| Title |
A Survey on High Performance Weather Forecasting System I/O Performance Measurement Tool |
| Authors |
최수혁(Soo-hyuck Choe) ; 정은성(Eun-Sung Jung) |
| DOI |
https://doi.org/10.5573/ieie.2022.59.4.47 |
| Keywords |
GloSea6; Weather forecasting system; Benchmarking/profiling; Survey |
| Abstract |
With the proliferation of computers, the era of using numerical forecasts for weather forecasts has arrived. With the advent of powerful computing resources, the complexity of numerical forecasting models continues to increase. At the same time, to improve the model's accuracy, available computer resources must be increased, and for this, research on optimization of CPU, GPU, and Storage I/O is indispensable. Therefore, in this paper, for the study of the numerical prediction model, Low GloSea6, which is a low-resolution weather prediction model that can operate even in small servers, is examined. Next, we do an in-depth investigation and analysis of MADbench2 (a benchmarking tool) and Darshan, CrayPat, NVIDIA NVXT, XIOS, Vidya, and Apollo (tools to identify the I/O behavior of scientific applications). In addition, the latest programming and performance visualization tools such as TFProf are investigated and analyzed. Based on this, we discuss applying benchmarking/profiling tools to future research on improving I/O performance of Low GloSea6.. |