机构地区: 华南理工大学土木与交通学院土木工程系
出 处: 《中山大学学报(自然科学版)》 2008年第S2期89-92,共4页
摘 要: 提出了求解矩阵特征值的GPU(图形处理器)实现方法,分别用基于GPU的幂法和QR法求解矩阵的最大特征值和所有特征值。基于GPU的计算与基于CPU的计算相比较,证实其计算精度较好,运算时间比基于CPU的运算时间快2.7~7.6倍。 A GPU(graphics processing unit) implementation was presented for solving eigenvalues of a matrix.The power method and the QR method based on the GPU are used to solve the largest eigenvalue and all eigenvalues of a given matrix.The computations are compared with those by the CPU,and it is found that the computation accuracy is good,and the running time on the GPU is faster than that on the CPU by a factor of 2.7~7.6.
领 域: [自动化与计算机技术] [自动化与计算机技术]