When dealing with large models, CPU processing is limited by its memory bandwidth. Graphical processing units (GPUs) offer far higher throughputs and include hundreds of parallel cores. This reduces the bottleneck, offering great performance improvements for certain types of problem.
The transient solver, integral equation solver, transmission line matrix (TLM) solver and particle-in-cell (PIC) solver all support GPU processing using the NVIDIA® Tesla series of GPUs. With the transient solver, integral equation solver and TLM solver, up to 8 GPU cards can be used in parallel, allowing even greater speed-up.