论文部分内容阅读
长期以来，ｓｐｅｅｄｕＰ一直被视为衡量并行处理性能的主要指标之一。不论是并行计算机系统的设计者，还是并行算法的设计者，均非常重视ｓｐｅｅｄｕｐ指标。那么，ｓｐｅｅｄｕｐ能否像人们想象的那样正确地描述并行处理的性能呢？迄今为止，人们对这一问题尚缺乏认识。本文从ｓｐｅｅｄｕｐ的定义出发，结合实例，全面分析了ｓｐｅｅｄｕｐ度量并行处理的性能所存在的问题以及可能导致的错误，还讨论了ｓｐｅｅｄｕｐ的适应条件。
SpeeduP has long been regarded as one of the main metrics for measuring parallel processing performance. Whether it is a designer of parallel computer systems or a designer of parallel algorithms, speedup metrics are highly valued. So, speedup as people imagine the right to describe the performance of parallel processing it? So far, people still lack of understanding of this issue. Starting from the definition of speedup and combining with examples, this paper comprehensively analyzes the problems existing in performance of speedup metric parallel processing as well as possible errors, and also discusses the speedup adaptation conditions.