ABSTRACT
With the rapid data scale growth, machine-learning algorithms are widely adopted to deal with big data. It naturally becomes the focus of researchers and software developers how to run such algorithms efficiently on modern many-core platforms. In this chapter, two unique hardware platforms, Intel® Single-Chip Cloud Computer (SCC) and Intel® Xeon® Phi™ coprocessor, are employed to accelerate the backpropagation (BP) neural network and lead to a better performance for data processing. Furthermore, we also explored dynamic voltage and frequency scaling (DVFS) on the many-core platform in searching of the suitable configuration to aid the processing of big data when the power/energy savings need to be taken into consideration.
