Parallel implementations of the simplex algorithm
Richard J. Marciano, Teodor Rus · 2003
Three parallel implementations of the simplex algorithm on three different parallel architectures are presented and compared. Each machine is the representative of one class of parallel computers. Performance comparisons and the major difficulties encountered by users of these machines are given. The potential for parallel programming of the array processors is investigated with the massively parallel processor (MPP) machine. Multiprocessor systems with asynchronous shared memory are studied by implementing the simplex algorithm on the Encore machine in both the process creation by fork() and the tasking environment. The class of supercomputers represented by the Alliant FX-8 minisupercomputer where a Fortran compiler can parallelize and vectorize DO loops is considered.>