共查询到19条相似文献,搜索用时 61 毫秒
1.
特征线方法(MOC)在求解堆芯规模中子输运方程时面临计算时间长的问题,加速和并行算法是目前研究的热点。基于MOC在特征线和能群层面的并行特性,采用统一计算设备构架(CUDA)编程规范,实现了基于图形处理器(GPU)的并行二维MOC算法。测试了菱形差分和步特征线法分别在双精度、混合精度及单精度浮点运算下的计算精度、效率及GPU加速效果。采用性能分析工具对GPU程序性能进行了分析,识别了程序性能瓶颈。结果表明:菱形差分和步特征线法在不同浮点运算精度下均表现出良好的计算精度;相比于CPU单线程计算,GPU加速效果在双精度和单精度情况下分别达到35倍和100倍以上。 相似文献
2.
直接模拟方法(DSM)是一种新的用于求解三维瞬态中子输运问题的方法.该方法通过直接模拟瞬态过程中系统内中子和缓发中子先驱核的动态行为来求解核反应堆动力学问题.由于该方法取消了现有方法的各种近似,具有普适通用性.本文在详细研究该方法的基础上,开发了相应的瞬态分析程序TMCC,并进行了算例的验证. 相似文献
3.
CPU-GPU异构系统为加速全堆芯特征线方法(MOC)精细计算提供了方法和思路。在实现基于CPU-GPU异构系统的二维MOC异构并行算法基础上,提出了性能分析模型,识别了影响异构并行算法并行效率的主要因素;针对识别到的性能影响因素,实现了输运计算与数据传递相互掩盖,提升了异构并行算法的整体并行效率。数值结果表明:程序具备良好的计算精度;数据传递(MPI通信和CPU与GPU之间的数据拷贝)是影响异构并行算法并行效率的主要因素;实现输运计算与数据传递相互掩盖后,程序性能和强并行效率均有所提升;5异构节点(包含20块GPU)并行时,程序整体效率提升达8%,强并行效率从87%提升到95%;相比CPU节点并行计算,4个CPU-GPU异构节点整体性能优于20个CPU节点。 相似文献
4.
5.
6.
7.
8.
9.
10.
12.
13.
14.
俄罗斯商用压水堆VVER和大多数实验堆均采用了六角形紧凑型栅格布置,为了实现VVER和六角形实验堆的高保真数值模拟分析,本文基于数值反应堆物理计算程序(NECP-X)开展了六角形堆芯高保真计算方法研究和程序开发。首先,将全局-局部耦合共振自屏计算方法拓展至六角形堆芯,实现六角形堆芯燃料棒的全堆芯高精度共振计算;其次,基于2D/1D耦合输运计算方法研究了六角形堆芯的高保真计算方法;最后,为了提高全堆芯计算的计算效率,研究了基于区域分解松耦合的非结构网格的粗网有限差分(CMFD)加速方法,可以实现以矩形、六角形和其他多边形栅元为基础的pin-by-pin CMFD 加速。为了验证六角形堆芯高保真计算方法的精度和效率,计算了六角形C5G7基准问题,并分析了六角形输运计算方法的计算精度和CMFD方法的加速效果;将NECP-X程序应用于西安脉冲堆的2D全堆芯计算,与蒙特卡罗程序的结果对比表明NECP-X程序计算得到的特征值和功率分布均具有较高精度。因此,本文建立的六角形堆芯高保真计算方法可以应用于六角形堆芯的分析计算。 相似文献
15.
16.
GPU加速三维特征线方法的研究 总被引:3,自引:0,他引:3
三维特征线方法可以精确求解任意几何堆芯的稳态多群中子输运方程,但同时也具有收敛慢、计算时间长的不足,需要研究相应的加速手段.图形处理器(GPU)计算由于具有速度快,能耗低的优点,被认为是未来高性能计算发展的方向之一.研究GPU计算加速三维特征线方法,并将其应用到三维特征线程序TCM中.借助统一计算设备架构(CUDA)的GPU计算,中央处理器(CPU)负责内存分配、有效增殖系数keff和源分布计算等逻辑性强或归约计算的处理,GPU执行特征线射线扫描细网求解细网通量.计算结果表明,经改写后的程序具有良好的加速效果. 相似文献
17.
《Journal of Nuclear Science and Technology》2013,50(9):1156-1171
The Simplified PN (SPN) method is applied to the axial solution of the two-dimensional (2-D) method of characteristics (MOC) solution based whole core transport calculation. A sub-plane scheme and the nodal expansion method (NEM) are employed for the solution of the one-dimensional (1-D) SPN equations involving a radial transverse leakage. The SPN solver replaces the axial diffusion solver of the DeCART direct whole core transport code to provide more accurate, transport theory based axial solutions. In the sub-plane scheme, the radial equivalent homogenization parameters generated by the local MOC for a thick plane are assigned to the multiple finer planes in the subsequent global three-dimensional (3-D) coarse mesh finite difference (CMFD) calculation in which the NEM is employed for the axial solution. The sub-plane scheme induces a much less nodal error while having little impact on the axial leakage representation of the radial MOC calculation. The performance of the sub-plane scheme and SPN nodal transport solver is examined by solving a set of demonstrative problems and the C5G7MOX 3-D extension benchmark problems. It is shown in the demonstrative problems that the nodal error reaching upto 1,400 pcm in a rodded case is reduced to 10pcm by introducing 10 sub-planes per MOC plane and the transport error is reduced from about 150pcm to 10pcm by using SP3. Also it is observed, in the C5G7MOX rodded configuration B problem, that the eigenvalues and pin power errors of 180 pcm and 2.2% of the 10 sub-planes diffusion case are reduced to 40 pcm and 1.4%, respectively, for SP3 with only about a 15% increase in the computing time. It is shown that the SP5 case gives very similar results to the SP3 case. 相似文献
18.
The method of characteristics (MOC) has great geometrical flexibility but poor computational efficiency in neutron transport calculations. The generalized minimal residual (GMRES) method, a type of Krylov subspace method, is utilized to accelerate a 2D generalized geometry characteristics solver AutoMOC. In this technique, a form of linear algebraic equation system for angular flux moments and boundary fluxes is derived to replace the conventional characteristics sweep (i.e. inner iteration) scheme, and then the GMRES method is implemented as an efficient linear system solver. This acceleration method is proved to be reliable in theory and simple for implementation. Furthermore, as introducing no restriction in geometry treatment, it is suitable for acceleration of an arbitrary geometry MOC solver. However, it is observed that the speedup decreases when the matrix becomes larger. The spatial domain decomposition method and multiprocessing parallel technology are then employed to overcome the problem. The calculation domain is partitioned into several sub-domains. For each of them, a smaller matrix is established and solved by GMRES; and the adjacent sub-domains are coupled by “inner-edges”, where the trajectory mismatches are considered adequately. Moreover, a matched ray tracing system is developed on the basis of AutoCAD, which allows a user to define the sub-domains on demand conveniently. Numerical results demonstrate that the acceleration techniques are efficient without loss of accuracy, even in the case of large-scale and strong scattering problems in complex geometries. 相似文献