Junichiro Makino

Heterogeneous multi-computer system: a new paradigm of parallel processing

HMCS (Heterogeneous Multi-Computer System) is a new parallel processing platform combining massively parallel processors for continuum simulation and particle simulation to realize multi-scale computational physics simulations. We constructed a prototype system of HMCS with a general purpose scientific parallel processor CP-PACS and a gravity calculation parallel processor GRAPE-6 connecting them via commodity-base parallel network. On the prototype of HMCS, a microscopic gravity calculation on GRAPE-6 and a macroscopic radiation hydrodynamics calculation on CP-PACS are performed simultaneously to realize detailed simulation on computational astrophysics. In this paper we report the overall concept, design and implementation of HMCS as well as the result of a novel computational simulation for galaxy formation.

GRAPE project for a dedicated tera-flops computer

We are constructing a one tera-flops machine dedicated to astronomical many-body problems. It consists of parallelized GRAPE machines connected to a host workstation. The GRAPE machines only calculate forces between particles in the system by pipeline architecture. We designed and fabricated LSI chips for it, and about 2000 chips are being connected in parallel. The machine will be in operation by summer of 1995. General concept and features of the machine, mode of parallelization, and their merits are discussed in addition to scientific objectives of the project.<<ETX>>

High-Accuracy Treecode Based on Pseudoparticle Multipole Method

We invented the pseudoparticle multipole method (P2M2), a method to express multipole expansion by a distribution of pseudoparticles. We can use this distribution of particles to calculate high order terms in both the Barnes-Hut treecode and FMM. The primary advantage of P2M2 is that it works on GRAPE. Although the treecode has been implemented on GRAPE, we could handle terms only up to dipole, since GRAPE can calculate forces from point-mass particles only. Thus the calculation cost grows quickly when high accuracy is required. With P2M2, the multipole expansion is expressed by particles, and thus GRAPE can calculate high order terms. Using P2M2, we realized arbitrary-order treecode on MDGRAPE-2. Timing result shows MDGRAPE-2 accelerates the calculation by a factor between 20 (for low accuracy) to 150 (for high accuracy). We parallelized the code so that it runs on MDGRAPE-2 cluster. The calculation speed of the code shows close-to-linear scaling up to 16 processors for N ≳ 106.