Level 0: (virtual networking) Orlando
The NCSA portable Immersadesk equipped with hardware to accommodate the DuoView CAVE library (=DuoDesk) will demo a real-time interactive animation of a relativistic gravitational lens. Cast as an entertaining partnership game, the application simulates network latency and other difficulties typically encountered when calibrating a distributed, persistent virtual environment.
Recall that the simulation of a gravitatinal lens requires recomputation of the entire scene for each frame since it depends on the position of the tracked viewer, the moving mass and the geometry of the scene. It is thus a challenge to solve suitable approximations to Einstein's equations in real-time. The computational limitations of the deskside Onyx is made apparent and remote computational assistance becomes necessary.
Level 1: (CAVE-to-CAVE) Orlando - Urbana
A connection (of modest bandwidth) is established between the CAVE at the NCSA in Urbana and the DuoDesk at SC98. Colleagues at the NCSA CAVE have access to high performance parallel computation of gravitational scene distortions. The results of the computational experiment are first reviewed at NCSA, and if correct, a compressed abstraction is sent to Orlando, and reconstructed locally. This needs to fit whatever bandwidth we can support at the moment.
Level 2: (Multi CAVE) Orlando - Urbana - Ballston - ...
We hope it will be possible for a third party to at least kibitz this scientific interaction, to participate orally, and to see "instant replays" in smooth animation.
In the near future, scientists, engineers and marketers, who are "on the road" with demanding VR displays, but with limited computational power, and uncertain network reliability, will wish to connect to home base and request concurrent (re)computation at the level of visual supercomputing. The home base will also be a VR display for the participation of colleagues not on the road. Third parties, such as funding agencies, board rooms, may wish to look into this exchange without degrading performance by requiring full participation.