On multigrid methods for the Navier-Stokes ComputerThe overall architecture of the multipurpose parallel-processing Navier-Stokes Computer (NSC) being developed by Princeton and NASA Langley (Nosenchuck et al., 1986) is described and illustrated with extensive diagrams, and the NSC implementation of an elementary multigrid algorithm for simulating isotropic turbulence (based on solution of the incompressible time-dependent Navier-Stokes equations with constant viscosity) is characterized in detail. The present NSC design concept calls for 64 nodes, each with the performance of a class VI supercomputer, linked together by a fiber-optic hypercube network and joined to a front-end computer by a global bus. In this configuration, the NSC would have a storage capacity of over 32 Gword and a peak speed of over 40 Gflops. The multigrid Navier-Stokes code discussed would give sustained operation rates of about 25 Gflops.
Document ID
19880055935
Acquisition Source
Legacy CDMS
Document Type
Conference Paper
Authors
Nosenchuck, D. M. (Princeton University NJ, United States)
Krist, S. E. (George Washington University Hampton, VA, United States)
Zang, T. A. (NASA Langley Research Center Hampton, VA, United States)
Date Acquired
August 13, 2013
Publication Date
January 1, 1988
Subject Category
Computer Systems
Meeting Information
Meeting: Multigrid methods: Theory, applications, and supercomputing