Phoronix: Would A Kickstarter Open-Source GPU Work?
The Kickstarter crowd-funding development platform has successfully worked for funding many computer games, a few interesting hardware projects, and other initiatives, but would it work for having an open-source graphics processor? A company may be turning to Kickstarter to open-source their 2D and 3D graphics designs...
is there any benefit of an opensource GPU? i dont see why it should be attractive project.
Well one would speculate it would have the same advantages of an opensource software project in that other companies are free to expand on/fork the work, or pick up where another company left off. Once something is open source its life can go on infinitely as long as there is enough interest, unlike proprietary projects which die a lonely death behind closed doors.
The fun thing with Kickstarter projects is you can contribute as little or as much as you like to projects just to see what they come up with not necessarily because what they are doing is going to be a superior product (especially on the first attempt). The OUYA console is a great example. This was always going to be a second rate device and I had no interest in it personally but many many people just wanted to see what they would come up with because thats half the fun.
As for this company asking Michael for his opinion sounds strange why wouldnt they just attempt the Kickstarter. I'm sure if field experts were asked about many of the succefully funded projects they would have told the projects not to waste there time on crowd funding but the only way to really know if people are interested is to try it. My guess would be they are trying to build up some hype.
I think if the GPU can compete on a level with Nvidia and AMD, then yes, it would be a good thing to do. As far as desktop graphics, I find it hard for myself to not stick with Nvidia simply because their Linux blob driver "just works" in almost every case I need it to. Nouveau has been improving but still doesn't quite cut it for me.
A GPU that will work consistently on Linux/Windows/Mac, no questions asked, is a surefire winner for me.
I think if they market the GPU properly to their prospective user base and can actually deliver on their promises, then it will be a huge success and will really stir things up in the GPU arena. Using "open source" as the sole means to push a bad product would be disaster.
I would through money behind it if it meets some specific requirements.
1. OpenGL 2 at the very very least. Much prefer 3.0 to be the minimum, but I'm aware patents and the like can get in the way of this.
2. Full Verilog Code being GPL'd, along with driver Code, no silly wrappers or fuse blobs.
3. Decent power management.
4. Support for the Linux Mir/Wayland. Would like to see Gallium framework used, unless for any reason it may be too restrictive.
5. No UVD? Point of open source card with smatterings of closed source chips on it.
6. Decent performance.. I don't expect an nvidia, amd killer, If it can outperform Intel's Iris, That would be handy.
7. Active corperate backing. I'd like to see it made, marketed, deals and talks made with Valve and any other interested gaming company.. I'd also like to see them contribute. (Think of a valve console with open Graphics )
8. OpenCL support.
9. Scalability would be nice, Desktop to phone if possible. But I can understand if the architecture doesn't initially support this.
10. Tutorials educational material to get any budding engineer working and contributing to this project.
I think that would be enough as a start. I hope I'm bot being too ambitious. The biggest part would have to be the corperate backing though. If they opensource then abandon it, hope for the hardware will be slim. However, if they release code with apropriate licensing, actively make the hardware, while contributing the open source community will back them up and contribute. Well at least I hope I wouldn't be alone in that.