Large scale QCD Monte Carlo calculations have typically been performed on either commercial supercomputers or specially built massively parallel computers such as Fermilab's ACPMAPS.Commodity clusters equipped with high performance networking equipment present an attractive alternative,achieving superior performance to price ratios and offering clear upgrade paths.We describe the construction and results to date of Fermilab's prototype production cluster,which consists of 80 dual Pentium Ⅲsystems interconnected with Myrinet networking hardware.We describe software tools and techniques we have developed for operating system installation and administration.We discuss software optimizations using the Pentium's built-in parallel computation facilities(SSE),Finally,we present short and long term plans for the construction of larger facilities.
展开▼
