ABSTRACT
In the last decade, with the booming development of information and communication technologies, human society has stepped into the era of big data. A huge amount of information is being generated every moment, accumulating into what can be regarded as “data gold mines.” Many types of information can be naturally abstracted as graph-structured data, for example, social network graphs, web link graphs, consumer–product relationship graphs, and the corresponding real-world problems can be naturally converted into graph computation problems. In recent years, as the size of graph-structured data is getting larger and larger, efficiently analyzing and processing large-scale graph-structured data can bring more and more significant scientific, economic, and social benefits, making large-scale graph computational problems a subject of widespread interest and attention in both academic and industrial circles. This chapter focuses on the customization and optimization of hardware accelerators for graph processing algorithms, starting with an introduction to the background of hardware acceleration for graph algorithms in Section 6.1, followed by an introduction to the basic computational principles in graph processing algorithms, and concluding with an introduction to typical computational systems and accelerator customization-related work for graph processing algorithms.
