Since Adreno is on UMA systems, this extension would come in extremely handy for more manual control from the user space for memory management. This extension is very capable, but with great power comes great responsibility.
There is the chance with this extension if you've got your code set up incorrectly to trample GPU state, so one must be careful.
My application uses this extension to get the biggest speed up in our buffer management compared to regular glBufferData, glBufferSubData, or glMapBufferRange. We use this extension on desktop with AMD, Nvidia, and Intel GPUs and they all have a considerable increase in speed due to how my application must update its buffers. We also use this with the Nvidia K1 on Android in order to bring a speedup.
Are there any plans, or can there be plans to bring this extension or a GLES equivalent to the Adreno 3xx/4xx series of GPUs? Being an UMA system it should be fairly trivial on the driver end I would suspect, just needs to be exposed correctly.
Thank you, I'm trying quite hard to make my application as quick as possible on the Adreno platform.
We are looking into this feature and find it interesting and potentially useful for inclusion in a future driver (no eta provided). Note that there's currently no OpenGL-ES extension for this functionality today