I just wanted to ask those reading and moderating this board what business opportunities they envisage for creators of software implementations of OpenGL ES (if any)?
Several members of Khronos are in the business of building OpenGL ES software renderers (and related activities). While I can’t speak for their business plans, my impression is that they are partnering with mobile device manufacturers and service providers to incorporate their software into e.g. cell phones and PDAs. Some of these companies also sell higher level software layers (gaming engines and the like) which will utilize OpenGL ES.
It is an interesting question how this will hold up once 3D hardware acceleration comes to the mobile device market, which looks like it will start happening in a big way in late 2004. There may be smaller devices (watch-tops? which continue to need software renderers, though.
Apart from the obvious areas of mobile phones and PDAs, there are also plenty of possibilities in military applications, automotive industry, civil avionics and various emerging industries. Also, there are interesting opportunities for hybrid SW/HW implementations (working on DSP, transputers and whatnot).
Writing an OpenGL ES implementation is not particularly tricky; there are currently many in-house “prototype” implementations. However, writing an industrial strength one requires a lot of hard work and expertise.
OpenGL ES is mainly targeted for embedded devices, and most of the markets there (i.e. areas where you can actually have a solid business) are very concerned about reliability and robustness. Having a product accepted in this space requires a huge amount of testing.
To give some indication about the amount of work involved, we spent approx. 60 man-months on building our OpenGL ES 1.0 implementation, ~70% of the time being spent on testing-related activities. And this was our 13th or 14th rasterizer implemented during the last decade, so we weren’t complete newbies. On top of this we of course have to do quite a lot of integration work per CPU/OS/device.
So, writing an implementation for your own fun, or maybe to be used for several games, is probably not that difficult, although I’m not sure where the money would come from.
o Ville Miettinen
o Hybrid Graphics, Ltd.
This topic was automatically closed 183 days after the last reply. New replies are no longer allowed.