Parallel Computing with the R Language in a Supercomputing Environment

Abstract

R is an open-source programming language and software environment for statistical computing and graphics. During the last decade a great deal of research has been conducted on parallel computing techniques with the R language. Two packages (snow and Rmpi) stand out as particularly useful for general use on computer clusters and the multicore package for the use on multi-core machines. This article describes the operation of the R language at the supercomputer HLRB2 hosted at the Leibniz-Rechenzentrum in Munich, Germany. Additional, a small benchmark is provided and the article explains and discusses two parallel biostatistical applications calculated at the HLRB2. The indirect comparison of interaction graph example outlines the requirements for more than 10.000 processors.