An Optimization Strategy for Improving Throughput of GPU Global Memory

作者：Wang, Yanhua^*; Qiao, Jianzhong; Lin, Shukuan; Zhao, Tinglei

来源：Journal of Grey System, 2018, 30(2): 42-56.

• Grey Clustering
• Grey Target Decision-making
• Cobweb Area
• Global Memory
• Congestion
• GPU

Multiple global memory access may lead to serious bottlenecks in GPU (Graphic Processing Unit) kernels. Global memory access congestion brings low throughput as well as bad performance. In the paper, the crucial characteristics of global memory access are analysed. Then a global memory access congestion judging model based on grey clustering is proposed, which can make classification for the congestion degree of global memory access. After analyzing the congestion objects and choosing the access data, optimization is carried out by a grey target decision model based on cobweb area. So the congestion is relieved. The proposed model is evaluated with several benchmarks on NVIDIA GTX 750. Comparing with the original kernels, experimental results demonstrate that the model can achieve 11.09% improvement of global memory throughput averagely.

• 出版日期2018
• 单位东北大学