Parallel visualization for molecular dynamics simulations

✍ Scribed by J. Kitowski; K. Boryczko

• Publisher: Elsevier Science
• Year: 1999
• Tongue: English
• Weight: 48 KB
• Volume: 121-122
• Category: Article
• ISSN: 0010-4655
• DOI: 10.1016/s0010-4655(06)70083-2

✦ Synopsis

Scientific visualization is an important aspect of numerically intensive computing. While a vast amount of simulation algorithms exists, visualization systems are still a field of scientific interest (e.g. [1,).

Molecular dynamics algorithms (and others, like dissipative particle dynamics) while used for, e.g., thermal instabilities (e.g. ) generate huge amounts of data. Visualization is the only way to cope with the results and due to memory and performance requirements parallelization is a method to use.

The proposed parallel visualization algorithm incorporates two different strategies for explicit parallelization. Domain decomposition, in which a computational box is devided into parts forms independently parts of the screen with particles. The functional decomposition (organized in a pipeline fashion) prepares vizualization frames for presentation in advance.

In practice two pilot systems were elaborated. The first, PIMD, is originally written in C language and incorporates Motif 1.2 (with XI 1) as well as MPI libraries. The second makes use of Visual3 [5] (and Fortran77), OpenGL and PVM software.

The algorithms are suitable for networks of workstations and parallel computers. For exanaple, for PIMD we obtain 50-60% efficiency on SGI Origin2000.