Historically, compilers have been able to translate high level sequential programs into a parallel form which is useful for generating optimal code, but then have subsequently thrown away that parallel information with the result that the final output is sequential target machine code. In our case this would be unfortunate since our target machine HPS I can execute concurrently the operations which comprise the parallel form that the compiler started with. This paper describes our work in compiling C programs for the HPS I. The objective is to enable the HPS I to exploit the irregular parallelism available in the source program.
|Original language||English (US)|
|Number of pages||9|
|Journal||Proceedings of the Hawaii International Conference on System Science|
|State||Published - Dec 1 1986|
ASJC Scopus subject areas
- Computer Science(all)