|
The purpose of this paper is to describe the tools and techniques in use at the NASA Ames Research Center for performing visualization of computational aerodynamics, for example visualization of flow fields from computer simulations of fluid dynamics about vehicles such as the Space Shuttle. The hardware used for visualization is a high-performance graphics workstation connected to a super computer with a high speed channel. At present, the workstation is a Silicon Graphics IRIS 3130, the supercomputer is a CRAY2, and the high speed channel is a hyperchannel. The three techniques used for visualization are post-processing, tracking, and steering. Post-processing analysis is done after the simulation. Tracking analysis is done during a simulation but is not interactive, whereas steering analysis involves modifying the simulation interactively during the simulation. Using post processing methods, a flow simulation is executed on a supercomputer and, after the simulation is complete, the results of the simulation are processed for viewing. This is by far the most commonly used method for visualization of computational aerodynamics. The next two methods are much more desirable, yet much less common given the current state of supercomputer and workstation evolution and performance. Both of these are more sophisticated methods because they involve analysis of the flow codes as they evolve. Tracking refers to a flow code producing displays that give a scientist some indication how his experiment is progressing so he could, perhaps, change some parameters and then restart it. Steering refers to actually interacting with the flow codes during execution by changing flow code parameters. (Steering methods have been employed for grid generation pre-processing as well to substantially reduce the time it takes to construct a grid for input to a flow solver). When the results of the simulation are processed for viewing by distributing the process between the workstation and the supercomputer, it is called distributed processing. This paper describes the software in use and under development at NASA Ames Research Center for performing these types of tasks in computational aerodynamics. Workstation performance issues, benchmarking, and high-performance networks for this purpose are also discussed as well as descriptions of other hardware for digital video and film recording. A new software environment, FAST, is introduced that is currently being developed at NASA Ames for implementation on workstations that will be procured in the latter half of 1989. This modular software environment will take advantage of the multiple processor and large memory configurations and other features as specified in the NASA RFP for these workstations and is a natural evolution of the techniques described in this paper.
|
