GPU computing approach

The GPU computing approach uses the computers graphics card to perform of the computational intensive calculations. This approach is based on CUDA^TM by NVIDIA and can be executed on NVIDIA graphic cards with Com­pute Capability 3.0 or higher.

Some GPUs have a significantly higher processing power when performing single-precision floating-point calculations rather than double-precision float­ing-point calculations. The default setting for the flexible mesh based engines is to use double-precision floating-point calculations, because these are more accurate than single-precision floating-point calculations. However, it is possi­ble to force the flexible mesh based engines to use single precision floating-point calculations. This is done by ticking off the “Use single precision GPU calculations” checkbox in the Simulation Settings tab found in File -> Options -> User Settings.