Skip to content

GPU/OpenCL Driver Setup

CompuCell3D GPU solvers use OpenCL. The solvers are included with the CompuCell3D packages, but the operating system must expose a working OpenCL GPU device through the vendor driver.

After installing or updating a GPU driver, reboot the computer before testing CompuCell3D GPU solvers.

Driver packages and installation steps change over time. The notes below are intended as a practical starting point, but the best source of current installation instructions is always the GPU vendor documentation for your exact hardware, operating system, and driver version.

Windows

NVIDIA GPUs

For most Windows users with NVIDIA GPUs, install the current NVIDIA driver or CUDA Toolkit from NVIDIA. The CUDA Toolkit installer includes the NVIDIA driver and an OpenCL library component.

  1. Open Device Manager and confirm that the GPU is listed under Display adapters.
  2. Install the driver using one of these NVIDIA options:
  3. Recommended for most users: NVIDIA driver downloads
  4. For developer or build systems: NVIDIA CUDA Toolkit downloads
  5. Use the default installer options unless you have a site-specific IT policy.
  6. Reboot.
  7. Open Command Prompt or PowerShell and run:
nvidia-smi

If nvidia-smi prints the GPU model and driver version, the NVIDIA driver is installed correctly. If CompuCell3D still does not find an OpenCL GPU device, reinstall the NVIDIA driver with the clean-install option enabled.

AMD GPUs

AMD cards need AMD Software: Adrenalin Edition or Radeon PRO Software, depending on the GPU family. Do not install NVIDIA CUDA for AMD cards.

  1. Open Device Manager and confirm the AMD GPU model under Display adapters.
  2. Download the appropriate AMD driver:
  3. For Radeon consumer GPUs, use AMD Windows drivers.
  4. For Radeon PRO/workstation GPUs, use the Radeon PRO driver listed for your exact card on the same AMD support site.
  5. Run the AMD installer. A Full Install or Minimal Install is fine for most users; Driver Only can also work if you do not need AMD's control panel.
  6. Reboot.
  7. If CompuCell3D does not find the AMD OpenCL device, use AMD's cleanup utility, reboot, and reinstall the current driver for the exact GPU model.

AMD also provides the HIP SDK for developer workflows on supported AMD GPUs, but ordinary CompuCell3D users should start with the standard AMD graphics driver first.

Ubuntu Linux

These instructions focus on Ubuntu. Commands should be run in a terminal.

Install clinfo first; it is the simplest way to verify that OpenCL is visible to user applications:

sudo apt update
sudo apt install clinfo

After the vendor driver is installed and the system has been rebooted, run:

clinfo | less

Look for at least one GPU platform and device. If clinfo only lists a CPU device, or reports no platforms, the GPU OpenCL runtime is not installed correctly.

NVIDIA GPUs on Ubuntu

Ubuntu's ubuntu-drivers tool is the most reliable default choice because it selects distribution-packaged drivers that match the kernel and Secure Boot configuration.

sudo apt update
sudo apt upgrade
sudo reboot

After reboot:

sudo ubuntu-drivers list
sudo ubuntu-drivers install
sudo reboot

For workstation or compute-focused systems, Ubuntu also supports the GPGPU/server driver path:

sudo ubuntu-drivers list --gpgpu
sudo ubuntu-drivers install --gpgpu
sudo reboot

Verify the driver:

nvidia-smi
clinfo | less

If nvidia-smi works but clinfo does not show an NVIDIA OpenCL platform, install the matching NVIDIA OpenCL ICD package for the driver branch shown by nvidia-smi. For example, if Ubuntu installed the 535 driver branch:

sudo apt install nvidia-opencl-icd-535
sudo reboot

Use the branch number that matches your installed driver.

AMD GPUs on Ubuntu

For AMD GPUs, the most reliable OpenCL route is AMD's packaged driver stack from the AMD support site or ROCm documentation. AMD GPU support depends strongly on the exact card model and Ubuntu version, so verify that your GPU is supported before changing drivers.

For Ubuntu 24.04, AMD's ROCm quick-start flow currently starts with:

wget https://repo.radeon.com/amdgpu-install/7.2.4/ubuntu/noble/amdgpu-install_7.2.4.70204-1_all.deb
sudo apt install ./amdgpu-install_7.2.4.70204-1_all.deb
sudo apt update

For Ubuntu 22.04:

wget https://repo.radeon.com/amdgpu-install/7.2.4/ubuntu/jammy/amdgpu-install_7.2.4.70204-1_all.deb
sudo apt install ./amdgpu-install_7.2.4.70204-1_all.deb
sudo apt update

Then install the AMD GPU kernel driver and ROCm user-space packages:

sudo apt install "linux-headers-$(uname -r)" "linux-modules-extra-$(uname -r)"
sudo apt install amdgpu-dkms
sudo apt install python3-setuptools python3-wheel
sudo usermod -a -G render,video $LOGNAME
sudo apt install rocm
sudo reboot

After reboot, verify OpenCL visibility:

groups
clinfo | less

The groups output should include render and video. If it does not, log out and back in, or reboot again.

Notes for Linux Driver Troubleshooting

  • Avoid mixing vendor .run installers with Ubuntu-packaged GPU drivers unless you know exactly why you need that setup.
  • Secure Boot can prevent unsigned kernel modules from loading. If the GPU disappears after installing drivers, check Secure Boot and kernel module signing.
  • If multiple OpenCL implementations are installed, clinfo is the quickest way to see which platforms are visible.
  • On laptops with hybrid graphics, make sure the discrete GPU is enabled in BIOS/firmware or the vendor control panel.

Official References