| 引用本文: | 姜春雷,张树清.CPU-GPU协同加速Kriging插值的负载均衡方法.[J].国防科技大学学报,2015,37(5):35-39,.[点击复制] |
| JIANG Chunlei,ZHANG Shuqing.A load balancing method in accelerating Kriging algorithm on CPU-GPU heterogeneous platforms[J].Journal of National University of Defense Technology,2015,37(5):35-39,[点击复制] |
|
| |
|
|
| 本文已被：浏览 9109次 下载 6872次 |
| CPU-GPU协同加速Kriging插值的负载均衡方法 |
| 姜春雷1,2, 张树清1 |
|
(1.中国科学院 东北地理与农业生态研究所， 吉林 长春 130102;2.中国科学院大学， 北京 100049)
|
| 摘要: |
| Kriging插值算法被广泛应用于地学各领域，有着极其重要的现实意义，但在面对大规模输出网格及大量输入采样点时，不可避免地遇到了性能瓶颈。利用OpenCL和OpenMP在异构平台上实现了CPU与GPU协同加速普通Kriging插值。针对Kriging插值中采样点的不规则分布及CPU和GPU由于体系结构差异对其的不同适应性，提出一种基于不同设备间计算性能的差异和数据分布特点的负载均衡方法。试验结果表明，该方法能有效提高普通Kriging插值速度，同时还能节约存储空间和提高访存效率。 |
| 关键词: 通用计算图形处理器 开放运算语言 Kriging插值 负载均衡 |
| DOI：10.11887/j.cn.201505006 |
| 投稿日期：2015-06-24 |
| 基金项目:国家自然科学基金资助项目（41271196）；中国科学院重点部署资助项目（KZZD-EW-07-02） |
|
| A load balancing method in accelerating Kriging algorithm on CPU-GPU heterogeneous platforms |
| JIANG Chunlei1,2, ZHANG Shuqing1 |
|
(1.Northeast Institute of Geography and Agroecology, Chinese Academy of Sciences, Changchun 130102, China;2.
2. University of Chinese Academy of Sciences, Beijing 100049, China)
|
| Abstract: |
| Kriging interpolation algorithm is of great practical significance and is widely applied to various fields of geoscience. However, Kriging interpolation would inevitably encounter the performance bottleneck when the output grid or input samples increase. Implemented with OpenCL and OpenMP, the ordinary Kriging interpolation was accelerated on heterogeneous platforms: GPU and CPU. By considering the performance difference of CPU and GPU on the densities of samples, a new load balancing method of LBCPDD (Load Balancing based on Computation Performance and Data Distribution) was proposed, in which not only hardware performance but also data distribution characteristics were taken into account. Experiment results show that LBCPDD method can effectively enhance the speed of ordinary Kriging, save memory space and improve the efficiency of memory access. |
| Keywords: general purpose graphics processor units open computing language Kriging interpolation load balancing |
|
|
