面向γ辐射屏蔽的图形处理器并行加速方法及点核积分框架研究

摘要/Abstract

摘要： 面向γ辐射屏蔽计算对高精度与高吞吐的并行需求,系统梳理了图形处理器(GPU)并行加速技术的关键进展。围绕确定性方法与随机方法的核心计算环节,归纳GPU在蒙特卡罗输运、离散纵标法(SN)以及点核积分中的加速要点,并提出面向点核积分的GPU实现框架,涵盖任务粒度与线程映射、共享/常量/纹理内存协同、核函数重构以及主机—设备流水化与数据驻留策略。进一步综述CPU-GPU协同、三维辐射场实时仿真与可视化闭环,以及与深度学习的代理/降噪与物理约束融合等方向的代表性进展,分析负载不均衡、统一虚拟内存页迁移、非结构化几何访存与数值一致性验证等瓶颈问题。结果表明,GPU在厚屏蔽穿透、多次散射与体素化大场景中显著压缩计算时间,并支撑近实时的三维场重建与交互式评估。

关键词: γ辐射, GPU并行, 点核积分, 屏蔽计算

Abstract: This review synthesizes recent advances in GPU (Graphics Processing Unit)——based acceleration for gamma-ray shielding and dose computation. Focusing on the computational kernels of deterministic and stochastic approaches, acceleration strategies for Monte Carlo transport, discrete ordinates (SN), and point-kernel integration are summarized. Building on these insights, we formalize a GPU-oriented framework for point-kernel methods that covers task granularity and thread mapping, cooperative use of shared/constant/texture memories, kernel refactoring, and host & device pipelining with data residency. We further survey CPU-GPU co-processing, real-time 3D radiation-field simulation with visualization feedback, and learning-based surrogates/denoisers with physics-informed constraints, identifying practical bottlenecks including load imbalance, unified memory page migration, unstructured-geometry memory access, and verification/validation of numerical consistency. Evidence indicates that GPUs can markedly shorten time-to-solution for thick-shield penetration, multiple scattering, and large voxelization, enabling field reconstruction and interactive assessment in near real-time.

Key words: gamma radiation, GPU acceleration, point-kernel integration, shielding calculation

中图分类号:

TL72

于辰悦, 张洪健, 张立国. 面向γ辐射屏蔽的图形处理器并行加速方法及点核积分框架研究[J]. 辐射防护通讯, 2025, 45(6): 14-20.

YU Chenyue, ZHANG Hongjian, ZHANG Liguo. GPU-accelerated methods for gamma-ray shielding and a point-kernel integration framework[J]. RADIATION PROTECTION BULLETIN, 2025, 45(6): 14-20.

参考文献

[1] CHENG Peng, QI Shuang, GUAN Xian,et al. Forward calculation of airborne gamma 3D radiation fields based on rapid coupling method of point kernel integrals[J]. Journal of Environmental Radioactivity, 2025, 282: 107603.
[2] XU Zhihui, LI Mengkun, ZOU Bowen, et al. A GPU-based point kernel gamma dose rate computing code for virtual simulation in radiation-controlled area[J]. Nuclear Engineering and Technology, 2023, 55(6): 1966-1973.
[3] 武祯,路伟,鄢书畅,等.GPU加速光电耦合输运蒙卡程序研发及应用[J].哈尔滨工程大学学报,2022, 43(11): 1649-1656+1663.
[4] Michalakes J, Vachharajani M. GPU acceleration of numerical weather prediction[J]. Parallel Processing Letters, 2008, 18(4): 531-548.
[5] Bell N, Garland M. Implementing sparse matrix-vector multiplication on throughput-oriented processors[C]//Proceedings of the Conference on High Performance Computing Networking, Storage and Analysis. Portland: IEEE, 2009: 1-11.
[6] Govindaraju N K, Lloyd B, WANG Wei, et al. Fast computation of database operations using graphics processors[C]//Proceedings of the ACM SIGMOD International Conference on Management of Data. Paris: ACM, 2004: 215-266.
[7] Badal A, Badano A. Accelerating Monte Carlo simulations of photon transport in a voxelized geometry using a massively parallel graphics processing unit[J]. Medical Physics, 2009, 36(11): 4878-4880.
[8] 卢风顺,宋君强,银福康,等.CPU/GPU协同并行计算研究综述[J].计算机科学, 2011,38(3):5-9+46.
[9] 龚春叶.面向异构体系结构的粒子输运并行算法研究[D].长沙:国防科学技术大学,2011.
[10] NVIDIA. NVIDIA CUDA统一计算设备架构编程模型[M]. Santa Clara: NVIDIA, 2007.
[11] Santhanam A P, Min Y, Neelakkantan H, et al. A multi-GPU real-time dose simulation software framework for lung radiotherapy[J]. International Journal of Computer Assisted Radiology and Surgery, 2012, 7(5): 705-719.
[12] Hissoiny S, Ozell B, Bouchard H, et al. GPUMCD: A new GPU-oriented Monte Carlo dose calculation platform[J]. Medical Physics, 2011, 38(2): 754-764.
[13] JIA Xun, GU Xuejun, Graves Y J, et al. GPU-based fast Monte Carlo simulation for radiotherapy dose calculation[J]. Physics in Medicine and Biology, 2011, 56(22): 7017-7031.
[14] 王静.基于并行技术的外照射剂量实时精确评估方法研究[D]. 合肥: 中国科学技术大学, 2015.
[15] SU Lin, YANG Youming, Bednarz B,et al. ARCHERRT—A GPU-based and photon-electron coupled Monte Carlo dose computing engine for radiation therapy: Software development and application to helical tomotherapy[J]. Medical Physics, 2014, 41(7): 071709.
[16] 杨子辉.大规模辐射场景三维实时仿真关键技术研究[D].合肥:中国科学技术大学,2016.
[17] 袁成前,祝美英,邓云李,等.基于虚拟现实的三维辐射场可视化仿真[J].核电子学与探测技术,2022, 42(6):967-973.
[18] 贺淑相.三维辐射场高效计算方法与可视化研究[D].北京:华北电力大学,2017.
[19] ZHANG Xinyang, ZHANG Hui, WANG Jian, et al. Deep learning-based fast denoising of Monte Carlo dose calculation in carbon ion radiotherapy[J]. Medical Physics, 2023, 50(12): 7314-7323.
[20] Witte M, Yan K, Unkelbach J, et al. A deep learning based dynamic arc radiotherapy photon dose engine[J]. Physics and Imaging in Radiation Oncology, 2024, 39: 100658.
[21] WANG Y, ZHANG Y, LI X, et al. A physically constrained Monte Carlo-neural network for BNCT dose calculation[J]. Medical Physics, 2024, 51(9): e535-e548.
[22] ZHOU Peng, CHANG Yankui, LI Shijun, et al. Clinical application of a GPU-accelerated Monte Carlo dose verification for Cyberknife M6 with Iris collimator[J]. Radiation Oncology, 2024, 19(1): 86.
[23] Bousse A, Marechal E, Reader A J. A review on low-dose emission tomography post-processing with deep learning[J]. Physics in Medicine & Biology, 2024, 69(17): 17TR01.
[24] 李仕军. 质子碳离子放疗实时蒙特卡罗剂量计算方法和应用[D]. 合肥: 中国科学技术大学, 2024.
[25] 杨旭辉. 基于积累因子优化的离散纵标-点核积分耦合方法研究[D]. 哈尔滨: 哈尔滨工程大学, 2023.
[26] 高克坤. 基于粒子事件的GPU蒙卡粒子输运并行计算方法研究及程序开发[D]. 衡阳: 南华大学, 2024.
[27] LI Teng, Narayana V K, El-Ghazawi T. Exploring graphics processing unit (GPU) resource sharing efficiency for high performance computing[J]. Computers, 2013, 2(4): 176-214.
[28] Karplus M. Development of multiscale models for complex chemical systems: from H+H₂ to biomolecules (Nobel Lecture)[J]. Angewandte Chemie (International ed. in English), 2014, 53(38): 9992-10005.
[29] 包彦, 贺清明, 曹良志, 等. 基于NECP-MCX的点核积分方法研究与程序研发[C]//中国核科学技术进展报告(第七卷)——中国核学会2021年学术年会论文集第6册(计算物理分卷、核物理分卷、核聚变与等离子体物理分卷、辐射物理分卷、核工程力学分卷). 北京: 中国原子能出版社, 2021: 12-19.
[30] 杨磊. γ辐照装置辐射场优化设计研究[D]. 衡阳: 南华大学, 2016.
[31] 李华, 赵原, 刘立业, 等. 基于MCNP对γ射线吸收剂量累积因子的计算与研究[J]. 辐射防护, 2017, 37(3): 161-168.