Matrix Inversion on a PIM (Processor-in-Memory)
Jae Chul Cha, Sandeep K. S. Gupta · 2008
Many computational applications in science and engineering involve solving matrix inversions with large data sets. To reduce the computation time, parallel architectures are used with appropriate data partitioning and placement schemes. In this paper, matrices are inverted on a PIM (processor-in-memory) architecture by applying a blockwise analytical matrix inversion scheme in a recursive manner, and whose performance is evaluated and analyzed.