This module looks at the rationale behind the approach taken by Intel with regards to the Intel® Xeon Phi coprocessor HW abstraction, programming model options, using standardized libraries such as Intel® MKL, high level view of offloading code, Intel® CilkTM Plus array notation and task parallelism, writing code for pure auto-vectorization, semi-automatic vectorization by annotation, and a comparison of architecturally independent vs dependent programming approaches.

Section 1: Intel® approach, HW abstraction