| 基于CUDA实现MRRR算法并行 |
| |
| 引用本文： | 汪丽杰,赵永华.基于CUDA实现MRRR算法并行[J].计算机科学,2012,39(3):286-289. |
| |
| 作者姓名： | 汪丽杰 赵永华 |
| |
| 作者单位： | (中国科学院计算机网络信息中心超算中心 北京 100190) |
| |
| 基金项目： | 国家自然科学基金(60873113);国家高技术研究发展计划(863)(2010AA012301);国家重点基础研究发展规划(973)(2011CB309702)资助 |
| |
| 摘 要： | MRRR(Multiple Relatively Robust Representations)算法是求解对称三对角矩阵本征值问题高效、精确的算法之一。在分析MRRR算法及CUDA(Compute Unified Device Architecture)并行体系结构的基础上,针对算法的可并行性,采用单指令多线程并行方式实现了基于CUDA的MRRR算法并行,并从存储结构方面优化算法。实验结果显示,与LAPACK库中串行MRRR实现相比,并行方法在保证精度的基础上获得了20倍的加速比,进而从计算精度和计算时间上说明MRRR算法适合在GPU上并行。
|
| 关 键 词： | MRRR 并行 CUDA 本征问题 |
Parallel Realization of the MRRR Algorithm Based on CUDA |
| |
| WANG Li-jie ZHAO Yong-hua.Parallel Realization of the MRRR Algorithm Based on CUDA[J].Computer Science,2012,39(3):286-289. |
| |
| Authors: | WANG Li-jie ZHAO Yong-hua |
| |
| Affiliation: | WANG Li-jie ZHAO Yong-hua(Department of Supercomputing Center,Chinese Academy of Sciences,Beijing 100190,China) |
| |
| Abstract: | The algorithm of multiple relatively robust representations(MRRR) is one of the fastest and most accurate
algorithms. After analyzing the MRRR algorithm and CUDA parallel architecture, parallel MRRR algorithm based on
CUDA was given, and explored the optimization in memory structure. Compared with LAPACK's MRRR implementa-
lion this parallel method provides 20-fold speedups. This result demonstrates the algorithm can be mapped efficiently
onto GPU. |
| |
| Keywords: | MRRR Parallel CUDA Eigenproblem |
| 本文献已被 CNKI 等数据库收录！ |
|
点击此处可从《计算机科学》下载全文 |
