Scheduling DAGs on asynchronous processors

• Authors: Michael A. Bender;Cynthia A. Phillips
• Venue: Proceedings of the nineteenth annual ACM symposium on Parallel algorithms and architectures
• Year: 2007

This paper addresses the problem of scheduling a DAG of unit-length tasks on asynchronous processors, that is, processors having different and changing speeds. The objective is to minimize the makespan, that is, the time to execute the entire DAG. Asynchrony is modeled by an oblivious adversary, which is assumed to determine the processor speeds at each point in time. The oblivious adversary may change processor speeds arbitrarily and arbitrarily often, but makes speed decisions independently of any random choices of the scheduling algorithm. This paper gives bounds on the makespan of two randomized online firing-squad scheduling algorithms, All and Level. These two schedulers are shown to have good makespan even when asynchrony is arbitrarily extreme. Let W and D denote, respectively, the number of tasks and the longest path in the DAG, and let πave denote the average speed of the p processors during the execution. In All each processor repeatedly chooses a random task to execute from among all ready tasks (tasks whose predecessors have been executed). Scheduler All is shown to have a makespan Tp= Θ(Wpπave), when WD ≥ p log p Θ((log p)α Wpπave + (log p) 1-α Dπave), when WD= p(log p)1-2α, for α ∈ [0, 1] Θ (Dπave, when WD ≤ plog p, both expected and with high probability. A family of DAGs is exhibited for which this analysis is tight. In Level each of the processors repeatedly chooses a random task to execute from among all critical tasks (ready tasks at the lowest level of the DAG). This second scheduler is shown to have a makespan of.