Abstract:

With the development of the semiconductor technology, the number of transistors integrated on a chip keeps increasing. Consequently, the computation and memory capacity of graphics processing units improve rapidly. So far, the floatingpoint computing capacity of GPUs has greatly exceeded that of CPUs, and the potential of GPUs in the nongraphic computing field, especially in high performance computing, has attracted more and more researchers’ attention. This paper gives an introduction to the principles of the general purpose computation on GPUs and the latest research results about architecture and the programming model of GPGPU from both the research community and industry.

Key words: GPGPU;architecture;programming model