Performance analysis of an implementation of the Beam and Warming implicit factored scheme on the NCUBE hypercube

A production 3-D Beam and Warming implicit Navier Stokes code has been implemented on the NCUBE hypercube using the grid allocation scheme of J. Bruno and P.R. Capello (see Proc. 3rd Conf. on Hypercube Concurrent Computers and Applications, p.1073-87, 1988). Predicted (32-b) performance on 1024 nodes is 67.1 MFLOPS. Efficiencies of 70% are attainable for implicit algorithms, although constant-memory scaled performance is found to decrease with increasing number of nodes, unlike the case for explicit implementations