| 通用图形处理器GPGPU的并行计算研究 |
| |
| 引用本文： | 张鹏博,郭兵,黄义纯,曹亚波.通用图形处理器GPGPU的并行计算研究[J].单片机与嵌入式系统应用,2017,17(8). |
| |
| 作者姓名： | 张鹏博 郭兵 黄义纯 曹亚波 |
| |
| 作者单位： | 四川大学计算机学院,成都,610065 |
| |
| 基金项目： | 国家自然科学基金重点项目，国家自然科学基金资助项目 |
| |
| 摘 要： | 随着图形处理器(GPU)从仅用来进行图形图像渲染,脱离成为并行计算平台通用图形处理器(GPGPU),其计算能力越来越强,本文在研究GPGPU体系结构的基础上对GPGPU并行计算线程调度进行深入研究,阐述了GPU线程调度原理,揭示了SIMT调度模式的不足.通过公式推导阐述了系统功耗与系统运行频率的关系.
|
| 关 键 词： | 大数据 并行计算 线程调度 GPU节能 |
Research on Parallel Computing of General Purpose CPGPU |
| |
| Zhang Pengbo,Guo Bing,Huang Yichun,Cao Yabo.Research on Parallel Computing of General Purpose CPGPU[J].Microcontrollers & Embedded Systems,2017,17(8). |
| |
| Authors: | Zhang Pengbo Guo Bing Huang Yichun Cao Yabo |
| |
| Abstract: | The Graphics Processing Unit (GPU) only is been used to the graphics image rendering,so it becomes a parallel computing platform General Purpose GPU(GPGPU),the parallel computing power is growing,and the application is more and more widely.Based on the study of GPGPU architecture,the GPGPU parallel computing thread scheduling is studies,and the principle of GPU thread scheduling and the deficiency of SIMT scheduling mode are introduced.The relationship between system power consumption and system operating frequency is expounded by formula. |
| |
| Keywords: | big data parallel computing thread scheduling GPU energy saving |
| 本文献已被 万方数据 等数据库收录！ |
|
