太湖之光上基于神威OpenACC的GTC-P移植与优化研究.PDF

太湖之光上基于神威 OpenACC 的 GTC-P 移植与优化研究 1 1,2 1 3 3 3 王一超 林新华 蔡林金 William Tang Stephane Ethier Bei Wang 1,4 2 Simon See 松岗聪 1 上海交通大学 高性能计算中心，上海 200240 2 东京工业大学，日本 3 普林斯顿大学 等离子体物理实验室，美国 4 NVIDIA ，新加坡 Porting and Optimizing GTC-P on TaihuLight Supercomputer with Sunway OpenACC 1 1,2 1 3 3 3 1,4 2 Yichao Wang , James Lin , Linjin Cai , William Tang , Stephane Ethier , Bei Wang , Simon See , Satoshi Matsuoka 1 (Center for High Performance Computing, Shanghai Jiao Tong University, Shanghai 200240, China) 2 (Tokyo Institute of Technology, Japan) 3 (Princeton Plasma Physics Lab, Princeton University, USA) 4 (NVIDIA, Singapore) Corresponding author: James Lin, E-mail: james@, Abstract Sunway TaihuLight with its sustainable performance achieving 93PFlops is now the No.1 supercomputer in the latest Top500 list. It provides a high-level directive language called Sunway OpenACC that is compatible with OpenACC 2.0 standard with some customized extensions to help developers port OpenACC applications to this system. GTC-P is a discovery-science-capable real-world application code based on the Particle-In-Cell (PIC) algorithm that is well-established in the HPC area. The motivation of this paper is to evaluate Sunway OpenACC by porting GTC-P on TaihuLight with the associated objective of doing so with Sunway OpenACC and then comparing the differences between Sunway OpenACC and standard OpenACC 2.0. Since the Sunway OpenACC compiler cannot at present deal with the performance bottleneck of GTC-P when it is directly ported onto TaihuLight, we have applied three optimizations on an intermediate version of t