| CPU-GPU并行矩阵乘法的实现与性能分析 |
| |
| 引用本文： | 程豪,张云泉,张先轶,李玉成.CPU-GPU并行矩阵乘法的实现与性能分析[J].计算机工程,2010,36(13):24-26,29. |
| |
| 作者姓名： | 程豪 张云泉 张先轶 李玉成 |
| |
| 作者单位： | 1. 中国科学院软件研究所并行计算实验室,北京100190;中国科学院研究生院,北京100190
2. 中国科学院软件研究所并行计算实验室,北京,100190 |
| |
| 基金项目： | 国家自然科学基金资助重点项目，国家"863"计划基金资助项目，国家重大科研装备研制项目 |
| |
| 摘 要： | 实现ATI平台上的CPU-GPU混合并行DGEMM，采用在GPU和CPU上同时进行计算的方法来提高运算性能。实验结果证明，当矩阵规模较大时，在AMD Phenom II X4 940和ATI FireStream 9270平台上，混合DGEMM性能相对于单独使用GPU平均提升了16%。通过实验验证了混合DGEMM性能、加速比、任务分配比例的估算方法，并探讨了影响混合DGEMM性能的各种因素。
|
| 关 键 词： | 混合并行 GPU技术 DGEMM程序 加速比 |
Implementation and Performance Analysis of CPU-GPU Parallel Matrix Multiplication |
| |
| CHENG Hao,ZHANG Yun-quan,ZHANG Xian-yi,LI Yu-cheng.Implementation and Performance Analysis of CPU-GPU Parallel Matrix Multiplication[J].Computer Engineering,2010,36(13):24-26,29. |
| |
| Authors: | CHENG Hao ZHANG Yun-quan ZHANG Xian-yi LI Yu-cheng |
| |
| Affiliation: | (1. Lab of Parallel Computing, Institute of Software, Chinese Academy of Sciences, Beijing 100190; 2. Graduate University of Chinese Academy of Sciences, Beijing 100190) |
| |
| Abstract: | The implementation of the CPU-GPU hybrid DGEMM is carried out on the ATI platform to improve the computing performance by computing both on GPU and CPU. Experimental results show that when matrix size is large, its performance on AMD Phenom II X4 940 and ATI FireStream 9270 platform, compared with using GPU alone, can be improved 16% on average. The evaluation method is verified along with the discussion of the factors which impact the hybrid DGEMM performance. |
| |
| Keywords: | hybrid parallelization GPU DGEMM speedup ratio |
| 本文献已被 维普 万方数据 等数据库收录！ |
| 点击此处可从《计算机工程》浏览原始摘要信息 |
|
点击此处可从《计算机工程》下载全文 |
|
