An array language for data parallelism: Definition, compilation, and applications

Abstract

The programming of SIMD machines that strongly support data parallelism, such as the Connection Machine, 1 presents new challenges for language, compiler, and algorithm designers. We propose an array language that captures many of the abstractions that are necessary for the effective programming of such machines, thereby liberating the user from having to specify low-level details. Consequently, this new language, ALP, allows for efficient compilation using state-of-the-art techniques, achieving hand-code quality. We demonstrate the effectiveness of our approach by two examples which show that despite being an array language, ALP does not restrict expressiveness to rigidly regular computational structures. © 1991 Kluwer Academic Publishers.