This project plans to investigate how to best use multiple workstations to support modern, highly interactive applications. The initial focus is on the real-time rendering of 3D computer generated scenes. These applications present two key new problems in the use of multi-computers: they stress resources in the machines (e.g., memory bandwidth and outboard video processing capability) other than just processor speed and network bandwidth, and "good performance" means meeting the short time deadline by which the next frame must be produced rather than minimizing the duration of a fairly long execution. The focus of this project is on resource allocation: how to divide system-wide resources among the components of the problem that must be done so that performance objectives can be met. The current effort is aimed at solving this problem in an environment where both the work to be performed and the total resources that are available are relatively constant from frame to frame. A longer term goal of the work is to investigate the use of replication to help assure that deadlines will continue to be met in more dynamic environments.

Project Start
Project End
Budget Start
1997-08-01
Budget End
2001-07-31
Support Year
Fiscal Year
1997
Total Cost
$229,431
Indirect Cost
Name
University of Washington
Department
Type
DUNS #
City
Seattle
State
WA
Country
United States
Zip Code
98195