CUDA is a NVIDIA's universal parallel computing architecture to help you solve complex computing problems with NVIDIA GPUs. You can install the CUDA toolkit by referring to the following instructions. For more information on how to select a CUDA version, see How to Select a Driver, Library, or Software Version.
Before You Begin
The GPU cloud server must be bound to the EIP.
Installing CUDA in a Linux Operating System
1. Get the download link of the CUDA installation package. Visit CUDA download website, select the corresponding CUDA version, select the operating system and installation package, and copy the CUDA toolkit download link. This topic takes the CentOS operating system and CUDA 11.4.0 as an example.
2. Enter the following command to download the CUDA installation package.
wget https://developer.download.nvidia.com/compute/cuda/11.4.0/local_installers/cuda_11.4.0_470.42.01_linux.run
3. Enter the following command to install the CUDA toolkit.
sudo sh cuda_11.4.0_470.42.01_linux.run --silent --toolkit --samples
4. After configuring the environment variables, check the CUDA version. If the following output is displayed, the CUDA is installed successfully.
echo 'export PATH=/usr/local/cuda/bin:$PATH' | sudo tee /etc/profile.d/cuda.sh source /etc/profile nvcc -V
Installing CUDA in a Windows Operating System
1. Download the CUDA installation package. Visit CUDA download website and select your desired CUDA version. This topic take the Windows Server 2016 x86_64 version as an example.
2. Double-click the cuda_11.4.0_471.11_win10 file to start the installation.
3. Select the installation path and click OK.
