Performance Comparison of a Two-Dimensional Elliptic Test Problem on Intel Xeon Phis
Ishmail A. Jabbie, George Owen, Benjamin Whiteley, Jonathan S. Graf, Matthias K. Gobbert, Samuel Khuvis · Maryland Shared Open Access Repository (USMAI Consortium) · 2016
The Intel Xeon Phi is a many-core processor with a theoretical peak performance of approximately 1 TFLOP/s in double precision. This project contrasts the performance of the second generation Intel Xeon Phi, code-named Knights Landing (KNL), to the rst-generation Intel Xeon Phi, code-named Knights Corner (KNC), as well as to a node with two CPUs as baseline reference. The benchmark code solves the classical elliptic test problem of the two-dimensional Poisson equation with homogeneous Dirichlet boundary that is prototypical for the computational kernel in many numerical methods for partial differential equations. The results show that the KNL can perform approximately four times faster than the KNC in native mode or two CPUs, provided the problem ts into the 16 GB of on-chip MCDRAM memory of the KNL.