PARALLEL COMPUTING IN A NETWORK OF WORKSTATIONS
R. Ogrodowczyk, Krzysztof Murawski · DOAJ (DOAJ: Directory of Open Access Journals) · 2004
In this paper we describe a few architectures and software for parallel-processing computers. We have tested a cluster constructed with the use of MPI. All tests have been performed for one- and two-dimensional magneto-hydrodynamic plasma. We have concluded from the results of these tests that a simple problem should be run in a sequential node, as its execution time does not essentially decrease with the number of processors used. At the same time, the execution time of a complex problem decreases significantly with the number of processors. In the case of twodimensional plasma the acceleration factor has reached the value of 3.7 with the use of 10 processors.