A comparison of the CM with the DAP for lattice gauge theory

Abstract

Lattice gauge theory is one of the most challenging large-scale scientific computations; a state of the art calculation requires at least 10 14 floating-point operations, necessitating the use of advanced architecture massively parallel computers such as the Connection Machine (CM) made by Thinking Machines Corporation (TMC), and the Distributed Array Processor (DAP) made in the past by International Computers Limited (ICL) and currently by active Memory Technology (AMT). The most important gauge theory to be solved is that descrining the sub-nuclear world of high energy physics: Quantum Chromodynamics (QCD). The simples example of a gauge theory is Quantum Electro-dynamics (QED), the theory which describes the interaction of electrons and photons. Simulation of QCD requires computer software very similar to that for the simpler QED problem. Thus, as a first step towards computer simulation of QCD, we have developed code for QED on the CM, and compared this with similar code for the DAP. Experience with the DAP allows us to predict performances for QCD code on the CM, showing the latter to be a very serious proposition for such large-scale scientific computations.