The communication challenge for MPP: Intel Paragon and Meiko CS-2
Parallel Computing
In search of clusters (2nd ed.)
In search of clusters (2nd ed.)
Modeling Communication Overhead: MPI and MPL Performance on the IBM SP2
IEEE Parallel & Distributed Technology: Systems & Technology
Performance Evaluation of the Quadrics Interconnection Network
Cluster Computing
Communication overhead for space science applications on the Beowulf parallel workstation
HPDC '95 Proceedings of the 4th IEEE International Symposium on High Performance Distributed Computing
A Design Study of Alternative Network Topologies for the Beowulf Parallel Workstation
HPDC '96 Proceedings of the 5th IEEE International Symposium on High Performance Distributed Computing
Performance Analysis of a Myrinet-Based Cluster
Cluster Computing
HPL performance prevision to intending system improvement
ISPA'04 Proceedings of the Second international conference on Parallel and Distributed Processing and Applications
Modeling message-passing overhead on NCHC formosa PC cluster
GPC'06 Proceedings of the First international conference on Advances in Grid and Pervasive Computing
GREENCOM-CPSCOM '10 Proceedings of the 2010 IEEE/ACM Int'l Conference on Green Computing and Communications & Int'l Conference on Cyber, Physical and Social Computing
Hi-index | 0.00 |
In this paper, we propose an improved model for predicting HPL (High performance Linpack) performance. In order to accurately predict the maximal LINPACK performance we first divide the performance model into two parts: computational cost and message passing overhead. In the message passing overhead, we adopt Xu and Hwang's broadcast model instead of the point-to-point message passing model. HPL performance prediction is a multi-variables problem. In this proposed model we improved the existing model by introducing a weighting function to account for many effects such that the proposed model could more accurately predict the maximal LINPACK performance Rmax. This improvement in prediction accuracy has been verified on a variety of architectures, including IA64 and IA32 CPUs in a Myrinet-based environment, as well as in Quadrics, Gigabits Ethernet and other network environments. Our improved model can help cluster users in estimating the maximal HPL performance of their systems.