Hi, it’s been forever since I’ve had the chance to work on a powervr chip ( dreamcast ), so I’m quite thrilled that I’ve got the opportunity again. I remember back in the day it was quite a feature to not have a frame buffer and instead use the memory for textures or what not.
What’s the situation like these days? It came to my attention since the app I’m working on doesn’t require a glClear but I do notice I can create a hall of mirror effect so there must be a persistent frame buffer there.
I’m wondering is there an optimization to be had somewhere? Any detailed docs on this stuff?
Thanks!
Hi,
One of the big advantages of tile-based deferred rendering is that you don’t need an external depth buffer or multisample buffers. However, there is always a persistent framebuffer since that is required by the scanout hardware that sends the image to the screen. The framebuffer is typically double-buffered.
You should always use glClear unless you really need the contents of the previous frame as background. Note however that you’re only guaranteed to get the contents of the previous frame if the EGL_SWAP_BEHAVIOR attribute of the surface equals EGL_BUFFER_PRESERVED.
