Using GPU Acceleration for Maxwell 3D Solvers

Maxwell 3D solvers can be accelerated by using a GPU accelerator. Numerical factorization of symmetric matrices is supported for GPU acceleration.

You can also enable distributed GPUs for Eddy Current Frequency Sweeps.

Hardware Requirements

For information on supported graphics cards, see the Ansys Platform Support webpage.

Enabling GPUs for Maxwell 3D solvers

  1. Go to Tools > Options > HPC and Analysis Options.

  2. In the Options tab, set the Enable GPU option to True.

    HPC and Analysis Options window with Enable GPU setting selected

Setting Up Nondistributed GPU for Maxwell 3D Solvers

From the Analysis Configuration window's Machines tab, in the Machines for Distributed Analysis section, set GPUs to 1.

Analysis Configuration window with Machines tab selected and number of GPUs set to 1

Enabling Distributed GPUs for Eddy Current Frequency Sweep

To set up distributed GPUs for eddy current frequency sweeps:

  1. Go to Tools > Options > HPC and Analysis Options and click Edit, or select the Simulation tab of the ribbon, and click Analysis Config to edit Analysis Configurations.
  2. Uncheck Use Automatic Settings, select the Job Distribution tab, then select Frequencies.

    Analysis Configuration window with Job Distribution tab selected and Frequencies enabled

  3. Go to Tools > Options > HPC and Analysis Options, select the Options tab, and set Enable GPU to True.

    HPC and Analysis Options window with Options tab selected and Enable GPU set to true

  4. Set up the machines for distributed GPU usage.

    For example, the following figure shows two simulation jobs set to run on the local host; and up to two GPUs can be used for acceleration depending on the availability of licenses and GPU cards.

    Analysis Configuration window with Machines tab selected and two cores enabled

Related Topics 

Setting HPC and Analysis Options for Maxwell and RMxprt Designs