一种基于Matrix的QR分解向量化方法

提出一种基于Matrix的Givens旋转的QR分解向量化方法。针对Matrix的体系结构特点,对向量数据访存和计算进行优化,使计算均衡分布到各个向量处理单元;设计双缓冲DMA的数据传输策略,使得内核的计算与DMA数据搬移的时间完全重迭,内核始终处于峰值计算,从而取得最佳的计算效率。实验结果表明,该方法能够取得较高的计算效率和性能加速比。

A vectorization method of QR decomposition based on Matrix

We propose a vectorization method of QR decomposition with Givens rotation on Matrix processors. According to the systematic characteristics of Matrix architecture, the computation tasks are evenly distributed to all vector processing elements by optimizing the memory access to vector data and calculation. We also design a double DMA buffering scheme to smooth the data transfers, which can fully overlap the kernel computation time and the DMA data transfer time so that the kernel computation is always at its peak speed and the best computation efficiency is achieved. Experimental results show that the proposal can achieve higher computation efficiency and performance speedup.