Research is proposed to investigate the efficacy of systolic compression/expansion (SEC) load balancing for scalable concurrent visualization systems (SCVS). The approach used is based on a Newtonian derivation for the equation of motion describing the message-passing flow between neighboring multicomputer computation elements. An assessment of graphical primitive quantum size representation for geometry, silicon process technology, and message-passing rates will lead to an effective metric for predicting the instantaneous load-balance, and systolically managing the update of SCVS display subsystem. The SCVS will be simulated via a physical concurrency emulation of the logically concurrent process structure required to implement a message-passing system for rendering graphical datasets in a shared-memory. This simulation will serve as the foundation for load balance monitoring studies based on a parametric test matrix of input conditions, such as user transformation rate, and geometric quanta size representation.

Agency
National Science Foundation (NSF)
Institute
Division of Industrial Innovation and Partnerships (IIP)
Type
Standard Grant (Standard)
Application #
9161259
Program Officer
Ritchie B. Coryell
Project Start
Project End
Budget Start
1992-01-01
Budget End
1992-09-30
Support Year
Fiscal Year
1991
Total Cost
$49,850
Indirect Cost
Name
Parallel Software Group
Department
Type
DUNS #
City
Santa Clara
State
CA
Country
United States
Zip Code
95054