一种易于硬件实现的嵌入式GPU三角形光栅化算法
An Embedded GPU Triangle Rasterization Algorithm for Hardware Implementation
-
摘要: 在嵌入式GPU中, 三角形光栅化的性能是决定图形处理器性能的一个重要因素.传统的三角形光栅化算法需要处理大量的无关像素, 会降低GPU的处理速度.本文针对采用瓦片渲染的嵌入式GPU, 提出一种基于瓦片的递归的三角形光栅化算法, 三角形光栅化会引起图形的几何走样, 本文利用超级采样中的反走样算法, 只用加法和移位实现了RGSS反走样算法.用C++语言实现了所提出的三角形光栅化及反走样算法.并在Xilinx zc706开发板上对算法进行了验证.结果表明, 光栅化算法可完成对任意三角形的渲染, RGSS反走样算法处理后的三角形抗锯齿效果明显, 该算法易于硬件实现, 适用于采用瓦片渲染的嵌入式GPU中.Abstract: In Embedded GPU, Triangle rasterization is an important factor in determining the performance of graphics processors. Traditional algorithms generate several redundant pixels and cut down the efficiency of GPU. A recursive triangular rasterization algorithm is proposed for tile-based embedded GPU. In addition, the triangular rasterization will cause the aliasing of the graph. The super-sampling anti-aliasing algorithm, RGSS, is used and implemented only by addition and shift operations. The C + + language is used to implement the triangular rasterization and anti-aliasing algorithm. The results show that the rasterization algorithm can complete the rendering of any triangle, and the triangular anti-aliasing effect of RGSS anti-aliasing algorithm is obvious. Anti-aliasing algorithm can significantly improve the rasterization of the anti-aliasing effect and suitable for using in tiled base rendering embedded GPU.