栅元有效共振积分的CUDA算法设计与实现

J4 ›› 2016, Vol. 38 ›› Issue (02): 224-230.

栅元有效共振积分的CUDA算法设计与实现

任成磊1,蒲鹏2,韩定定1

（1.华东师范大学多维度信息处理重点实验室,上海 200241；
2.华东师范大学信息科学技术学院,上海 200241）

收稿日期:2015-02-28 修回日期:2015-05-21 出版日期:2016-02-25 发布日期:2016-02-25

Design and implementation of CUDA algorithm
for calculating effective resonance integral

REN Chenglei1,PU Peng2,HAN Dingding1

(1.Shanghai Key Laboratory of Multidimensional Information Processing,East China Normal University,Shanghai 200241;
2.School of Information Science and Technology,East China Normal University,Shanghai 200241,China)

Received:2015-02-28 Revised:2015-05-21 Online:2016-02-25 Published:2016-02-25

摘要/Abstract

摘要：

核反应堆中需要实时精确地计算堆芯和增殖材料的有效共振积分或群截面来实现反应堆的安全控制。整个计算过程因为涉及大量的积分运算和庞大的核素截面数据,采用常规的计算方法,计算时耗相当大。基于统一计算设备架构（CUDA）平台,利用图形处理器（GPU）的计算能力,对整个计算过程进行并行化分解,多线程同时运算,大幅度提升计算速度,降低时耗。实验结果表明,在GPU上并行计算所得结果与原始数据没有明显差异,且加速效果显著。

关键词: 慢化源递推公式, 有效共振积分, CUDA, 并行计算, GPU

Abstract:

Nucleon reactors need realtime accurate calculation of effective resonance integral or groupaveraged effective cross section to achieve the safety control of reactors. Because the whole calculation process involves a lot of integral operations and huge section data, the time consumption of conventional calculation methods is considerably huge . Based on the compute unified device architecture (CUDA) platform, the whole calculation process with multithreaded operations at the same time can greatly improve the computing speed and reduce the time consumption with the help of the computing power of the graphic processing unit (GPU). Experimental results show that the parallel computing results on the GPUs have no significant difference from the original data, and the proposal improves the speedup significantly.

Key words: moderated source recurrence formula;effective resonance integral;CUDA;parallel computing;GPU

任成磊1,蒲鹏2,韩定定1. 栅元有效共振积分的CUDA算法设计与实现[J]. J4, 2016, 38(02): 224-230.

REN Chenglei1,PU Peng2,HAN Dingding1. Design and implementation of CUDA algorithm
for calculating effective resonance integral [J]. J4, 2016, 38(02): 224-230.

参考文献［6］

［1］	Kier P H,Rabba A A.RABBLE,A program for computation of resonance absorption in multiregion reactor cells［R］. Argonne National Lab,1967.
［2］	Wen M W. GPU computing gems Jade edition［M］.New York:Elsevier,2011.
［3］	Zeng Bo,Lei Youcheng,Wang Congzhi,et al.CUDAbased acoustic radiation force imaging algorithm［J］.Computer Engineering and Application,2014,48(18):249254.(in Chinese)
［4］	Roshd M H M,Chow H C.GETRANSA two dimensional integral transport code［R］.Ontario:Atomic Energy of Canada Ltd,Sheridan Park,Power Projects,1978.
［5］	Ma Yichen,Gao Wei,Tang Fuchu,et al.Numerical solution of threedimension integral transport equation—calculation of neutron collision probabilities［J］.Journal of Engineering Mathematics,1984,1(1):110128.(in Chinese)
［6］	Hu Yongming,Jing Xingqing,Wang Guilin.Effective resonance integrals calculations using recuurence formulas in reactor cells［J］.Nuclear Power Engineering,1982,3(3):3444:005.(in Chinese)
［7］	Lane R K,Nordheim L W,Sampson J B.Resonance absorption in materials with grain structure［J］.Nuclear Science and Engineering,1962,14:390396.
［8］	Adir J,Lathrop K D.Theory of methods used in the GGC4 multigroup cross section code:Report GA9021［R］.Calif:General Automic Div,General Dynamics Corp, 1968.
［9］	Dyos M W.The construction of statistical neutron resonances［J］.Nuclear Science and Engineering,1968,34(2):181188.
［10］	Teuchert E.Resonanz absorption in einer zweifach heterogenen Anordnung Kugelfrmiger Brennelemente［J］.Nukleonik,1968,11:6875.
［11］	Kuncir G F.A program for the calculation of resonance integrais:Report GA2525［R］.Calif:General Atomic Div,Dynamics Corp,1961.
［12］	Nordhem L W.A program of research and calculations of resonance absorption:Report GA2527［R］.Calif:General Atomic Div,General Dynamics Corp,1961.
［13］	Sterens C A,Smith C V.A computer program for evaluating resonance absorption including resonance overlap:Report GA6637［R］.Calif:General Atomic Div,General Dynamics Corp,1965.
［14］	Marowka A.On performance analysis of a multithreaded application parallelized by different programming models using Intel VTune［M］∥Parallel Computing Technologies.Berlin:Springer Berilin Heidelberg,2011:317331.
	附中文参考文献：
［3］	曾博,雷友诚,王丛知,等.基于CUDA 的声辐射力弹性成像算法研究［J］.计算机工程与应用,2014,48(18):249254.
［5］	马逸尘,高伟,唐富初,等.三维中子积分输运方程的数值解——中子碰撞概率的计算［J］.工程数学学报,1984,1(1):110128.
［6］	胡永明,经荥清,王桂林.用递推公式计算栅元有效共振积分［J］.核动力工程,1982,3(3):3444.

[1]	苏丽，孙彦猛，张博为，杨先博，朱颖. 一种基于Hadoop+CUDA实现相关器的方法[J]. J4, 20160101, 38(01): 46-51.
[2]	刘屹成, 刘晓燕, 严馨. 并行平衡级联支持向量机[J]. 计算机工程与科学, 2023, 45(07): 1170-1177.
[3]	陈强, 谭林, 王云丽, 肖靖. 工业区块链中基于CUDA的数据并行处理方法[J]. 计算机工程与科学, 2022, 44(12): 2102-2110.
[4]	臧照虎, 李晨, 王耀华, 陈小文, 郭阳. 面向众核系统的层次化栅栏同步机制[J]. 计算机工程与科学, 2022, 44(11): 1901-1908.
[5]	张勇, 张曦, 万云博, 何先耀, 赵钟, 卢宇彤. 非结构有限体积CFD计算的网格重排序优化[J]. 计算机工程与科学, 2022, 44(10): 1721-1729.
[6]	朱浩, 周博洋, 卢雪山, 杜溢墨. OpenCL计算软件栈评估[J]. 计算机工程与科学, 2021, 43(12): 2105-2114.
[7]	范培勤, 过武宏, 韩梅, 唐帅, 张驰, . 水声环境特征参数并行预报方法研究[J]. 计算机工程与科学, 2021, 43(11): 1920-1925.
[8]	罗磊, 陈照云, 王俪璇. 用户QoS感知的GPU集群深度学习任务动态调度[J]. 计算机工程与科学, 2021, 43(08): 1331-1340.
[9]	龚昊, 刘莹, 冯建周, 赵仁良, 冷佳旭, . 基于GPU加速的脉冲多普勒雷达信号处理[J]. 计算机工程与科学, 2021, 43(07): 1141-1149.
[10]	焦育威, 王鹏, 辛罡, . 基于采样尺度自适应的多尺度量子谐振子优化算法并行化[J]. 计算机工程与科学, 2021, 43(07): 1200-1209.
[11]	俞茂学, 贾东宁, 魏志强, 许佳立, 马广浩. 一种基于国产异构众核处理器的C++智能源码转换框架[J]. 计算机工程与科学, 2021, 43(06): 997-1005.
[12]	丁哲昭, 储根深, 胡长军, 李扬. 基于申威众核处理器的圣维南求解程序的并行与优化[J]. 计算机工程与科学, 2021, 43(05): 820-829.
[13]	孙兆鹏, 周宽久. 基于PCIe的高性能FPGA-GPU-CPU异构编程架构[J]. 计算机工程与科学, 2021, 43(04): 641-651.
[14]	魏金晖, 李晨, 鲁建壮. 多GPU系统虚实地址转换架构研究[J]. 计算机工程与科学, 2021, 43(02): 228-234.
[15]	张亮, 常旭, 秦志楷, 沈立. 量子线路模拟器QuEST在多GPU平台上的性能优化[J]. 计算机工程与科学, 2021, 43(01): 17-23.