Optimizing the LINPACK Algorithm for Large-Scale PCIe-Based CPU-GPU Heterogeneous Systems
文献类型：期刊论文
| 作者 | Tan, Guangming; Shui, Chaoyang; Wang, Yinshan; Yu, Xianzhi; Yan, Yujin |
| 刊名 | IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS
 |
| 出版日期 | 2021-09-01 |
| 卷号 | 32期号:9页码:2367-2380 |
| 关键词 | Pipeline processing Graphics processing units Computer architecture Supercomputers Clustering algorithms Programming Optimization LINPACK algorithm software pipeline performance model heterogeneous computing cluster |
| ISSN号 | 1045-9219 |
| DOI | 10.1109/TPDS.2021.3067731 |
| 英文摘要 | There is a widening gap between GPU and other components (CPU, PCIe bus and communication network) in heterogeneous parallel system. The gap forces us to orchestrate cooperative execution among these components much more carefully than ever before. By taking the LINPACK benchmark as a case study, this article proposes a fine-grained pipelining algorithm on large-scale CPU-GPU heterogeneous cluster systems. First, we build an algorithmic model that reveals a new approach to GPU-centric and fine-grained pipelining algorithm design. Then, we present four model-driven pipelining algorithms that incrementally squeeze bubbles in the pipeline so that it is occupied by more useful floating-point calculations. The algorithms are implemented on both the AMD and NVIDIA GPU platforms. The finally optimized LINPACK program achieves 107 PFlops on 25, 600 GPUs (70 percent floating-point efficiency). Several insights have been drawn to suggest tradeoff of algorithm design, programming support, and architecture design. |
| 资助项目 | National Key Research and Development Program of China[2018YFB0204400] ; National Key Research and Development Program of China[XDC01030000] ; National Key Research and Development Program of China[XDC05010100] ; National Natural Science Foundation of China[62032023] ; National Natural Science Foundation of China[61972377] ; National Natural Science Foundation of China[61702483] ; CAS[QYZDJ-SSW-JSC035] |
| WOS研究方向 | Computer Science ; Engineering |
| 语种 | 英语 |
| WOS记录号 | WOS:000638398900004 |
| 出版者 | IEEE COMPUTER SOC |
| 源URL | [http://119.78.100.204/handle/2XEOYT63/16647]  |
| 专题 | 中国科学院计算技术研究所期刊论文_英文 |
| 通讯作者 | Shui, Chaoyang |
| 作者单位 | Chinese Acad Sci, Inst Comp Technol, State Key Lab Comp Architecture, Beijing 100864, Peoples R China |
| 推荐引用方式 GB/T 7714 | Tan, Guangming,Shui, Chaoyang,Wang, Yinshan,et al. Optimizing the LINPACK Algorithm for Large-Scale PCIe-Based CPU-GPU Heterogeneous Systems[J]. IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS,2021,32(9):2367-2380. |
| APA | Tan, Guangming,Shui, Chaoyang,Wang, Yinshan,Yu, Xianzhi,&Yan, Yujin.(2021).Optimizing the LINPACK Algorithm for Large-Scale PCIe-Based CPU-GPU Heterogeneous Systems.IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS,32(9),2367-2380. |
| MLA | Tan, Guangming,et al."Optimizing the LINPACK Algorithm for Large-Scale PCIe-Based CPU-GPU Heterogeneous Systems".IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS 32.9(2021):2367-2380. |
入库方式： OAI收割
来源：计算技术研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。