Development of Lattice QCD Tool Kit on Cell Broadband Engine Processor

TL;DR

This paper presents an implementation of SU(3) matrix multiplication optimized for the Cell Broadband Engine, achieving significant speed improvements and detailing the tuning process for high-performance lattice QCD computations.

Contribution

It introduces a specialized code for SU(3) matrix multiplication on Cell/B.E. and reports optimized performance metrics and tuning procedures.

Findings

Single precision matrix multiplication reaches 227 GFLOPS on QS20.

Data transfer efficiency is 4.6% of hardware peak.

Achieved 7.4% of the theoretical peak speed for the calculation.

Abstract

We report an implementation of a code for SU(3) matrix multiplication on Cell/B.E., which is a part of our project, Lattice Tool Kit on Cell/B.E.. On QS20, the speed of the matrix multiplication on SPE in single precision is 227GFLOPS and it becomes 20GFLOPS {this vaule was remeasured and corrcted.} together with data transfer from main memory by DNA transfer, which is 4.6% of the hardware peak speed (460GFLOPS), and is 7.4% of the theoretical peak speed of this calculation (268.77GFLOPS). We briefly describe our tuning procedure.