Abstract　 　　 

　Intel has recently released the Xeon Phi coprocessor aiming for providing an x86 based MIC (Many Integrated Core) architecture. Compared with popular CPU/GPU hybrid, the Xeon/Xeon Phi combination bears quite a bit common features in boosting float point operations for highly scalable parts of computations via integrating enormous number of microprocessors. However, there are much more fundamental differences between the two heterogeneous architectures, the most significant of which is that cores in Xeon Phi are all of the x86 compatible micro-architecture. Together with the on-chip Linux-like OS and encapsulated communication interfaces, Xeon Phi enables simple and flexible coding, compiling and building strategies that are unavailable on GPU. In this talk, the coding skills and some examples are presented to demonstrate how to transplant CPU based codes onto the Xeon/Xeon Phi platform. Some problems resulting from the imperfect support of C++ language are also addressed.

  Dr. Dali Kong obtained his B.Sc. degree in 2008 from Dept. of Astronomy, Nanjing University, China. He then obtained his PhD degree in 2012, under the supervision of Prof. Keke Zhang, from Dept. of Mathematics, University of Exeter, UK. His main areas of researches are theoretical and computational fluid mechanics, high performance computing and planetary sciences. He is also devoted to biological fluid systems where viscosity is dominant.