A shared memory model on a cluster PCs

摘要

The GFLOPS project's aim is to develop a parallel architecture as well its software environment to implement scientific applications efficiently. This goal can be achieved only with a real collaboration among the architecture, the compiler and the programming language. This paper investigates the C//Parallel language and the underlying programming model on a global address space architecture. The main advantage of our paradigm is that it allows a unique framework to express and optimize both data and control parallelism at user-level. The evaluation of C//has been conducted on a cluster of PCs.