Titlebook: Job Scheduling Strategies for Parallel Processing; IPDPS 2000 Workshop, Dror G. Feitelson,Larry Rudolph Conference proceedings 2000 Springe

MUMP

System Utilization Benchmark on the Cray T3E and IBM SP,alled the E.ective System Performance (ESP) test, designed to provide a utilization metric that is transferable between systems and illuminate the e.ects of various scheduling parameters. Results with discussion are presented for the Cray T3E and IBM SP systems together with insights obtained from s

LAIR

中和

脆弱帶來

Melatonin

用不完

Overthrow

Load Balancing for Minimizing Execution Time of a Target Job on a Network of Heterogeneous Workstatate and size are “random? on a NOWs. In such an environment, partitioning (load balancing) a target job based on only the .rst order moments (means) of system parameters is not optimal. In this paper, it is proposed to consider the second order moments (standard deviations) also in load balancing in

Grasping

Adaptive Selection of Partition Size for Supercomputer Requests,A, an AppLeS application scheduler, to adaptively craft the request tobe submitted to the supercomputer based on the current state of the system. Wedemonstrate that SA significantly improves a job’s turn-around time in a varietyof scenarios. We also identify how the state of the system, the characte

失誤

Adaptive Selection of Partition Size for Supercomputer Requests,monstrate that SA significantly improves a job’s turn-around time in a varietyof scenarios. We also identify how the state of the system, the characteristicsof the job, and the quality of the information made available to SA influenceits performance.

ETHER