News for hackers. Stuff that matters.
Re:Android AOSP maintainer quits by krelian on Wednesday August 07, 2013 @03:31PM (#6175064)
What's the actual risk for the vendor when releasing the drivers? Isn't all the magic in the chip itself and driver only serves as an interface?
Re:Android AOSP maintainer quits by iyulaev on Wednesday August 07, 2013 @05:59PM (#6175943)
I suspect it's a combination of support, documentation, and trade secrets. I've worked with numerous vendors over the last decade and the support strategies vary dramatically. Some vendors put up very nice datasheets, manuals, source code, etc, while some give you almost nothing at all, and instead ship an engineer with the product to help you design it in. The rationale is that these chips are very complicated, applications vary dramatically, and it's sometimes better to increase the quantity of manpower (in engineer form) for your favorite 5 customers (that make up most of your market anyway) rather than to increase the quantity of documentation so that you can also sell to Joe & Jane Hobbyist.
Re:Android AOSP maintainer quits by Osiris on Wednesday August 07, 2013 @03:55PM (#6175189)
I have no experience with GPUs, but I once read something that indicated that the driver, in fact, has a lot of proprietary code in it. For example, nVidia often releases driver updates after a new game is released that can give 10-25% performance boosts in those games. That's not because it's running the GPU faster, that's because of optimizations within the driver on how GPU commands get processed by the GPU.

Someone with more knowledge about this please correct me if I'm wrong.

Re:Android AOSP maintainer quits by VLM on Wednesday August 07, 2013 @05:27PM (#6175765)
"Isn't all the magic in the chip itself"

Yeah thats the problem.

There exists a patent troll who owns patents for using pink unicorns to render pixels. Vendor ships a chip that uses red horses to render pixels... close enough for a legal patent battle? Who knows. But if they try to keep it a secret, maybe it'll all work out.

Re:Android AOSP maintainer quits by nileshtrivedi on Wednesday August 07, 2013 @07:53PM (#6176470)
That explains not open-sourcing the drivers. But it doesn't explain not distributing the binaries - which is the case here.

Edit: Fixed typo.

Re:Android AOSP maintainer quits by kalleboo on Thursday August 08, 2013 @10:24AM (#6179062)
What makes no sense to me is what's the risk for the vendor in letting someone redistribute the driver's binary blobs. Any competitor who wants to reverse-engineer the code can already extract it from the released product anyway!
Re:Android AOSP maintainer quits by joezydeco on Wednesday August 07, 2013 @03:42PM (#6175115)
What if the driver is only unlocking part of the GPU's capabilities? A vendor can spin one chip for n customers, all paying different royalty and licensing fees based on need.
Re:Android AOSP maintainer quits by kbenson on Wednesday August 07, 2013 @03:57PM (#6175196)
But isn't the problem they aren't allowing the binary blob to be released? There's little reason to believe someone is going to reverse engineer and open up capabilities of a chip in a binary blob (at least little reason to believe that will be more likely because it's released rather than pulled directly from a device).