MPI-2 one-sided usage and implementation for read modify write operations: A case study with HPCC

Abstract

MPI-2's One-sided communication interface is being explored in scientific applications. One of the important operations in a one sided model is readmodify-write. MPI-2 semantics provide MPLPut, MPLGet and MPI_Accumulate operations which can be used to implement read-modify-write functionality. The different strategies yield varying performance benefits depending on the underlying one-sided implementation. We use HPCC Random Access benchmark which primarily uses read-modijy-write operations as a case study for evaluating the different implementation strategies in this paper. Currently this benchmark is implemented based on MPI two-sided semantics. In this work we design and evaluate MPI-2 versions of the HPCC Random Access benchmark using one-sided operations. To improve the performance, we explore two different optimizations: (i) software based aggregation and (ii) hardware-based atomic operations. We evaluate our different approaches on an InfiniBand cluster. The software based aggregation outperforms the basic one sided scheme without aggregation by a factor of 4.38. The hardware based scheme shows an improvement by a factor of 2.62 as compared to the basic one sided scheme. © Springer-Verlag Berlin Heidelberg 2007.