The CP-PACS Project and Lattice QCD Results

Abstract

Lattice QCD is a fundamental theory of quarks and gluons which are constituents of hadrons such as protons and pions. Numerical studies of lattice QCD have developed significantly during the past decade in parallel with the development of computers. Of particular importance in this regard has been the construction of dedicated QCD computers (see for reviews Ref. 1)) and the move of commercial vendors toward parallel computers in recent years. In Japan the first dedicated QCD computer was developed in the QCDPAX project. 2) The QCDPAX computer with a peak speed of 14 GFLOPS is actually the 5th computer in the PAX project, 3) which pioneered the development of parallel computers for scientific and engineering applications in Japan. The CP-PACS project was conceived as a successor of the QCDPAX project in the early summer of 1991. The project name CP-PACS is an acronym for Computational Physics by a Parallel Array Computer System. The aim of the project was to develop a massively parallel computer for carrying out research in computational physics with primary emphasis on lattice QCD. In this article after a brief description of lattice QCD and the background of the project in §2, we present an overview of the CP-PACS project in §3, and describe characteristics of the CP-PACS computer in §4. The performance of the computer for lattice QCD applications as well as for the LINPACK benchmark are also given. Main results in lattice QCD are given in §5. Section 6 is devoted to conclusions. §2. Lattice QCD and background of the project Lattice QCD is a fundamental theory of quarks and gluons defined in terms of the path-integral formalism of quantum theory on a 4-dimensional hyper-cubic lattice. The lattice spacing plays a role of an ultra-violet cutoff. The infinite volume limit and the continuum limit should be taken in order to get physical quantities.