Study of Optimal Parallel Computation for Molecular Dynamics

碩士 === 國立臺北科技大學 === 機電整合研究所 === 91 === The thesis use PC Cluster and Parallel computational PROM algorithms to reduce the CPU time of Molecular Dynamics program of “ar.for” that is modified by NTFL lab. The PC Cluster system, composed by two PC, runs by Linux OS kernel. By using MPICH (me...

Full description

Bibliographic Details
Main Authors: Banjamin Dong, 董仲桓
Other Authors: C. S. wang
Format: Others
Language:zh-TW
Published: 2003
Online Access:http://ndltd.ncl.edu.tw/handle/94298481022512778528
id ndltd-TW-091TIT00651011
record_format oai_dc
spelling ndltd-TW-091TIT006510112016-06-13T04:16:31Z http://ndltd.ncl.edu.tw/handle/94298481022512778528 Study of Optimal Parallel Computation for Molecular Dynamics 以平行計算在分子動力學上計算最適性之研究 Banjamin Dong 董仲桓 碩士 國立臺北科技大學 機電整合研究所 91 The thesis use PC Cluster and Parallel computational PROM algorithms to reduce the CPU time of Molecular Dynamics program of “ar.for” that is modified by NTFL lab. The PC Cluster system, composed by two PC, runs by Linux OS kernel. By using MPICH (message transfer interface program) and FORTRAN compiler language to recompose the sequence code as parallel code We develop the PROM algorithms, including Parallization, Ram sequence continue, Optimization of anti-symmetric matrix, and Manipulation transfer buffer zone. It can save both the computational CPU time and the transfer time in buffer zone. We explored the CPU time, data exchange time and the parallel efficiency by our PROM algorithm. There are two important tasks to reduce the CPU time and transfer time. Firstly, by packing transfer data or replace transfer with calculation. Secondly, by balancing the calculation load for both CPU which runs the parallel program. By utilizing PROM algorithms, the CPU time will be reduced to 24.4% which is compared with original code C. S. wang 王金樹 2003 學位論文 ; thesis 103 zh-TW
collection NDLTD
language zh-TW
format Others
sources NDLTD
description 碩士 === 國立臺北科技大學 === 機電整合研究所 === 91 === The thesis use PC Cluster and Parallel computational PROM algorithms to reduce the CPU time of Molecular Dynamics program of “ar.for” that is modified by NTFL lab. The PC Cluster system, composed by two PC, runs by Linux OS kernel. By using MPICH (message transfer interface program) and FORTRAN compiler language to recompose the sequence code as parallel code We develop the PROM algorithms, including Parallization, Ram sequence continue, Optimization of anti-symmetric matrix, and Manipulation transfer buffer zone. It can save both the computational CPU time and the transfer time in buffer zone. We explored the CPU time, data exchange time and the parallel efficiency by our PROM algorithm. There are two important tasks to reduce the CPU time and transfer time. Firstly, by packing transfer data or replace transfer with calculation. Secondly, by balancing the calculation load for both CPU which runs the parallel program. By utilizing PROM algorithms, the CPU time will be reduced to 24.4% which is compared with original code
author2 C. S. wang
author_facet C. S. wang
Banjamin Dong
董仲桓
author Banjamin Dong
董仲桓
spellingShingle Banjamin Dong
董仲桓
Study of Optimal Parallel Computation for Molecular Dynamics
author_sort Banjamin Dong
title Study of Optimal Parallel Computation for Molecular Dynamics
title_short Study of Optimal Parallel Computation for Molecular Dynamics
title_full Study of Optimal Parallel Computation for Molecular Dynamics
title_fullStr Study of Optimal Parallel Computation for Molecular Dynamics
title_full_unstemmed Study of Optimal Parallel Computation for Molecular Dynamics
title_sort study of optimal parallel computation for molecular dynamics
publishDate 2003
url http://ndltd.ncl.edu.tw/handle/94298481022512778528
work_keys_str_mv AT banjamindong studyofoptimalparallelcomputationformoleculardynamics
AT dǒngzhònghuán studyofoptimalparallelcomputationformoleculardynamics
AT banjamindong yǐpíngxíngjìsuànzàifēnzidònglìxuéshàngjìsuànzuìshìxìngzhīyánjiū
AT dǒngzhònghuán yǐpíngxíngjìsuànzàifēnzidònglìxuéshàngjìsuànzuìshìxìngzhīyánjiū
_version_ 1718302866640207872