摘要
为提高层状地层空间数据插值的计算速度和实现系统负载平衡,在采用动态调整并行粒度策略的基础上开发混合粒度的并行Kriging插值算法.在此算法中,根据CPU与地层层数之间的数量关系动态调整并行计算的粒度,优先使用粗粒度的并行计算策略,同时利用中粒度并行算法平衡负载.基于Windows2000操作系统搭建了PC机群并行环境,大量的试验统计数据表明该方法具有良好的可扩展性和并行计算效率.
In order to improve the computing speed of spatial interpolation used for stratified stratum,and to achieve the load balancing within the different number of stratum layers and computing nodes,a novel parallel Kriging interpolation with multigranularity is developed based on the strategy of dynamic adjustment of the parallel granularity. In this algorithm, according to the quantitative relationship between the number of CPUs with this of stratum layer, the parallel granularity is adaptively adjusted, in which the coarse granularity Kriging interpolation is priority used while the medium granularity Kriging is adopted to balance the computing load. The PC cluster architure is setup using the personal computer with Windows 2000 OS, the large number of test statistics show that the method has a good scalability and a parallel computing efficiency.
出处
《小型微型计算机系统》
CSCD
北大核心
2009年第1期152-155,共4页
Journal of Chinese Computer Systems
