Performance comparison between a massive SMP machine and clusters

In this brief report we compared the performance of selected scientific applications on sun-test a massive SMP 32-core machine with blade and zebra, the two main partitions of the production cluster at SISSA. In tables 1 and 2 some details on the hardware are presented. Information on the interconnection network is available in table 3 along with some network performance results, obtained with the Intel MPI Benchmark. Our main aim was to asses if a single massive multicore machine is a viable alternative to Beowulf-type clusters when running scientific applications. No considerations regarding hardware costs and power consumption were made. Two identical sun-test machines were also connected via high speed Infiniband network to see if the code scalability was good among two such nodes.