Next: Introduction

Using Heterogeneous Clusters of PCs
for Parallel Computing

Zdenek Bittnar, Jaroslav Kruis, Jirí Nemecek, Borek Patzák, Daniel Rypl


Department of Structural Mechanics
Faculty of Civil Engineering
Czech Technical University in Prague
Thákurova 7, 166 29 Prague, Czech Republic

Abstract:

Typical engineering design offices are equipped with several PC computers with different processor power, available memory and disk capacity. The computers are usually connected with fast Ethernet. The computing performance is relatively high with respect to the demands of design process involving CAD, structural analysis, etc. This makes the offices well equipped for high performance computing provided relevant parallel software is available. The aim of this contribution is to review the progress in this area with the special attention to the latest results of our group.

In order to develop parallel application, suitable communication software has to be installed on individual computers. Most common are PVM and MPI, both available as public domain software running under Windows NT as well as Linux operating systems.

Although there may be different combinations of hardware and software forming a parallel computing environment it can be shown that the results (computational time, speedup, efficiency) achieved on such cluster are competitive to results obtained on special parallel computers, as IBM SP2, SGI Origin, etc. This will be demonstrated on applications in mesh generation and analysis of highly nonlinear systems by implicit as well as explicit algorithms.

It is important to realize that from engineering point of view the scalability of the algorithm is not the only criterion to judge the efficiency of parallel application. In many case, the ability to analyze extremely large problems not solvable on individual machines is of primary interest.





Next: Introduction

Daniel Rypl
2005-12-03