Abstract
We discuss a new, stand-alone MPI benchmark program (for now called mpicroscope) for assessing the performance of collective communication patterns, in particular of the standard MPI 2.2 collective operations. The benchmark is intended to facilitate comparisons between different ways of expressing the same functionality towards performing automatic detection of violations of self-consistent MPI performance guidelines. The benchmark can be used by MPI library developers for assessing the relative quality of new algorithms and implementations. It can also be used to make users aware of aspects of their MPI library where performance (portability) problems may lurk. The current version of the benchmark automatically detects, for any measured communication pattern, two universal, self-consistent guidelines that encourage monotone and split-robust performance.
The benchmark aims to employ sound benchmarking procedures and is controlled via command-line options. It covers the MPI 2.2 collective operations, and a number of alternative patterns that express MPI collective functionality. In contrast to many other benchmarks data can be structured as described by an MPI derived datatype. We present results from a small InfiniBand cluster with a vendor MPI library, showing performance guideline violations that were detected and highlighted with mpicroscope.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Worsch, T., Reussner, R., Augustin, W.: On Benchmarking Collective MPI Operations. In: Kranzlmüller, D., Kacsuk, P., Dongarra, J., Volkert, J. (eds.) PVM/MPI 2002. LNCS, vol. 2474, pp. 271–279. Springer, Heidelberg (2002)
Augustin, W., Worsch, T.: Usefulness and Usage of SKaMPI-Bench. In: Dongarra, J., Laforenza, D., Orlando, S. (eds.) EuroPVM/MPI 2003. LNCS, vol. 2840, pp. 63–70. Springer, Heidelberg (2003)
Fox, G.C., et al.: Solving Problems on Concurrent Processors: General Techniques and Regular Problems. Prentice-Hall (1988)
Gropp, W.D., Lusk, E.: Reproducible Measurements of MPI Performance Characteristics. In: Margalef, T., Dongarra, J., Luque, E. (eds.) PVM/MPI 1999. LNCS, vol. 1697, pp. 11–18. Springer, Heidelberg (1999)
Gropp, W., Hoefler, T., Thakur, R., Träff, J.L.: Performance Expectations and Guidelines for MPI Derived Datatypes. In: Cotronis, Y., Danalis, A., Nikolopoulos, D.S., Dongarra, J. (eds.) EuroMPI 2011. LNCS, vol. 6960, pp. 150–159. Springer, Heidelberg (2011)
Gropp, W.D., Kimpe, D., Ross, R., Thakur, R., Träff, J.L.: Self-consistent MPI-IO Performance Requirements and Expectations. In: Lastovetsky, A., Kechadi, T., Dongarra, J. (eds.) EuroPVM/MPI 2008. LNCS, vol. 5205, pp. 167–176. Springer, Heidelberg (2008)
Grove, D., Coddington, P.: Precise MPI performance measurement using MPIBench. In: Proceedings of HPC Asia (2001)
Hamid, N.A.W.A., Coddington, P.D., Vaughan, F.: Comparison of MPI benchmark programs on shared memory and distributed memory machines (point-to-point communication). International Journal on High Performance Computing Applications 24(4), 469–483 (2010)
Hoefler, T., Schneider, T., Lumsdaine, A.: Accurately measuring collective operations at massive scale. In: 22nd IEEE International Symposium on Parallel and Distributed Processing (IPDPS), pp. 1–8 (2008)
Lastovetsky, A., Rychkov, V., O’Flynn, M.: MPIBlib: Benchmarking MPI Communications for Parallel Computing on Homogeneous and Heterogeneous Clusters. In: Lastovetsky, A., Kechadi, T., Dongarra, J. (eds.) EuroPVM/MPI 2008. LNCS, vol. 5205, pp. 227–238. Springer, Heidelberg (2008)
Mierendorff, H., Cassirer, K., Schwamborn, H.: Working with MPI Benchmarking Suites on ccNUMA Architectures. In: Dongarra, J., Kacsuk, P., Podhorszki, N. (eds.) PVM/MPI 2000. LNCS, vol. 1908, pp. 18–26. Springer, Heidelberg (2000)
MPI Forum. MPI: A Message-Passing Interface Standard. Version 2.2, September 4 (2009), http://www.mpi-forum.org
Reussner, R.: Using SKaMPI for developing high-performance MPI programs with performance portability. Future Generation Computer Systems 19(5), 749–759 (2003)
Reussner, R., Sanders, P., Prechelt, L., Müller, M.S.: SKaMPI: A Detailed, Accurate MPI Benchmark. In: Alexandrov, V.N., Dongarra, J. (eds.) PVM/MPI 1998. LNCS, vol. 1497, pp. 52–59. Springer, Heidelberg (1998)
Reussner, R., Sanders, P., Träff, J.L.: SKaMPI: A comprehensive benchmark for public benchmarking of MPI. Scientific Programming 10(1), 55–65 (2002)
Thakur, R., Gropp, W.D., Rabenseifner, R.: Improving the performance of collective operations in MPICH. International Journal on High Performance Computing Applications 19, 49–66 (2004)
Träff, J.L.: An Improved Algorithm for (Non-commutative) Reduce-Scatter with an Application. In: Di Martino, B., Kranzlmüller, D., Dongarra, J. (eds.) EuroPVM/MPI 2005. LNCS, vol. 3666, pp. 129–137. Springer, Heidelberg (2005)
Träff, J.L., Gropp, W.D., Thakur, R.: Self-consistent MPI Performance Requirements. In: Cappello, F., Herault, T., Dongarra, J. (eds.) PVM/MPI 2007. LNCS, vol. 4757, pp. 36–45. Springer, Heidelberg (2007)
Träff, J.L., Gropp, W.D., Thakur, R.: Self-consistent MPI performance guidelines. IEEE Transactions on Parallel and Distributed Systems 21(5), 698–709 (2010)
Turner, D., Oline, A., Chen, X., Benjegerdes, T.: Integrating New Capabilities into NetPIPE. In: Dongarra, J., Laforenza, D., Orlando, S. (eds.) EuroPVM/MPI 2003. LNCS, vol. 2840, pp. 37–44. Springer, Heidelberg (2003)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2012 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Träff, J.L. (2012). mpicroscope: Towards an MPI Benchmark Tool for Performance Guideline Verification. In: Träff, J.L., Benkner, S., Dongarra, J.J. (eds) Recent Advances in the Message Passing Interface. EuroMPI 2012. Lecture Notes in Computer Science, vol 7490. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-33518-1_15
Download citation
DOI: https://doi.org/10.1007/978-3-642-33518-1_15
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-33517-4
Online ISBN: 978-3-642-33518-1
eBook Packages: Computer ScienceComputer Science (R0)