Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network

  Advanced Search  

Task Reweighting on Multiprocessors: Efficiency versus Aaron Block and James H. Anderson

Summary: Task Reweighting on Multiprocessors: Efficiency versus
Aaron Block and James H. Anderson
Department of Computer Science
University of North Carolina at Chapel Hill
We consider the problem of task reweighting in fair-scheduled multiprocessor systems wherein each task's
processor share is specified using a weight. When a task is reweighted, a new weight is computed for it
that is then used in future scheduling. Task reweighting can be used as a means for consuming (or making
available) spare processing capacity. The responsiveness of a reweighting scheme can be assessed by comparing
its allocations to those of an ideal scheduler that can reweight tasks instantaneously. A reweighting scheme is
fine-grained if any additional per-task "error" (in comparison to an ideal allocation) caused by a reweighting
event is constant. Similarly, a reweighting scheme is coarse-grained if the additional "error" per reweighting
event is non-constant. While fine-grained reweighting produces only a small amount of error when changing
task weights, it has a worst-case time complexity of (NlogN), where N is the number of tasks. If the number
of tasks exceeds the number of processors, then such time complexity is larger than that of coarse-grained
reweighting, which is (MlogN), where M is the number of processors. In this paper, we construct two new
reweighting algorithms that are hybrids of fine- and coarse-grained reweighting that have time complexity
less than (NlogN) and produce less error than current coarse-grained techniques. We also present an
experimental evaluation of these scheme to compare their relative advantages.


Source: Anderson, James - Department of Computer Science, University of North Carolina at Chapel Hill


Collections: Computer Technologies and Information Sciences