Improved performance of QCD code on ALiCE

Abstract

We present results for the performance of QCD code on ALiCE, the Alpha-Linux Cluster Engine at Wuppertal. We describe the techniques employed to optimise the code, including the metaprogramming of assembler kernels, the effects of data layout and an investigation into the overheads incurred by the communication.

Comments: 3pp. Latex2e using packages espcrc2 and graphicx with 7 PostScript figures. Contribution to Lattice2002(machines)