Výpočetní platforma Intel Xeon Phi

Abstract

This thesis discusses the possibility of parallel programming on Intel Xeon Phi coprocessor. Describes in detail the architecture, its features, advantages, disadvantages, possibilities of use and the overall issue of high parallel programming today. The thesis also contains a brief description of other technologies such as OpenCL and CUDA, comparing them with Xeon Phi at the hardware level, but also from a programmer's perspective, which focuses on compilers, programming languages, libraries, including individual examples of source code. The second part is more practical when it is made detailed test of coprocessor. There is explain the issue of compiling and running computing applications in a variety of modes for a few specific examples, which are then under a different set parameters tested. In addition, a sample of some useful functions and features to facilitate work with coprocessor.