#TITLE_ALTERNATIVE#

• Main Author: SAPUTRA (NIM : 10206026); Pembimbing : Dr. Rizal Kurniadi, ADY
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/14679
• Institution: Institut Teknologi Bandung

Parallel computation is widely used in many study field, including in physical system computation. The purpose of parallel computation is to optimizing the usability of computer and network resources to get faster computation. Conjugate gradient method is a numerical method based on Krylov’s Subspace that commonly use to solve a symmetric-positif-definite linear equation in many scientific computation. This Final Assignment research purposing in optimizing conjugate gradient method with parallel computation in distributed computing system. <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> The description begin with teoritical insight of physical problem that involving linear equation system, conjugate gradient method theoritical background and explanation of program design and the infrastructure of distributed computing system which used. Physical problem which become focus on this research is study of Poisson equation in 2-dimensional potential distributon. <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> <br /> This research prove the effect of processor number addition in multicore and multinode computer environment, the change of size problem, the change of distribution scheme and the change of block size. CG parallel program are evaluated in their speedup, granularity and efficiency. It is proven that CG multicore computation still has better performance than CG multinode computation. CG multinode performance is strongly influenced of its communication performance.