Please use this identifier to cite or link to this item:
|Title:||Towards predictive modeling of message-passing communication||Authors:||March, V.
|Issue Date:||2009||Citation:||March, V., Murali, V., Teo, Y.M., See, S., Himer, J.T. (2009). Towards predictive modeling of message-passing communication. 2009 11th IEEE International Conference on High Performance Computing and Communications, HPCC 2009 : 482-487. ScholarBank@NUS Repository. https://doi.org/10.1109/HPCC.2009.94||Abstract:||Communication has been shown to be a performance bottleneck and a limiting factor of many large parallel applications. As such, predicting the application scalability necessitates a communication performance model. This paper investigates the LogGP communication performance model for predicting message-passing communications when the system configuration (i.e., number of nodes) is varied. The cost functions for the message-passing operations are based on MVAPICH2 1.0, and the experiments are conducted on the Ranger system using up to 256 nodes connected with an InfiniBand network. For point-to-point communications, we observe that the LogGP model accurately predicts the communication performance. However, the results for three collective operations, i.e., MPI-Barrier, MPI-Alltoall, and MPI-Bcast, are varying. For MPI-Bcast, the LogGP model is able to predict its scalability up to 256 nodes, and the prediction error is at most a factor of two on 256 nodes. For the remaining collectives, the scalability - bar that of MPI-Alltoall on small messages (m = 2 bytes) - is predicted by LogGP, but the prediction error for 256 nodes is 3.5-12 times of the measured performance. © 2009 IEEE.||Source Title:||2009 11th IEEE International Conference on High Performance Computing and Communications, HPCC 2009||URI:||http://scholarbank.nus.edu.sg/handle/10635/40458||ISBN:||9780769537382||DOI:||10.1109/HPCC.2009.94|
|Appears in Collections:||Staff Publications|
Show full item record
Files in This Item:
There are no files associated with this item.
checked on Apr 7, 2019
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.