Scheduling under dynamic speed-scaling for minimizing weighted completion time and energy consumption

Abstract

International audienceSince a few years there is an increasing interest in minimizing the energy consumption of computing systems. However in a shared computing system, users want to optimize their experienced quality of service, at the price of a high energy consumption. In this work, we address the problem of optimizing and designing mechanisms for a linear combination of weighted completion time and energy consumption on a single machine with dynamic speed-scaling. We show that minimizing linear combination reduces to a unit speed scheduling problem under a polynomial penalty function. In the mechanism design setting, we define a cost share mechanism and study its properties, showing that it is truthful and the overcharging of total cost share is bounded by a constant

Power‐aware speed scaling in multiprocessor systems

“…(Using (17) and Lemma 4) Case III: When At-ALG is (dΦ p /dt) = (dΦ pO /dt) + (dΦ pA /dt). Using (9) and (11), it gives…”

“…A number of results [4,[7][8][9][10] for online clairvoyant scheduling problems are obtained for the objective of minimising the flow time plus energy. The researchers have good contribution [2,[11][12][13] for the clairvoyant job scheduling problems where the objective is to minimise the weighted/magnitude-based flow time plus energy. The scheduling problems under the nonclairvoyant setting for the objective of minimising the flow time plus energy are studied by some researchers [14][15][16][17].…”

Mechanism Design for Aggregating Energy Consumption and Quality of Service in Speed Scaling Scheduling

“…From the previous section, we know that the game admits a pure Nash equilibrium, and by symmetry of the players, in fact two pure Nash equilibria. Following [10], the first one is…”

“…To break the symmetry we consider only points where the social cost of S21 is minimal among the two profiles. In [10] we showed that these are precisely the points that satisfy For such points, we ask whether one of the players wants to deviate from S12, i.e., the other potential equilibrium, with larger social cost. Our experiments indicate that, for each player j ∈ {1, 2}, there is a threshold t j , which depends on p 2 , such that player j wants to deviate if and only if w 2 ≤ t j .…”

“…In a companion paper [10] we study this game from the point of view of the game regulator, in a different setting. The players announce with their job their priority factors, and the regulator gets to decide on the completion time of the jobs.…”