Performance Optimization of the HPCG Benchmark on the Sunway TaihuLight Supercomputer
Yulong Ao, Chao Yang, Fangfang Liu, Wanwang Yin, Lijuan Jiang, Qiao Sun · ACM Transactions on Architecture and Code Optimization · 2018
In this article, we present some key techniques for optimizing HPCG on Sunway TaihuLight and demonstrate how to achieve high performance in memory-bound applications by exploiting specific characteristics of the hardware architecture. In particular, we utilize a block multicoloring approach for parallelization and propose methods such as requirement-based data mapping and customized gather collective to enhance the effective memory bandwidth. Experiments indicate that the optimized HPCG code can sustain 77% of the theoretical memory bandwidth and scale to the full system of more than 10 million cores, with an aggregated performance of 480.8 Tflop/s and a weak scaling efficiency of 87.3%.