjoy it.

Josphat Mutai 12:05 am

About Josphat Mutai

Red Hat Certified Engineer with more than 2 years experience managing critical Cloud Infrastructures and Services. Passionate about Technology and loves testing and reviewing gadgets and Open-source tools.

Author Archive Page

Have anything to say?

Your email address will not be published. Required fields are marked *

All comments are subject to moderation.