江门中微子实验（JUNO）做为一个多功能的中微子观测平台，其主要物理目标包括中微子质量测序和震荡参数的精确测量等。因该实验规模巨大，其离线软件面临很大计算压力。针对JUNO探测器模拟和事例重建软件进行优化，采用GPU加速技术将能显著提升软件运行速度。结合GPU计算技术现状和发展趋势，我们选择NVIDIA Tesla GPU和CUDA平台做为研发方向。通过对探测器模拟和事例重建软件运行模式的分析，并与CUDA轻线程计算结构做比较和匹配，提出合理的CUDA开发方案并预期能获取理想的加速效果。此外，我们还会将CUDA环境与JUNO离线软件平台SNiPER集成，使开发者和用户在CPU和GPU计算之间都能平滑过渡。

The Jiangmen Underground Neutrino Observatory (JUNO) is a multiple purpose neutrino experiment to determine neutrino mass hierarchy and precisely measure oscillation parameters. JUNO is a huge project, and it will occupy a great amount of computing resources. We take the GPU computing into account, and expect a significant improvement of the executing efficiency to the JUNO detector simulation and event reconstruction. According to the actuality and prospect of GPU computing technology in industry circles, we prefer the NVIDIA Tesla GPU and CUDA platform as our research direction. The detector simulation and event reconstruction procedures are generally reviewed. The main executing part can be well mapped to the CUDA lightweight threads architecture. An appropriate design with CUDA is proposed, which will provide excellent acceleration result. At the same time, we will integrate CUDA into the SNiPER software platform, so that developers and users are able to switch between CPU and GPU computing smoothly.