The Intel® Xeon Phi™ coprocessor platform has a software stack that enables new programming models, including offload of computation from host processor to the Intel® Xeon Phi™ coprocessor to improve response time and/or throughput. A new paper shares draws on insights from a multi-year, intensive development effort to answer common questions, why offload to a coprocessor is useful, how it is specified, and conditions for the profitability of offload. You’ll learn about: the software architecture and design of the offload compiler runtime, key performance features and their impact for a set of directed micro-benchmarks and larger workloads.
Download the paper here. (pdf)