high performance computing on graphics processing units: hgpu.org

Initially, computers were invented as devices to speed-up computations and facilitate the performance of repetitive mathematical operations. Their wider application in different areas upgraded this basic role and converted computers from calculating machines into devices for processing large amounts of data, with processing understood in a very general sense. The wonderfully large and ever increasing variety of applications based on various forms of data and information processing imposes various challenges, and computer hardware is supposed to provide the appropriate answers. This is the motivation for the surprisingly fast evolution of computing power in the relatively short history of contemporary computers. It seems that presently the opening of graphics processing units (GPU) for general purpose computations (GPGPU) is an answer that can meet the high demands for large computing power in certain applications. Their efficient use can provide a short break in the inevitable race for entirely new computing technologies that are necessary and offer some extended time for the underlying research work. This chapter discusses first the rationales that lead to the development of computing resulting in the appearance of the GPGPU and the GPU as the underlying technological platform. Then, we present the foundations of GPU architecture and the GPU programming frameworks up to the extent that is necessary for understanding the applications of GPUs discussed in this book.