Parallelization of Temperature Distribution Simulations for Semiconductor and Polymer Composite Material on Distributed Memory Architecture

摘要

The implementations of parallel algorithms in solving partial differential equations (PDEs) for heat transfer problems are based on the high performance computing using distributed memory architecture. In this paper, the parallel algorithms are exploited finite difference method in solving multidimensional heat transfer problem for semiconductor components and polymer composite materials. Parallel Virtual Machine (PVM) and C language based on Linux operating system are the platform to run the parallel algorithms. This research focused on Red-Black Gauss Seidel (RBGS) iterative method. Parallel performance evaluations in terms of speedup, efficiency, effectiveness, temporal performance and communication cost are analyzed.