Linux-based High Performance Computing Clusters Design and Implementation

As components of computers and network are increasing to achieve commercialization and standardization,the performance of PCs or workstations is increasing while the price is becoming cheaper.At the same time,Linux kernel and open-source cluster middleware technology are becoming increasingly mature and stable.Clusters are mainstream high performance computing platforms gradually.The use of high performance computing clusters to prototype,debug,and run parallel applications is becoming an increasingly popu-lar alternative to using specialized,especially expensive,supercomputing platforms.It is full of significance for scientific and engineering computing applications to research on deployment high performance computing clusters based on PCs or workstations rapidly as well as its reliability and manageability.This paper provides a case study about OSCAR cluster with five PCs nodes and fast Ethernet interconnect,and then run some simple parallel examples on it.