Optimizing Matrix Multiplication on Intel® Xeon Phi TH x200 Architecture

Matrix multiplication is ubiquitous in scientific computing. From computational science to machine learning, a large and diverse set of applications rely on the performance of general matrix-matrix multiplication (GEMM) subroutines. The Intel ^® Math Kernel Library(R) provides highly optimized GEMM subroutines that take full advantage of the available parallelism and vectorization in both Intel ^® Xeon ^® and Intel ^® Xeon Phi(TM) processors. In this paper we discuss the optimization of GEMM subroutines for the Intel ^® Xeon Phi ^TM x200 (code-named Knights Landing).