主要还是让A(i,j) B(i,j)进行移动A(i,i+j),B(i+j,j)，然后进能直接相乘。具体的内容可以看并行算法导论，-Matrix multiplication with the Cannon Algorithm: The Cannon Algorithm for matrix multiplication was presented in the course “Parallel and Distributed Algorithms “by Dr. Klauck. Sh

CUDA Application Design and Development，CUDA程序设计与开发，最近刚出不久的一本书，里面的技术很新，相比上一代Fermi架构，新的Tesla并行架构，并行效能效果进一步增强。-CUDA Application Design and Development Author(s): Rob Farber Published: 14 Nov 2011 As the computer industry retools to leverage mass