> Did you mean to email me privately?  If not, feel free to forward my
> reply to the list.  In fact, I'm sure many would like to participate
> in this discussion.

Yeah, I guess it went private by mistake.  I'll just leave the whole
conversation quoted in this email then so others can follow if they
want.

> On 3/3/06, tsuraan <[EMAIL PROTECTED]> wrote:
> > > The "PCI code" is a piece of copyrightable material.  It amounts to a
> > > human-readable representation of circuitry whose purpose is to
> > > communicate between your design and a host computer.  To use it, you
> > > would have to "link" the PCI logic to your design.
> > >
> > > This is where the viral nature of the GPL takes effect:  If you use
> > > this board and our PCI core to develop some other product, you are
> > > designing something that is inseparably dependent on our copylefted
> > > work.  The instant you "release" this composite design, you are bound
> > > by the terms of the GPL (or any other license by which you get this
> > > IP).
> >
> > I would certainly say that if I write some verilog that uses your PCI
> > controller, then redistribution of that code requires adherence to the
> > PCI controller's source license.  By the terms of the GPL, "compiled"
> > forms of the design (bitfiles?) would also have to respect your
> > license.  But, if I take those bitfiles and use them to burn in an
> > ASIC, is selling that piece of hardware equal to distributing
> > software?
>
> Logic in chips is copyrightable.  That's been established.  In
> addition, if all you did was resynthesize the Verilog for the ASIC,
> then clearly, the logic in the ASIC is a derivative work of the
> Verilog code.  (All you've done is translate it, mechanically no
> less.)
>
> > I don't actually know the physical process of how an ASIC
> > is formed, so maybe the bit files are just there in a PROM on the
> > ASIC, but if it's actual hardware that's being shaped based off the
> > data in the bitfile, then I don't know if its distribution has to
> > follow the license of the original source.
>
> I would think so.
>
> > I know I can sell NVidia graphics cards without having a license to
> > the NVidia GPU's firmware and hardware designs.
>
> Yeah, but you didn't COPY the GPU.  By selling an nVidia chip, you are
> just reselling something you've already paid nVidia for.  On the other
> hand, if you were to use steal their HDL code or use a microscope to
> analyze their ASIC and then just copied it, you'd be in violation of
> copyright.

That distinction does make sense to me.  There is quite a difference
between selling a device and selling copies of a device :)

> > I don't really see
> > how you could keep an unethical company from doing the same with
> > hardware made from your design.  I'm not trying to annoy you or weasel
> > out of your license; I just really want to see the open graphics
> > project succeed, and my hardware design skills are somewhat lacking...
>
> Yes.  Proving that someone's copied your chip is very challenging.
> This is surely one of the major reasons why hardware developers do not
> share their HDL.  The OGP and Traversal are taking a major risk doing
> it this way.  We have copyright law on our side, but physics is
> another matter.
>
> You might say that one of the reasons that we're releasing our
> hardware HDL under GPL is that the community demanded it.  If that
> turns out to be a bad decision, and it ruins our plans for developing
> open hardware, they have only themselves to blame.

Yeah, I do hope it doesn't turn out to be a bad idea.  My biggest fear
is still lack of interest on the part of customers, but  with things
like $20k pci controllers being sold by Traversal, maybe you guys
really do have a good chance at making it in the business side :)
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)

Reply via email to