测量超级计算机工作负载的带宽
Measuring Bandwidth for Super Computer Workloads
摘要：
并行计算在从研究到重大问题解决的几乎所有领域都发挥着重要作用。迄今为止，许多研究都集中在并行处理领域。如今，它将其使用扩展到最终用户应用程序，如GPU以及多核处理器开发。带宽测量对于资源管理和研究现有超级计算机系统的各种性能因素是必不可少的，这将有助于更好地利用系统，因为超级计算机非常少，并且它们的资源应该得到适当的利用。本文以一台超级计算机LANL的实际工作量跟踪为例，说明了如何用给定的参数估计带宽。
英文摘要：
Parallel computing plays a major role in almost all the fields from research to major concern problem solving purposes. Many researches are till now focusing towards the area of parallel processing. Nowadays it extends its usage towards the end user application such as GPU as well as multi-core processor development. The bandwidth measurement is essential for resource management and for studying the various performance factors of the existing super computer systems which will be helpful for better system utilization since super computers are very few and their resources should be properly utilized. In this paper the real workload trace of one of the super computers LANL is taken and shown how the bandwidth is estimated with the given parameters.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
