Exploiting Maximum Loop Parallelism by Using Heterogeneous Computing
Zeng Guo-su · Journal of Jiangxi Normal University · 2000
In this paper, the definition of maximum loop speedup has been presented, which is the metric of parallelism hidden in loop body. the the classes of Do-loop and their dependence as well as the parallelism they contained have been studied. How to exploit such parallelism under heterogeneous computing environment? The several approaches have been proposed in this paper, which are eliminating serial bottleneck by means of heterogeneous computing, heterogeneous Do-all-loop scheduling, heterogeneous Do- across scheduling. We find that, not only on theoretical analysis but also on experimental results, these schemes acquire better performance than in homogeneous computing.