摘要
对科学计算问题中的节能减排技术进行研究,在动态电压调节技术的基础上,提出一种节能的并行计算设计模型。该模型在保证整个并行任务总的执行时间不增加的基础上,充分适当延长非关键路径上任务的执行时间,从而降低处理器的功耗,达到节能的目的。采用高斯约当算法进行评测,结果证明,矩阵的维数越大,产生的任务越多,该并行设计模型的节能效果越明显。最高可节省38.4%。
The computation based on energy saving and emission reduction is the key technology of the green economy development in the future and will become the mainstream of future economic development. In this paper we carry out the study aiming at the technologies of energy saving and emission reduction in scientific computing, and present a design model for energy saving parallel algorithm based on dynamic voltage scaling (DVS) technology. Based on ensuring no extra total executing time in entire parallel tasks, the model fully and properly extends the execution time of tasks in non-crucial path so as to reduce the energy consumption in processor and achieve the goal of energy saving. Gauss Jordan algorithm is employed in evaluation and test, results show that the larger the matrix dimensionality, the more the tasks, and the more evident the energy saving effect of the parallel design model. The maximum savings can reach 38.4%.
出处
《计算机应用与软件》
CSCD
北大核心
2014年第7期18-20,27,共4页
Computer Applications and Software
基金
江苏省高校哲学社会科学研究重点项目(2010ZDIXM022)
关键词
绿色计算
科学计算
节能算法
分布式计算
高斯约当算法
Green computing
Scientific computing
Energy saving algorithm
Distributed computing
Gauss Jordan algorithm