Generic Parallel Array Programming in SAC

SaC is a purely functional array processing language for computationally inten-sive numerical applications. Its design aims at combining efficiency in program construction with efficiency in parallel program execution. We demonstrate the declarative, generic pro-gramming style of SaC by means of a small case study: 3-dimensional complex fast-Fourier transforms. The impact of abstraction on expressiveness, readability, and maintainability of code as well as on clarity of underlying mathematical concepts is discussed and compared with other approaches. We quantify the associated impact on runtime performance both in uniprocessor and in multiprocessor environments.