Abstract: With the rapid development of VLSI and graphics technology, especially the rapid development of 2D/3D graphics acceleration technology, GPU graphics acceleration has become the main issue of research. This paper designs the driver of light weight DirectFB graphics hardware acceleration system on FT. This paper, from the design and implementation of GPU graphics acceleration perspectives, introduces Radeon GPU graphics acceleration related principles and mainly researches GPU memory address space and command transport mechanism and others base technology implement. At last, this system is implemented.

Key words: DirectFB, Radeon GPU, graphics acceleration, GPU memory address space, command transport mechanism

摘要： 在超大规模集成电路和图形技术快速发展，特别是2D/3D图形加速技术高速发展的背景下，GPU图形加速已经成为人们研究的焦点。主要在飞腾平台上为轻量级的DirectFB图形系统设计了一种硬件加速驱动。从驱动的设计与实现角度介绍了Radeon GPU图形加速的相关原理，对飞腾平台上GPU主存地址空间和命令传送机制等关键技术进行了系统的研究，并最终实现了该系统。

关键词: DirectFB, adeon GPU, 图形加速, GPU主存地址空间, 命令传输机制