Efficient decomposition and performance of parallel PDE, FFT, Monte Carlo simulations, simplex, and sparse solvers (Q1186822)

Investigation of parallelization for several versions of the algorithms named in the title (PDE stands for red-black successive overrelaxation and alternating directions implicit methods, sparse solver for triangular sparse linear system) on two experimental shared memory parallel processors: An 8 processor variant for the VAX 6300 series and a M31, consisting of 16 Micro VAX CPUs. The two computers have scalar processors with different cache design. The presence of the caches forces to use as far as possible ``local'' data with corresponding consequences for the algorithms.