摘要
In this paper, a new loop transformation is proposed that can exploit parallelism in loops which cannot be found by traditional methods. Then the method is extended to show how to achieve maximum speedup of loops if there are infinite processors and how to balance the workload of parallel sections in loops i f there is fixed number of processors.
In this paper, a new loop transformation is proposed that can exploit parallelism in loops which cannot be found by traditional methods. Then the method is extended to show how to achieve maximum speedup of loops if there are infinite processors and how to balance the workload of parallel sections in loops i f there is fixed number of processors.
