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 read-modify-write. MPI-2 semantics provide MPI_Put, MPI_Get 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-modify-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.
This research is supported in part by Department of Energy’s grant #DE-FC02-06ER25749 and #DE-FC02-06ER25755; National Science Foundation’s grants #CNS-0403342 and #CCF-0702675; grants from Intel, Sun Microsystems, Cisco Systems, and Linux Networx; Equipment donations from Intel, Mellanox, AMD, Apple, IBM, Microway, PathScale, Silverstorm and Sun Microsystems.
This is a preview of subscription content, log in via an institution to check access.
Preview
Unable to display preview. Download preview PDF.
Similar content being viewed by others
References
Network Based Computing Laboratory, MVAPICH2, http://mvapich.cse.ohio-state.edu/
Argonne National Laboratory: MPICH2, http://www-unix.mcs.anl.gov/mpi/mpich2/
Booth, S., Mourao, F.E.: Single Sided MPI Implementations for SUN MPI. In: Supercomputing (2000)
Buntinas, D., Panda, D.K., Gropp, W.: NIC-Based Atomic Remote Memory Operations in Myrinet/GM. In: Workshop on Novel Uses of System Area Networks (SAN-1) (February 2002)
Dongarra, J., Luszczek, P.: overview of the hpc challenge benchmark suite. In: SPEC Benchmark Workshop (2006)
Garg, R., Sabharwal, Y.: Optimizing the HPCC randomaccess benchmark on blue Gene/L Supercomputer. ACM SIGMETRICS Performance Evaluation Review (2006)
InfiniBand Trade Association: InfiniBand Architecture Specification, Release 1.0 (October 24, 2000)
Jiang, W., J.Liu, Jin, H.W., Panda, D.K., Buntinas, D., R.Thakur, W.Gropp: Efficient Implementation of MPI-2 Passive One-Sided Communication on InfiniBand Clusters. EuroPVM/MPI (September 2004)
Nieplocha, J., Tipparaju, V., Apra, E.: An evaluation of two implementation strategies for optimizing one-sided atomic reduction. In: IPDPS (2005)
Liu, J., Jiang, W., Jin, H.W., Panda, D.K., Gropp, W., Thakur, R.: High Performance MPI-2 One-Sided Communication over InfiniBand. In: CCGrid 2004 (April 2004)
Mourao, F.E., Silva, J.G.: Implementing MPI’s One-Sided Communications for WMPI. newblock In: EuroPVM/MPI (September 1999)
Thacker, R.J., Pringle, G., Couchman, H.M.P., Booth, S.: Hydra-mpi: An adaptive particle-particle, particle-mesh code for conducting cosmological simulations on mpp architectures. High Performance Computing Systems and Applications (2003)
HPCC Benchmark Suite. http://icl.cs.utk.edu/hpcc
Gropp, W., Lusk, E., Skjellum, A.: Using MPI:Portable Parallel Programming with the Message Passing Interface, 2nd edn. MIT Press, Cambridge, MA (1999)
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 2007 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Santhanaraman, G., Narravula, S., Mamidala, A.R., Panda, D.K. (2007). MPI-2 One-Sided Usage and Implementation for Read Modify Write Operations: A Case Study with HPCC . In: Cappello, F., Herault, T., Dongarra, J. (eds) Recent Advances in Parallel Virtual Machine and Message Passing Interface. EuroPVM/MPI 2007. Lecture Notes in Computer Science, vol 4757. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-540-75416-9_36
Download citation
DOI: https://doi.org/10.1007/978-3-540-75416-9_36
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-75415-2
Online ISBN: 978-3-540-75416-9
eBook Packages: Computer ScienceComputer Science (R0)
Keywords
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.