An algorithm for the minimum value movement cost on a data flow multiprocessor system
暂无分享,去创建一个
In a data flow model, data movement cost within a processor is negligible in comparison with data movement cost among processors. We exploit the algorithm which partitions nodes of data flow graph into clusters considering the data dependency and edge cost in order to maintain the inherent parallelism of data flox graph and minimize total mount of data movement cost among processors.
[1] 유원희,et al. 데이타 플로우 언어의 설계에 관한 연구 , 1986 .