Computer Simulation Technology
 

GPU Computing - CUDA

Modern Graphics Processing Units feature a large number of processing cores, in the case of the NVIDIA® Tesla™ 10 series there are 240. CST uses the high degree of parallelization provided with NVIDIA® CUDA™ (Compute Unified Device Architecture) to speed up the simulation dramatically.

GPU computing is available for CST MWS's transient solver. You will be able CST MWS simulations even faster while maintaining access to the performance determining features such as Perfect Boundary Approximation (PBA)® and Thin Sheet Technology (TST)™. Many other important features of CST MWS are implemented, including to open boundaries, lossy metals, dispersive materials, and far field monitors. Currently single card, dual card, and quad card configurations are supported.

Speedup of solver loop for GPU computing compared to 2x Intel Xeon X5550.
Speedup of solver loop for GPU computing compared to a work station with 2x Intel Xeon X5550 (Nehalem) processors. The drop in speed up at high meshpoint numbers indicates the soft memory limit for the respective card configurations.