Abstract
The National Science Foundation released a set of application benchmarks that would be a key factor in selecting the next-generation high-performance computing environment. These benchmarks consist of six codes that require large amount of memory and work with large data sets. Here we study the complexity, performance and scalability of these codes on three SGI machines: a 512-processor Altix 3700, a 512-processor Altix 3700/BX2 and 512-processor dual-core based Altix 4700; and a 128-processor Cray Opteron cluster interconnected by the Myrinet network. We evaluated these codes for two different problem sizes using different numbers of processors. Our results show that some of these codes scale up very well as we increase the number of processors while others scaled up poorly. Also, one code achieved about 2/3 of the peak rate of an SGI Altix processor. Moreover, the dual-core system achieved comparable performance results to the single-core system.
Acknowledgements
This research was partially supported by the NASA Postdoctoral Program at Ames Research Center, administered by Oak Ridge Associated Universities through a contract with NASA. The author also likes to thank the staff at NASA Advanced Supercomputing Division for providing access to the machines there and assistance in running some of the codes.