Allwinner VPU support for the Linux kernel • Hacker News Board

Re: Allwinner VPU support for the Linux kernel

#21

Post by ZenoArrow » Sat, Feb 03, 2018, 3:59 PM UTC

joezydeco wrote: post #20

ZenoArrow wrote: post #18

Earlier quoted context omitted.

Patents for what? Algorithms? I've heard similar arguments/excuses made before, I'd like to understand if there's any legal precedence for them, or if it's more of a hypothetical risk. Lastly, with the 'supply of documentation' approach, does this (to your knowledge) carry the same legal risks?

The API would expose what IP cores are inside the GPU. A competitor could easily dissect the code and figure out what blocks and/or techniques are violating their patent rights.

> "The API would expose what IP cores are inside the GPU."

How would an open-source driver developed by a third party be less effective at doing so?

Re: Allwinner VPU support for the Linux kernel

#22

Post by joezydeco » Sat, Feb 03, 2018, 4:38 PM UTC

ZenoArrow wrote: post #21

joezydeco wrote: post #20

Earlier quoted context omitted.

The API would expose what IP cores are inside the GPU. A competitor could easily dissect the code and figure out what blocks and/or techniques are violating their patent rights.

> "The API would expose what IP cores are inside the GPU." How would an open-source driver developed by a third party be less effective at doing so?

The third-party is probing into a black box. They have no idea what is or isn't there. So it could be plausible deniability on the side of the chipmaker.

Re: Allwinner VPU support for the Linux kernel

#23

Post by squarefoot » Sat, Feb 03, 2018, 6:48 PM UTC

amluto wrote: post #9

ZenoArrow wrote: post #8

Earlier quoted context omitted.

> "Allwinner would be pretty foolish not to fully fund this endeavour" Why would they fund it? If they saw the potential benefits, they could probably open source their own closed source drivers and avoid this Kickstarter being needed, or at the very least provide documentation to speed this work along.

If I were Allwinner, I'd be quite nervous about open-sourcing my code. Who knows what patents it infringes?

This is why the recent proposal by Bunnie of trading transparency against liability would be so important.

Re: Allwinner VPU support for the Linux kernel

#24

Post by dsr_ » Sat, Feb 03, 2018, 7:18 PM UTC

Narishma wrote: post #16

bonyt wrote: post #6

Earlier quoted context omitted.

It's still useful for some applications. ATSC digital broadcasts in the U.S., for example, are in MPEG-2, as are the signals provided by many cable companies. I have an HDHomeRun Prime, and purchased the MPEG-2 HW Decoder license[1] for my Raspberry Pi 3 to make things a bit smoother. [1]: http://www.raspberrypi.com/mpeg-2-license-key/

Isn't the Raspberry Pi 3 powerful enough to do MPEG-2 decoding in software?

Not if you want it to do anything else at the same time, and not if you need de-interlacing of the 1080i streams that half the over-the-air channels use.

Re: Allwinner VPU support for the Linux kernel

#25

Post by ZenoArrow » Sun, Feb 04, 2018, 7:03 AM UTC

joezydeco wrote: post #22

ZenoArrow wrote: post #21

Earlier quoted context omitted.

> "The API would expose what IP cores are inside the GPU." How would an open-source driver developed by a third party be less effective at doing so?

The third-party is probing into a black box. They have no idea what is or isn't there. So it could be plausible deniability on the side of the chipmaker.

How about documentation then? Let's say Allwinner releases C header files, plus high level descriptions of what each function does. Do they still maintain plausible deniability if the hardware design infringes on patents?