基于负载均衡的空间线分组算法

The Algorithm of Spatial Line Grouping Based on Load Balancing

针对传统并行操作计算效率低的问题,提出以分组并行处理模式优化节点间的负载均衡。以表层浮标轨迹验证涡旋实验为例,给出面向不可分割空间线对象的快速分组方法,设计了两个分组调整算法。实验结果显示,算法可以使每个计算节点达到负载均衡。与串行计算的比较实验结果显示,算法具有较好的加速效果,且加速比随着计算节点个数的增加呈上升趋势。因此,基于负载均衡的空间线分组算法是对不可分割空间线的计算进行优化的有效途径。

To solve tbe problem of low computation efficiency in traditional parallel task operation, this paper pro- posed a grouping parallel processing mode for optimizing load balancing among computation nodes. The quick grou- ping algorithm for grouping the indivisible spatial line objects was applied to the vortex validation experiment, which is implemented by tracking surface buoy. Moreover, two group adaptation algorithms were given. The algorithm proposed in this paper balanced the load of each computation node and achieved perfect speedup. The speedup ratio increased with the numbers of computation nodes. The experimental results show that this spatial line grouping al- gorithm is an effective approach for optimizing the computation of indivisible spatial linear objects.