Some observations on optimal frequency selection in DVFS–based energy consumption minimization

Select |




Print


Babaii Rizvandi, Nikzad; Taheri, Javid; Zomaya, Albert


2011-08-03


Journal Article


Journal of Parallel and Distributed Computing


71


8


1154-1164


In the recent years, the issue of energy consumption in multiprocessor computer systems has attracted a great amount of attention. In response to this, many energy-aware scheduling algorithms have been developed primarily using the dynamic voltage-frequency scaling (DVFS) capability incorporated with recent commodity processors. Majority of these algorithms involve two passes: scheduling generation and slack reclamation. The former is to redistribute tasks among DVFS-enabled processors based on new cost function including makespan and power consumption; and, the latter is typically achieved by executing individual tasks with slacks in lower processor frequency. In this paper, a new slack reclamation algorithm is proposed by revising the energy reduction technique from a different perspective. Toward this end, firstly, the problem of a task slack reclamation using multiple selections of the processors’ frequencies is formulated. Secondly, several proofs are provided to show that (1) if working frequency set of processor is assumed to be continues, the optimal energy will always achieved using only one frequency, (2) for real processors with discrete set of working frequencies, the optimal energy is always achieved using two frequencies, and (3) these two frequencies are adjacent/neighbour when processor power consumption is a convex function of frequency. Thirdly, a novel algorithm to find best combination of frequencies to result the optimal energy is presented. The presented algorithm has been evaluated based on results obtained from experiments with three different sets of task graphs: 1,500 randomly generated task graphs, and 300 task graphs for of each of two real-world applications (Gauss-Jordan and LU decomposition). Results showed superiority of the proposed algorithm in comparison with other techniques.


Energy consumption, DVFS, high performance computing


https://doi.org/10.1016/j.jpdc.2011.01.004


nicta:4365


Babaii Rizvandi, Nikzad; Taheri, Javid; Zomaya, Albert. Some observations on optimal frequency selection in DVFS–based energy consumption minimization. Journal of Parallel and Distributed Computing. 2011-08-03; 71(8):1154-1164. https://doi.org/10.1016/j.jpdc.2011.01.004



Loading citation data...

Citation counts
(Requires subscription to view)