When faced with increasing performance needs, typical solutions to go for might be FPGAs (Field-Programmable Gate Arrays), GP-GPUs (General Programmable GPUs) or multicore CPUs. It is a tough challenge to get true performance out of any of these platforms. The achieved performance heavily depends on how you optimize your software for the specific hardware. It is very difficult to assess which platform will perform best. Well, that is exactly what we did!
