School of Engineering, Tokai University.
1117 Kitakaname Hiratsuka, 259-1292 Japan,
email:nshimizu @ keyaki . cc .u-tokai.ac.jp
In this paper I will present some numerical analysis performance evaluations with workstation clusters(with up to 64 workstations). For the sake of comparison with commercial parallel machines, I adapted my method to some of the NPB(NAS Parallel Benchmarks) such that CG, FT, MG. With process scheduling, direct transfer to user memory, segment directed task distribution, I achieved up to 4 times speedup compared to the pre-tuned programs. The result with 100BASE-T network also will be presented.