Progress toward the full implementation of a version of TACO3D, a three-dimensional thermal analysis code, for use on the CM-2 Connection Machine is described. A combination of a parallel conjugate gradient solver, which runs on the CM-2 and a serial calling program which resides on the VAX front-end, is implemented in order to assess the feasibility of utilizing this type of massively parallel computer in the engineering production environment. Preliminary results show that this implementation, running on a 16 K processor CM-2, is over five times faster than a single XMP processor in the solution of a ~30000-node three-dimensional application. Future plans to complete the optimum implementation of the entire code are also discussed
Published in:
Frontiers of Massively Parallel Computation, 1988. Proceedings., 2nd Symposium on the Frontiers of
Date of Conference: 10-12 Oct 1988