CAJ | 학술논문

根据ANSI/ANS-3.5-1998规定以及核电厂建模精度的提高，对核电模拟机仿真速度提出了更高的要求。但是目前已难以通过提升中央处理器（ CPU）频率的方式来提升现有模拟机的运算速度。与此同时，CPU/GPU异构计算融合了串行/并行计算，利用显卡（ GPU）的并行计算能力可极大提升现有桌面电脑的运算能力，目前已经广泛应用于科学研究。英伟达公司的CUDA平台被用于开发CPU/GPU异构计算应用程序，来提升核电厂全范围模拟机的仿真计算。通过核电厂全范围模拟机运行测试对比，证实使用CPU/GPU异构计算程序，能有效提升模拟机运行速度。
근거ANSI/ANS-3.5-1998규정이급핵전엄건모정도적제고，대핵전모의궤방진속도제출료경고적요구。단시목전이난이통과제승중앙처리기（ CPU）빈솔적방식래제승현유모의궤적운산속도。여차동시，CPU/GPU이구계산융합료천행/병행계산，이용현잡（ GPU）적병행계산능력가겁대제승현유탁면전뇌적운산능력，목전이경엄범응용우과학연구。영위체공사적CUDA평태피용우개발CPU/GPU이구계산응용정서，래제승핵전엄전범위모의궤적방진계산。통과핵전엄전범위모의궤운행측시대비，증실사용CPU/GPU이구계산정서，능유효제승모의궤운행속도。
According to the ANSI/ANS-3. 5-1998 regulations and the improvement of the nuclear power plant modeling accuracy, the nuclear power plant Full Scope Simulator ( FSS) is required to realize real-time simulation. However, it is hard to speed up the simulation by increasing the CPU frequency. Compared with traditional serial computing, CPU/GPU heterogeneous computing, as the combination of serial computing and parallel computing, will enable dramatic increases in computing performance of the present desktop computers by harnessing the parallel computing power of the Graphic Processing Unit ( GPU) . Nowadays, CPU/GPU heterogeneous computing is widely applied in scientific research. In this paper, CUDA, the parallel computing platform developed by NVIDIA corporation, was used for CPU/GPU heterogeneous computing program, which was applied to simulation computing of the nuclear power plant FSS. Compared with the original FSS, the test results verify that it is a feasible way to speed up FSS by CPU/GPU heterogeneous computing.