Abstract
MPI provides a set of primitives that allow processes to dynamically create communicators on the fly. This set of primitives can be exploited by the applications where only a certain group of processes need to participate at any given time. Also, these primitives play an important role in the context of dynamic process management of MPI-2. Special attention has to be paid in creating MPI communicators with InfiniBand’s hardware multicast support as it involves the high overhead of interaction between the application and an external multicast management entity. In this paper, we propose different design alternatives of efficiently creating the communicators dynamically. The basic idea behind the schemes proposed is to remove most of the overhead of the hardware multicast group construction from the critical path of the application. Our results indicate that by using Multicast Pool and Lazy approaches of group construction proposed in the paper, we can significantly reduce the overhead by a factor of as much as 4.8 and 3.9, repectively, compared to the Basic approach.
This research is supported in part by Department of Energy’s Grant #DE-FC02-01ER25506; National Science Foundation’s grants #CCR-0204429, #CCR-0311542 and #CNS-0403342; grants from Intel and Mellanox; and equipment donations from Intel, Mellanox, AMD and Apple.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
Similar content being viewed by others
References
Multicast collectives, http://vmi.ncsa.uiuc.edu
MVAPICH: MPI over InfiniBand Project, http://nowlab.cis.ohio-state.edu/projects/mpi-iba/
Bermudez, A., Casado, R., Quiles, F.J., Pinkston, T.M., Duato, J.: Evaluation of a Subnet Management Mechanism for InfiniBand Networks. In: Proceedings of ICPP (2003)
Bermudez, A., Casado, R., Quiles, F.J., Pinkston, T.M., Duato, J.: On the InfiniBand Subnet Discovery Process. In: Proceedings of Cluster Computing (2003)
Chen, H.A., Carrasco, Y.O., Apon, A.W.: MPI Collective Operations over IP Multicast. In: Workshop PC-NOW 2000 (2000)
InfiniBand Trade Association. InfiniBand Architecture Specification, Release 1.2 (October 2004), http://www.infinibandta.org
Liu, J., Mamidala, A.R., Panda, D.K.: Fast and Scalable MPI-Level Broadcast using InfiniBand’s Hardware Multicast Support. In: Proceedings of IPDPS (2004)
Mamidala, A.R., Liu, J., Panda, D.K.: Efficient Barrier and Allreduce InfiniBand Clusters using Hardware Multicast and Adaptive Algorithms. In: Proceedings of Cluster Computing (2004)
Sancho, J.C., Robles, A., Duato, J.: Effective Strategy to Compute Forwarding Tables for InfiniBand Networks. In: Proceedings of ICPP (2001)
Snir, M., Otto, S., Huss-Lederman, S., Walker, D., Dongarra, J.: MPI–The Complete Reference. The MPI-1 Core, 2nd edn., vol. 1. The MIT Press, Cambridge (1998)
Yuan, X., Daniels, S., Faraj, A., Karwande, A.: Group Management Schemes for Implementing MPI Collective Communication over IP-Multicast. In: The 6th International Conference on Computer Science and Informatics, Durham, NC, March 8-14 (2002)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2005 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Mamidala, A.R., Jin, HW., Panda, D.K. (2005). Efficient Hardware Multicast Group Management for Multiple MPI Communicators over InfiniBand. In: Di Martino, B., Kranzlmüller, D., Dongarra, J. (eds) Recent Advances in Parallel Virtual Machine and Message Passing Interface. EuroPVM/MPI 2005. Lecture Notes in Computer Science, vol 3666. Springer, Berlin, Heidelberg. https://doi.org/10.1007/11557265_50
Download citation
DOI: https://doi.org/10.1007/11557265_50
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-29009-4
Online ISBN: 978-3-540-31943-6
eBook Packages: Computer ScienceComputer Science (R0)