Face-backed culling, clipping culling and deferred shading are wildly used in
embedded graphical rending pipeline. At the same time,the mobile devices are powered by batteries, so energy consumption is also an importation consideration in design. Energy consumption is proportional to the amount of memory access, tiling architectures proposed aim to reduce the memory traffic to reduce energy consumption, but the tiling architectures require additional buffer to store each triangle and some triangles may be rendered many times that causes the wasting of bandwidth, so it is still an argument. In
future work, we should to solve the problems above in tiling architectures. That time tile-base architecture will be very suitable to mobile devices. Although not as
well as culling to avoid data processing, compression is also a very important issue. Since texturing needs huge bandwidth, texture compression is naturally supported
in OpenGL ES. Z-buffer algorithm is simply and suitable for hardware implementation, but it requires large depth buffer and huge bandwidth. It is know that the depth values are linearly interpolated in screen space, so many compressions of depth buffer using the
property are proposed, but the depth buffer compression now widely used is only to reduce the bandwidth. So a compression proposed to reduce both the size of buffer and the bandwidth is of great significance.