Performance study of linearly connected parallel machines through simulation

The authors study the performance of linearly connected parallel machines by examining several job scheduling algorithms for these computers. The motivation for this study is the observation that the computational power of personal computers is poorly utilized, and if personal computers are connected in parallel, they must be used instead of expensive mainframe computers. They also study the common problem of sorting large files by using parallel computers and compare the performance with that of sequential computers. These simulations demonstrated that linearly connected personal computers can provide good computing power and speedup if the local computing load on these personal computers is not too high.<
>