A tunable collective communication framework on a cluster of SMPS

In this paper we investigate a tunable MPI collective communications library on a cluster of SMPs. Most tunable collective communications libraries select optimal algorithms for inter-node communication on a given platform. We add another layer of intra-node communications composed by several tunable shared memory operations. We explore the advantages of our approach, and discuss when to use our approach, when to switch to another approach on the shared memory layer. Experimental results indicate that collective communications designed by such an approach with proper tuning can outperform vendor implementations.

[1]  Dhabaleswar K. Panda,et al.  Fast collective operations using shared and remote memory access protocols on clusters , 2003, Proceedings International Parallel and Distributed Processing Symposium.

[2]  William Gropp,et al.  A High-Performance MPI Implementation on a Shared-Memory Vector Supercomputer , 1997, Parallel Comput..

[3]  Maciej Go.  Algorithms for Collective Communication Operations on Smp Clusters , 1999 .

[4]  David E. Culler,et al.  Multi Protocol Active Messages on a Cluster of SMP , 1997, ACM/IEEE SC 1997 Conference (SC'97).

[5]  Sathish S. Vadhiyar,et al.  ACCT: Automatic Collective Communications Tuning , 2000, PVM/MPI.

[6]  Lionel M. Ni,et al.  Performance Metrics and Measurement Techniques of Collective Communication Services , 1997, CANPC.

[7]  David A. Bader,et al.  SIMPLE: A Methodology for Programming High Performance Algorithms on Clusters of Symmetric Multiprocessors (SMPs) , 1998, J. Parallel Distributed Comput..

[8]  Rajeev Thakur,et al.  Improving the Performance of Collective Operations in MPICH , 2003, PVM/MPI.

[9]  Ramesh Subramonian,et al.  LogP: towards a realistic model of parallel computation , 1993, PPOPP '93.

[10]  Robert A. van de Geijn,et al.  Building a high-performance collective communication library , 1994, Proceedings of Supercomputing '94.

[11]  Steve Sistare,et al.  Optimization of MPI Collectives on Clusters of Large-Scale SMP's , 1999, SC.

[12]  William Gropp,et al.  Reproducible Measurements of MPI Performance Characteristics , 1999, PVM/MPI.

[13]  Robert A. van de Geijn,et al.  Fast Collective Communication Libraries, Please , 1995 .

[14]  Jeffrey S. Vetter,et al.  Communication characteristics of large-scale scientific applications for contemporary cluster architectures , 2002, Proceedings 16th International Parallel and Distributed Processing Symposium.

[15]  Werner Augustin,et al.  Benchmarking Collective Operations with SKaMPI , 2003 .

[16]  J. Watts,et al.  Interprocessor collective communication library (InterCom) , 1994, Proceedings of IEEE Scalable High Performance Computing Conference.

[17]  Jehoshua Bruck,et al.  CCL: A Portable and Tunable Collective Communication Library for Scalable Parallel Computers , 1995, IEEE Trans. Parallel Distributed Syst..

[18]  Lars Paul Huse.  Collective Communication on Dedicated Clusters of Workstations , 1999, PVM/MPI.

[19]  Sathish S. Vadhiyar,et al.  Automatically Tuned Collective Communications , 2000, ACM/IEEE SC 2000 Conference (SC'00).