arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2108.04002cs.DCcs.PF
收藏

准备在Exascale进行性能分析

Preparing for Performance Analysis at Exascale

Jonathon Anderson, Yumeng Liu, John Mellor-Crummey

摘要:

在分析执行度量时,新兴异构艾级平台的性能工具必须解决两个主要挑战。首先,大规模执行的度量可能会记录大量的性能数据。其次,并行程序的性能测量在两个方面是稀疏的:存在于任何上下文的度量集和存在于不同线程中的上下文集。对于GPU加速的应用程序,稀疏性的一个重要来源是,无数的GPU指标都不适用于许多CPU上下文中的任何一个。为了解决这些挑战,我们开发了一种新的流聚合方法来进行事后分析,该方法采用共享和分布式内存并行性来聚合来自应用程序的每个等级,线程和GPU流的稀疏性能测量,并将异构调用路径配置文件和跟踪属性添加到源代码。使用相同数量的资源,我们的方法分析了GPU加速的应用程序的大规模性能测量的速度比HPCToolkit的数量级快,其稀疏的分析结果是多达三个数量级小于HPCToolkit的密集表示的指标。

英文摘要:

Performance tools for emerging heterogeneous exascale platforms must address two principal challenges when analyzing execution measurements. First, measurement of large-scale executions may record mountains of performance data. Second, performance measurements for parallel programs are sparse in two ways: the set of metrics present for any context and the set of contexts present in different threads. For GPU-accelerated applications, an important source of sparsity is that none of the myriad of GPU metrics apply to any of the many CPU contexts. To address these challenges, we developed a novel streaming aggregation approach to postmortem analysis that employs both shared and distributed memory parallelism to aggregate sparse performance measurements from every rank, thread, and GPU stream of an application, and attributes heterogeneous call path profiles and traces to source code. Using the same amount of resources, our approach analyzes large-scale performance measurements of GPU-accelerated applications over an order of magnitude faster than HPCToolkit and its sparse analysis results are as much as three orders of magnitude smaller than HPCToolkit's dense representation of metrics.

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。

↑