View online
Abstract
Linux/pSeries clusters for HPC (High Performance Computing) are a new offering which is attracting more and more clients. Often, these clients are already using a Linux/xSeries cluster and would like to benefit from the outstanding 64-bit floating point performance and capability offered by Power4-based servers while still retaining Linux as their operating system.
This IBM Redpaper explains how to bring a collection of pSeries nodes and a Myrinet interconnect to a state where a true HPC production workload can be run in a Linux clustered environment. Already very popular with Linux/xSeries clusters, Myrinet is the high-speed and low-latency network that we will use for distributed memory applications on Linux/pSeries clusters.
Table of contents
Introduction
Cluster setup
Myrinet interconnect
Compiling and running MPI applications
A global cluster filesystem
Batch job scheduler
Future work
The authors of this redpaper
Acknowledgements
References
Appendix A: Building shared libraries
Appendix B: Building RPMs