Soldato
- Joined
- 9 Jan 2008
- Posts
- 6,513
- Location
- London/Camberley
My MSci project was a C++ program which we wanted to give literally as much CPU time of as possible. We ended up doing 12-day runs on dual Westmere Xeon X5650 boxes (12 cores). Even they were pretty slow compared to GPUs, if only we had time to port it to CUDA.
Same for my project as well. We needed more cores badly to get as many runs as possible and unfortunately it wasn't programmed to run on GPUs.