Parallel sphere rendering Page: 3 of 17
This article is part of the collection entitled: Office of Scientific & Technical Information Technical Reports and was provided to Digital Library by the UNT Libraries Government Documents Department.
The following text was automatically extracted from the image on this page using optical character recognition software:
Parallel Sphere Rendering
Michael Krogh, Charles Hansen, James Painter Guillaume Colin de Verdiere*
Advanced Computing Laboratory Centre d'Etudes de Limeil-Valenton
Los Alamos National Laboratory CEL-V/DMA/AIM
Los Alamos, New Mexico 87545 94195 Villeneuve-Saint-Georges, France
Sphere rendering is an important method for visualizing molecular dynamics data. This paper
presents a parallel divide-and-conquer algorithm that is almost 90 times faster than current graphics
workstations. To render extremely large data sets and large images, the algorithm uses the MIMD
features of the supercomputers to divide up the data, render independent partial images, and then
finally composite the multiple partial images using an optimal method. The algorithm and perfor-
mance results are presented for the CM-5 and the T3D.
Key Words: Parallel rendering, sphere rendering, compositing.
In recent years, massively parallel processors (MPPs) have proven to be a valuable tool for performing
scientific computation. Available memory on these types of computers is greater than that which is
found on most traditional vector supercomputers. For example, a fully populated 256 node T3D has
16 gigabytes of physical memory. A 1024 node CM-5 contains 32 gigabytes of physical memory. As a
result, scientists who utilize these MPPs can execute their three dimensional simulation models with
much greater detail than previously possible. While current simulations don't typically utilize the
entire memory systems of these machines, it is not uncommon for a data set from a single time-step in a
dynamic simulation to be in excess of several gigabytes. For example, molecular dynamics simulations of
structural materials have reached 600 million atoms [1, 2, 12]. While researchers don't usually perform
simulations with 100 million atoms, 10 million to 40 million atom simulations are becoming routine.
Figure 1 and Figure 2 show images of such data.
With such large data sets, visualization is an essential tool for analyzing the simulation output. Re-
searchers wish to gain insight into both macroscopic phenomenon as well as microscopic phenomenon.
Traditional methods, such as statistical analysis and browsing through the raw simulation data, aren't
adequate by themselves for analyzing data sets ranging in size from gigabytes to terabytes. Visualiza-
tion, because of its high bandwidth, enables a researcher to explore his or her data sets in a more timely
and fruitful manner .
*Author currently at the ACL through a grant from DGA/DRET
Here’s what’s next.
This article can be searched. Note: Results may vary based on the legibility of text within the document.
Tools / Downloads
Get a copy of this page or view the extracted text.
Citing and Sharing
Basic information for referencing this web page. We also provide extended guidance on usage rights, references, copying or embedding.
Reference the current page of this Article.
Krogh, M.; Hansen, C.; Painter, J. & de Verdiere, G.C. Parallel sphere rendering, article, May 1, 1995; New Mexico. (digital.library.unt.edu/ark:/67531/metadc794439/m1/3/: accessed November 12, 2018), University of North Texas Libraries, Digital Library, digital.library.unt.edu; crediting UNT Libraries Government Documents Department.