I'm working on software/hardware for business intelligence. I'm doing heterogenous parallel programming Intel/Nvidia. I use Linux with Intel & Nvidia compilers, libraries and tools. I have several machines but with 3.2 GHz quad-core and a low to mid range Nvidia video card as a compute processor, I can sustain 450 gflops. I use Intel SSDs as part of a hierarchical storage system to keep the processors supplied with data. I'm also testing using FPGAs for some processes.