Collaborative Data Analysis and Visualization


Project Description:

This project involves the augmentation of PCs and Macs in a large student lab operated by the University of Minnesota.s Academic Distributed Computing Services organization (see figure). When the students are not using the machines, they will be useful to us, as a result of this project, as (1) a high capacity and high aggregate bandwidth storage area network, (2) an image rendering farm, (3) a genomics data mining engine, and (4) a distributed computing platform and Grid computing testbed.


Accomplishments Fall/Winter 2002

HARDWARE

We have determined the hardware configuration for the PCs and tested its performance thoroughly. We now have a very inexpensive PC configuration that achieves performance levels on our applications that have amazed us. We consider the specially outfitted PC together with our software highly optimized for this system as a tool for scientific visualization that would not have been developed save for this project. To a $1000 reconditioned Dell PC we added dual ATA-133 disks, striped in software, so that the PC enjoys 100 MB/sec sustained performance from a 400 GB disk subsystem costing only $620. New Hardware Configuration

SOFTWARE

Because we had never had individual disk subsystems this fast in our lab before, we had to retune much of our visualization software to exploit this stunning new performance (and price/performance) level. Our retuned movie playing software plays full-screen animations from disk on this type of PC at speeds of 20 frames per second (which reflects the maximum rate at which the pixels of the frame buffer can be filled by the graphics engine). Our retuned movie image rendering software, HVR www.lcse.umn.edu/hvr, now performs the volume rendering of a cube of 512x512x512 voxels in just over 2 seconds.

USES

We demonstrated one of our test PCs so configured and with our improved software in the NCSA exhibit booth at the Supercomputing 2002 exhibit. We also put these new tools to use this last fall in a class for freshman in scientific cinematography, for the making of computer movies from scientific simulation data. Some of the better student movies are presented below.

These movies are DivX encoded for compression. You will need the DivX 5.0 codec, or later, in order to play them. One place to download the latest DivX codec is from http://www.divx.com/divx



AVI movie(16.3 MBytes) of fluid turbulence



AVI movie(7.2 MBytes) of fluid turbulence



AVI movie(5.7 MBytes) of fluid turbulence



Red AVI movie(6.7 MBytes) of a red giant star



AVI movie(4.5 MBytes) of a red giant star



AVI movie(16.9 MBytes) of a red giant star