Introduction to parallel computing: design and analysis of algorithms
Introduction to parallel computing: design and analysis of algorithms
Compiling Communication-Efficient Programs for Massively Parallel Machines
IEEE Transactions on Parallel and Distributed Systems
| Hi-index | 0.00 |
The parallelization of Fourier spectral and Chebyshev collocation method to the cylindrical form of Navier Stokes equations with one periodic and two non-periodic directions has been implemented. Matrix diagonalization technique applied to solve the elliptic equations which occur in the implicit steps of three-step time-splitting algorithm enables one to parallelize only the matrix-matrix multiplication routines. Parallel matrix multiplication algorithms of Canon and Fox have been thus explored and compared between each other in details. The speed up of up to 3.7 in the Intel Paragon parallel computer has been obtained compared with the CRAY YMP vector computer.