TY - JOUR

T1 - An effective load balancing policy for geometric-decaying algorithms

AU - Gil, Joseph

AU - Matias, Yossi

N1 - Funding Information:
3Part of this research was done while the author was at Tel Aviv University and at the University of Maryland Institute for Advanced Computer Studies, and was partially supported by NSF Grants CCR-9111348 and CCR-8906949. E-mail: matias@research.att.com.

PY - 1996/8/1

Y1 - 1996/8/1

N2 - Parallel algorithms are often first designed as a sequence of rounds, where each round includes any number of independent constant time operations. This so-called work-time presentation is then followed by a processor scheduling implementation on a more concrete computational model. Many parallel algorithms are geometric-decaying in the sense that the sequence of work loads is upper bounded by a decreasing geometric series. A standard scheduling implementation of such algorithms consists of a repeated application of load balancing. We present a more effective, yet as simple, policy for the utilization of load balancing in geometric-decaying algorithms. By making a more careful choice of when and how often load balancing should be employed, and by using a simple amortization argument, we show that the number of required applications of load balancing should be nearly constant. The policy is not restricted to any particular model of parallel computation, and, up to a constant factor, it is the best possible.

AB - Parallel algorithms are often first designed as a sequence of rounds, where each round includes any number of independent constant time operations. This so-called work-time presentation is then followed by a processor scheduling implementation on a more concrete computational model. Many parallel algorithms are geometric-decaying in the sense that the sequence of work loads is upper bounded by a decreasing geometric series. A standard scheduling implementation of such algorithms consists of a repeated application of load balancing. We present a more effective, yet as simple, policy for the utilization of load balancing in geometric-decaying algorithms. By making a more careful choice of when and how often load balancing should be employed, and by using a simple amortization argument, we show that the number of required applications of load balancing should be nearly constant. The policy is not restricted to any particular model of parallel computation, and, up to a constant factor, it is the best possible.

UR - http://www.scopus.com/inward/record.url?scp=0030210608&partnerID=8YFLogxK

U2 - 10.1006/jpdc.1996.0098

DO - 10.1006/jpdc.1996.0098

M3 - ???researchoutput.researchoutputtypes.contributiontojournal.article???

AN - SCOPUS:0030210608

VL - 36

SP - 185

EP - 188

JO - Journal of Parallel and Distributed Computing

JF - Journal of Parallel and Distributed Computing

SN - 0743-7315

IS - 2

ER -