Sparbit：一种新的具有计算代价和数据位置感知的MPI Allgather算法
Sparbit: a new logarithmic-cost and data locality-aware MPI Allgather algorithm
摘要：
集体操作被认为是提高exascale就绪和高性能计算应用程序的性能的关键。在本文中，我们专注于消息传递接口（MPI）Allgather多对多的集体，这是最多的调用和耗时的操作。用于此调用的每个MPI算法都受到不同的操作和性能限制，这可能包括仅适用于受限情况，随着进程数量的增长需要线性数量的通信步骤，内存副本和移位以确保正确的数据组织，以及非本地数据交换模式，其中大部分对总操作时间有负面影响。所有这些特征创造了一种环境，在这种环境中，没有算法对所有情况都是最好的，因此这意味着必须仔细选择替代方案来执行调用。考虑到这些方面，我们提出了条纹并行二叉树（Sparbit）算法，它具有最佳的延迟和带宽的时间成本，没有使用限制。它还保持了一个更本地的通信模式，最大限度地减少了由于长距离交流的延迟，允许提取更多的性能从当前系统相比，渐近等效的替代品。在最佳情况下，Sparbit在46.43%的测试用例上超过了传统MPI算法，平均（中位数）改进了34.7%（26.16%），最高达到84.16%。
英文摘要：
The collective operations are considered critical for improving the performance of exascale-ready and high-performance computing applications. On this paper we focus on the Message-Passing Interface (MPI) Allgather many to many collective, which is amongst the most called and time-consuming operations. Each MPI algorithm for this call suffers from different operational and performance limitations, that might include only working for restricted cases, requiring linear amounts of communication steps with the growth in number of processes, memory copies and shifts to assure correct data organization, and non-local data exchange patterns, most of which negatively contribute to the total operation time. All these characteristics create an environment where there is no algorithm which is the best for all cases and this consequently implies that careful choices of alternatives must be made to execute the call. Considering such aspects, we propose the Stripe Parallel Binomial Trees (Sparbit) algorithm, which has optimal latency and bandwidth time costs with no usage restrictions. It also maintains a much more local communication pattern that minimizes the delays due to long range exchanges, allowing the extraction of more performance from current systems when compared with asymptotically equivalent alternatives. On its best scenario, Sparbit surpassed the traditional MPI algorithms on 46.43% of test cases with mean (median) improvements of 34.7% (26.16%) and highest reaching 84.16%.
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。
