Well apparently nVidia support OpenGL ES with their desktop driver. I'm glad there is another option here now.
Originally Posted by Azpegath
They all do, it is a part of the OGL standard to support the same routines as GLES and provide it in the OGL driver. GLES 2 is supported by OGL 4.1+ and GLES 3 is 4.3+.
Originally Posted by b15hop
What exactly is the nature of the problem? For example:
Originally Posted by berarma
- is the issue an inability to have two drivers with the same name (or other such ID) on (some) systems? Can this not be fixed just by renaming the new API --- use the OpenGL SP (or whatever) driver with a different name and different name space...
- is the issue that even the newest chips do a lousy job of segregating state, so that it's not possible to have two "masters" both submitting orders to the chip because they'll collide and both want to control some non-guarded resource?
- is the issue sheer laziness? If the problem really IS software that's, say, more than 10 years old then, WTF, why can't you just route that old code to a purely software based renderer [it's not like it's going to demand leading edge performance] and then you can toss the legacy support not just from the driver but also from the HW.
The details are interesting because of, as always, the existence of Apple which has a demonstrated impatience with legacy stupidity and a willingness to do something about it far beyond the Wintel and Linux (and now orthodox ARM) worlds. If, as seems likely, Apple are going to ship their own GPU soon, there seems scope for both performance (smaller faster chip without legacy crap) and, more importantly, agility (faster design and verification of both the HW and the OGL driver) if they're willing to draw a line in the sand regarding what they support: Everything earlier than, say, GLES 2.0 routes to SW emulation (and gets abandoned in a year, when the line moves to GLES 3.0, or whatever).
Of course Apple's footling around with a better ARM GPU doesn't affect the Intel GPU world --- yet...
Looks like everyone focused on just the number of API functions and forgot about the other dozens of points stated in Rich's blog.
There is need for a new crossplatform API, not a revised OpenGL...