M. Frumkin and D. Riley (USA)
dynamic scheduling, performance, naviga tion, trajectory optimization.
Planning complex missions requires a number of pro grams to be executed in concert. The Application Navi gation System (ANS), developed in the NAS Division, can execute many interdependent programs in a distributed en vironment. We show that the ANS simplifies user effort and reduces time in optimization of the trajectory of a martian airplane. We use a software package, Cart3D, to evalu ate trajectories and a wave front algorithm to determine the optimal trajectory. ANS employs the "GridScape" to repre sent the dynamic state of the available computer resources. Then, ANS uses a scheduler to dynamically assign ready tasks to machine resources and the GridScape for tracking available resources and forecasting completion time of run ning tasks. We demonstrate system capability to schedule and run the trajectory optimization application with effi ciency exceeding 60% on 64 processors.
Important Links:
Go Back
