I came across a link to this article in the SL Forum: NVIDIA, AMD, Intel Explain How OpenGL Can Unlock 15x Performance Gains. That means if you are getting 20 FPS you should jump to 300 FPS.
The long version is the video and it is way techy:[youtube -bCeNzgiJ8I]
So, what are they saying? Loosely… They are creating sides of the render process that are not locked together. This means the CPU and GPU are mostly never waiting on each other. As it is now, single note pads (large blocks of memory} are being cleared, being drawn to by a single process to build up images, then used, and erased with the process starting over. Multiple processes need multiple note pads. The next process can’t use the resources until the preceding processes release the resources/note pads. The memory that makes up the note pads has to be moved around.
The newer idea is to change how processes work so they can be writing to and changing the buffer at the same time without waiting or a need to erase the buffer. Plus the buffer doesn’t need to move around.
The take away is that using the new ideas and features in OpenGL results in far fewer waits, less work, allows more things to be done at once, and more activities combined to run in parallel. The result is a huge gain in OpenGL render performance.
So, will these ideas be useable in Second Life? Probably. Will the Lab adopt them? Probably. When we will see the boost in performance? No clue. But, I would say not any time soon.
If I assume the Lab is just hearing about this at the GDC meet up, then it will likely be a year or more before it starts coming out the other end of the tube. But, that is pure speculation on my part.
Will a third party add it? I suppose there is a chance. But, most third party developers do not have people with experience in render engines. That is a part of the viewer few developers mess with. Even the Lab only has a small number of people that know render engines. So, I’m not hopeful that third parties will get out front on this issue. But, they may surprise us.
Apparently the biggest gains will be in the Android operating systems. That would be nice. But, it has potential for desktop games too.
If the Lab does change the engine to use the new techniques, the upgrade in performance would allow them to add more features that would currently drag the system down to unacceptable levels. So, I expect they will be interested. But, with a new CEO a project of this scope won’t start without his blessing. I doubt he will be starting any new massive projects until he has more of an idea of what SL is and where he wants to take it.