Using GPU Devices

imate can run on CUDA-capable GPU devices with the following installed:

1. NVIDIA graphic driver,

2. CUDA libraries.

CUDA Version

The version of CUDA libraries installed on the user’s machine should match the version of the CUDA libraries that imate package was compiled with. This includes matching both major and minor parts of the version numbers. However, the version’s patch numbers do not need to be matched.

Note

The imate package that is installed with either pip or conda already has built-in support for CUDA Toolkit. The latest version of imate is compatible with CUDA 12.2.x, which should match the CUDA version installed on the user’s machine.

Methods of Setting up CUDA and imate

There are three ways to use imate with a compatible version of CUDA Toolkit:

1. Install NVIDIA CUDA Toolkit with a CUDA version compatible with an existing imate installation. In this way, you can keep the imate package that is already installed with pip or conda.

2. Compile imate from the source for a specific version of CUDA to use an existing CUDA library. In this way, you can keep the current CUDA installation.

3. Use docker image with pre-installed imate, CUDA libraries, and NVIDIA graphic driver. This is the most convenient way as no compilation or installation of imate and CUDA Toolkit is required.

The above methods are described in the following sections:

Installing imate on GPU

• 1. Install NVIDIA CUDA Toolkit
  • 1.1. Install CUDA Runtime Libraries
    • 1.1.1. Add CUDA Repository
    • 1.1.2. Install Minimal CUDA Libraries
  • 1.2. Install NVIDIA Graphic Driver
  • 1.3. Install OpenMP
• 2. Compile imate from Source with CUDA
  • 2.1. Install C++ Compiler and OpenMP
  • 2.2. Install CUDA Compiler and Development Libraries
  • 2.3. Load CUDA Compiler on GPU Cluster
  • 2.4. Configure Compile-Time Environment Variables
  • 2.5. Enable Dynamic Loading (optional)
  • 2.6. Compile and Install
• 3. Use imate Docker Container on GPU
  • 3.1. Install Docker
  • 3.2. Install NVIDIA Container Toolkit
  • 3.3. Get imate Docker image
  • 3.4. Use imate Docker Container on GPU

Once you have set up imate on GPU devices, you can refer to the following sections to learn how to execute imate functions on GPUs:

Using imate on GPU

• 1. Inquiry GPU and CUDA with imate
  • 1.1. Locate CUDA Toolkit
  • 1.2. Detect NVIDIA Graphic Driver
  • 1.3. Detect GPU Devices
• 2. Run imate Functions on GPU
  • 2.1. A Simple Example
  • 2.2. Get Process Information
  • 2.3. Verbose Output
  • 2.4. Set Number of GPU Devices
• 3. Deploy imate on GPU Clusters
  • 3.1. Load Modules
  • 3.2. Interactive Session with SLURM
  • 3.3. Submit Jobs to GPU with SLURM