MPI PERFORMANCE AT CLUSTER NETWORKING GAUSS ELIMINATION CASE STUDY

• Main Author: Eko Kristus Pramuko, Andik
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/72822
• Institution: Institut Teknologi Bandung

Main goal of parallel computing are to accelarate the process for problem that big and complex. If the problem have been solving by single personal computer is take a long time to procceded. For instance, the process for the next day estimate wheather, which the process more than one day the result is nothing. So, the the problem in parallel programming is about speedup and efficiency. In this Thesis the parallel massage passing library used is MPI. The another earliest massage passing library is PVM. The method beeing used are for measured speedup and efficiency at the problem that can be comparation between sequential and parallel. Because the speedup and efficiency in parallel environtment are comparation between the sequential and parallel process speed. In this case the problem is Gauss elimination. Gauss elimination were choiced because data exchanged in Gauss elimination is very intensive. Composition between load that represented by matrix dimention, process time and number of computer can be get the optimized point of parallel.