🚀 go-pugleaf

The computational problem is molecular dynamics simulations. x,y,z
coordinates of the atoms in a molecule at different time steps are
one of the outputs produced by such simulations. These coordinates
are then used by a display program to display the molecule's dynamics
on a  workstation (Silicon Graphics INDIGO Elan 4.0.5A). As molecular
dynamics is computationally intensive, we are exploring ways to 
speed up the simulation using parallel computers (Intel Paragon, MasPar, 
nCUBE 2S). These machines can be accessed from the Silicon Graphics 
Workstations via ethernet. 

As the parallel computer calculates those configurations ( Roughly, a 
configuration is one set of x,y,z coordinates) we would like to be able 
to display them on the SGI. The idea is to demonstrate to a naive
audience that indeed parallelism is able to speed up the computations.
(You would perhaps agree with me that "naive audience" prefers graphics 
demos to algorithmic proofs :-) ). 

Previous experience on similar problems, information on public domain or 
commercial software, any other ideas on convincing the audience would be
greatly helpful. 

Vamsee
(vamsee@tamu.edu)