GPU通用计算及其在计算智能领域的应用

发布时间：2018-03-21 14:21

本文选题：计算智能　切入点：群体智能　出处：《智能系统学报》2015年01期 　论文类型：期刊论文

【摘要】：在日趋复杂的图形处理任务的推动下,GPU已经演化成为具有众多计算核心、计算能力强大的通用计算设备,并被越来越多地应用于图形处理之外的计算领域。GPU具有高并行、低能耗和低成本的特点,在数据并行度高的计算任务中,相比与传统的CPU平台有着显著的优势。随着GPU体系结构的不断演进以及开发平台的逐步完善,GPU已经进入到高性能计算的主流行列。GPU通用计算的普及,使个人和小型机构能有机会获得以往昂贵的大型、超级计算机才能提供的计算能力,并一定程度上改变了科学计算领域的格局和编程开发模式。GPU提供的强大计算能力极大地推动了计算智能的发展,并且已经在深度学习和群体智能优化方法等子领域获得了巨大的成功,更是在图像、语音等领域取得了突破性的进展。随着人工智能技术和方法的不断进步,GPU将在更多的领域获得更加广泛的应用。
[Abstract]:Driven by the increasingly complex graphics processing tasks, GPU has evolved into a universal computing device with numerous computing cores and powerful computing power, and has been increasingly applied to the computing field. GPU has high parallelism outside of graphic processing. The characteristics of low energy consumption and low cost, in the high degree of data parallelism of computing tasks, Compared with the traditional CPU platform, it has obvious advantages. With the continuous evolution of the GPU architecture and the gradual improvement of the development platform, GPU has entered the mainstream of high-performance computing. Providing individuals and small organizations with access to the computing power previously provided by expensive large, supercomputers, To a certain extent, it has changed the pattern of scientific computing and the powerful computing power provided by the programming development mode. GPU has greatly promoted the development of computing intelligence. And has achieved great success in sub-fields such as deep learning and swarm intelligence optimization, especially in images, With the development of artificial intelligence technology and methods, GPU will be widely used in more fields.
【作者单位】：北京大学机器感知与智能教育部重点实验室;北京大学信息科学技术学院;
【基金】：国家自然科学基金资助项目(61375119,61170057,60875080)
【分类号】：TP332

【共引文献】