Your system must meet the following requirements to use the GPU accelerator capability in Mechanical APDL. For information on the most recently tested GPU cards, see the GPU Accelerator Capabilities PDF on the Platform Support section of the Ansys Website.
The machine(s) being used for the simulation must contain at least one GPU card.
A minimum 16GB of on-card memory is recommended in order to achieve meaningful acceleration in simulations that can use the GPU card.
To achieve optimal performance, only GPU cards with significant double precision performance (FP64) are recommended for use with the sparse direct solver and eigensolvers based on the sparse solver (for example, Block Lanczos or subspace). The following cards are recommended:
Recommended Solvers Card Release Year Sparse (Direct) Iterative (PCG, etc.) Mixed AMD Instinct MI325X 2024 Y Y Y AMD Instinct MI300X 2023 Y Y Y AMD Instinct MI210 2022 Y Y Y AMD Instinct MI250, MI250X 2021 Y Y Y NVIDIA B100 2024 Y Y Y NVIDIA A800 2024 Y Y Y NVIDIA H100 2022 Y Y Y NVIDIA A30 2021 Y Y Y NVIDIA A100 2020 Y Y Y AMD Radeon Pro W7900, W7800 2023 N Y Y NVIDIA RTX 5880 Ada 2024 N Y Y NVIDIA RTX 5000 Ada, A4500 Ada 2023 N Y Y NVIDIA L40 2022 N Y Y NVIDIA RTX 6000 Ada 2022 N Y Y NVIDIA RTX A5500 2022 N Y Y NVIDIA A16, A10 2021 N Y Y NVIDIA RTX A5000, A4500, A4000 2021 N Y Y NVIDIA RTX A6000 2020 N Y Y NVIDIA A40 2020 N Y Y For NVIDIA GPU cards, note the following:
The Mechanical APDL installation provides the necessary CUDA 12.6 libraries to accelerate the FEA solution. No separate installation of CUDA is required.
The GPU driver is not installed by Mechanical APDL. The driver version must be 525.60.13 or newer.
For AMD GPU cards, note the following:
The Mechanical APDL installation provides the necessary HIP/ROCm 6.2.0 libraries to accelerate the FEA solution. No separate installation of HIP/ROCm is required. Note that your machine must have at least SUSE Linux Enterprise Server 15 SP2 to use the ROCm libraries.
The GPU driver is not installed by Mechanical APDL. The driver version must be 6.2.0 or newer. Some instructions are given below.
the following AMD link lists their supported operating systems (OS).
To install the AMD driver, you must perform the three-step procedure summarized below and described in detail with specific Linux commands for different operating systems on the AMD page: How to Install ROCm (amd.com).
Download the "amdgpu-install" script to your system using the appropriate commands for your operating system. Note that the version of amdgpu-install must come from the ROCm 6.2.0 page, and you only need to download the installer script (the full ROCm package is not needed).
Install the "kernel-mode driver" with the following command:
sudo amdgpu-install --usecase=dkms
Install the “libdrm-amdgpu” library to your operating system using the appropriate commands.
Add your user name (
LOGNAME) to the list of users in the video group through the following command.sudo usermod -a -G video $LOGNAME
To utilize a GPU device that is not on the recommended list of cards, set the following environment variable:
ANSGPU_OVERRIDE=1
This is most beneficial when you wish to run on newer GPUs that were not available at the time of release of this version of the Ansys program. If you choose to use this environment variable, you should ensure that the GPU device that you wish to use is sufficiently powerful, in terms of both double-precision compute power and on-card memory, to achieve meaningful acceleration for your simulation. Using this environment variable with an underpowered CPU may actually decelerate your simulation.
If you have issues setting up the GPU drivers, see Troubleshooting in the Parallel Processing Guide.