A scheduling model for reduced CPU energy
FOCS '95 Proceedings of the 36th Annual Symposium on Foundations of Computer Science
Speed scaling for weighted flow time
SODA '07 Proceedings of the eighteenth annual ACM-SIAM symposium on Discrete algorithms
Energy-efficient algorithms for flow time minimization
ACM Transactions on Algorithms (TALG)
Speed scaling with an arbitrary power function
SODA '09 Proceedings of the twentieth Annual ACM-SIAM Symposium on Discrete Algorithms
Optimal speed scaling under arbitrary power functions
ACM SIGMETRICS Performance Evaluation Review
Communications of the ACM
Scalably scheduling power-heterogeneous processors
ICALP'10 Proceedings of the 37th international colloquium conference on Automata, languages and programming
Competitive algorithms for due date scheduling
ICALP'07 Proceedings of the 34th international conference on Automata, Languages and Programming
Hi-index | 0.00 |
This paper considers several online scheduling problems that arise from companies with made-to-order products. Jobs, which are product requests, arrive online with different sizes and weights. A company needs to assign a due date for each job once it arrives, and complete the job by this due date. The (weighted) quoted lead time of a job equals its due date minus its arrival time, multiplied by its weight. We focus on companies that mainly rely on computers for production. In those companies, energy cost is a large concern. For most modern processors, its rate of energy usage equals sa, where s is the current speed and α 1 is a constant. Hence, reducing the processing speed can reduce the rate of energy usage. Algorithms are needed to optimize the (weighted) quoted lead time (for better user experience) and the energy usage (for a smaller energy cost). We propose an algorithm which is 4((log k)α-1 + α/α-1)-competitive for minimizing the sum of the quoted lead time and energy usage, where k is the ratio between the maximum to minimum job density. Here, the density of a job equals its weight divided by its size. We also consider the setting where we may discard a job by paying a penalty, and the setting of scheduling on a multiprocessor. We propose competitive algorithms for both settings.