The Community for Technology Leaders
Green Image
Issue No. 10 - October (2005 vol. 16)
ISSN: 1045-9219
pp: 956-965
Ali Pinar , IEEE Computer Society
<p><b>Abstract</b>—In many applications of parallel computing, distribution of the data unambiguously implies distribution of work among processors. But, there are exceptions where some tasks can be assigned to one of several processors without altering the total volume of communication. In this paper, we study the problem of exploiting this flexibility in assignment of tasks to improve load balance. We first model the problem in terms of network flow and use combinatorial techniques for its solution. Our parametric search algorithms use maximum flow algorithms for probing on a candidate optimal solution value. We describe two algorithms to solve the assignment problem with <tmath>\log W_T</tmath> and <tmath>|P|</tmath> probe calls, where <tmath>W_T</tmath> and <tmath>|P|</tmath>, respectively, denote the total workload and number of processors. We also define augmenting paths and cuts for this problem, and show that any algorithm based on augmenting paths can be used to find an optimal solution for the task assignment problem. We then consider a continuous version of the problem and formulate it as a linearly constrained optimization problem, i.e., <tmath> \min \|Ax\|_\infty, {\rm {s.t.}} Bx=d</tmath>. To avoid solving an intractable <tmath>\infty{\hbox{-}}{\rm{norm}}</tmath> optimization problem, we show that, in this case, minimizing the <tmath>2{\hbox{-}}{\rm{norm}}</tmath> is sufficient to minimize the <tmath>\infty{\hbox{-}}{\rm{norm}}</tmath>, which reduces the problem to the well-studied linearly constrained least squares problem. The continuous version of the problem has the advantage of being easily amenable to parallelization. Our experiments with molecular dynamics and overlapped domain decomposition applications proved the effectiveness of our methods with significant improvements in load balance. We also discuss how our techniques can be extended to heterogeneous parallel computers.</p>
Parallel computing, load balancing, flexibly assignable tasks, maximum flow, constrained least squares.

A. Pinar and B. Hendrickson, "Improving Load Balance with Flexibly Assignable Tasks," in IEEE Transactions on Parallel & Distributed Systems, vol. 16, no. , pp. 956-965, 2005.
91 ms
(Ver 3.3 (11022016))