摘要
数据流计算机开销过大的基本原因是并行的粒度太细。本文研完在数据流计算模型中引入复合函数级并行的开发,讨论它在SDS模型的抽象机上的实现方法,并介绍以此为基础的模型机SDS-1系统。
The basic reason of large overhead in a dataflow computer is little parallel granularity. In this paper the parallelism in compound function level is introduced into dataflow computing model. Its implementation in a abstract machine of SDS model is also discussed by describing the SDS-1 system.
出处
《计算机学报》
EI
CSCD
北大核心
1989年第5期321-327,共7页
Chinese Journal of Computers
