Moded Flat Ghc for Data-parallel Programming (extended Abstract) 1 Regular Computation in Concurrent Logic Programming

Concurrent logic languages have been used mainly for the (parallel) processing of rather irregular symbolic applications. However, since concurrent logic languages are essentially general-purpose, they should be applicable to problems with regular structures and their data-parallel processing as well. This paper studies the possibility of massively parallel processing in concurrent logic programming, focusing on arrays and its data-parallel processing. \We hope the simplicity of GHC will make it suitable for a parallel computation model as well as a programming language. The exibility of GHC makes its eecient implementation diicult compared with CSP-like languages. However, a exible language could be appropriately restricted in order to make simple programs run eeciently. On the other hand, it would be very diicult to extend a fast but innexible language naturally." | 5] (1985) Concurrent logic languages have focused mainly on the parallel processing of symbolic applications with rather irregular structures 8, 4]. However, real-life parallel symbolic applications (such as machine learning) may involve a lot of numerical computation as well. We anticipate that future symbolic languages should provide certain support of high-performance computing. Whether concurrent logic languages can evolve in this direction deserves studying in depth. Previous approaches to irregular parallel symbolic processing involving numerical computation were mostly multilingual. The whole computation was coordinated by symbolic languages, while numerical computation was programmed in conventional languages and called via foreign-language interface. This may be a promising approach in the short run, with the obvious advantage of the reuse of existing numerical software,