Math Kernel Library Automatic Offload for Intel® Xeon Phi™ Coprocessor

Intel® Math Kernel Library (MKL) includes a unique new Automatic Offload (AO) feature that enables computationally intensive Intel® MKL functions called in user code to benefit from attached Intel® Xeon Phi™ coprocessors automatically and transparently. This special feature allows users to leverage the additional computational resources provided by the coprocessor without changes to Intel MKL calls in their codes or modifications to the way they compile and link. AO allows programmers to use the same usage model they are familiar with, while enjoying performance scaling from multi-cores to many cores.

Learn how to use this feature: Download the white paper (PDF, 360KB) in the attachment.

For more complete information about compiler optimizations, see our Optimization Notice.