Algorithms for Parallel Computing
Effective use of large parallel
computers depends upon the availability of scalable parallel algorithms and
implementations. Sandia has a long history of research and development activity
in this area, with a focus on algorithms to enable parallel science and engineering
simulations. While the research activities have been spread across a number
of application areas, many of the fruits of these efforts can be found in
the Zoltan tool for parallel data management.
 |
|
The
Zoltan toolkit is used for dynamic load balancing and data migration in the Sandia
code CHISELS, a feature-scale modeler for semiconductor surface growth.
The computational grid is refined around the deposition surface (shown
as a red line). Zoltan's Recursive Coordinate Bisection algorithm generates
partitions (indicated by the colored regions) that
equally distribute grid cells to processors while maintaining geometric locality
of objects in processors. (Image courtesy of Steve Plimpton and Larry Musson.) |
Areas of Research:
Program Contact: Suzanne L. K. Rountree
Return to Top of the Page

Newsnotes | Info
and Events (internal - SNL only) | Open-Source
Software Downloads | Privacy
and Security
Sandia National
Laboratories Home Page - External or Internal
(SNL only)
Maintained by: Bernadette M. Watts
Modified on:
May 6, 2008
|