[Openchrome-users] A bounty for XVMC on the CX700?

[Harald Welte]

Hi all,

On Thu, Sep 11, 2008 at 12:54:05AM +0200, Xavier Bachelot wrote:
> > Yeah, I am still patiently awaiting this. Discussions w/ VIA have
> > gotten me nowhere, as VIA's employees want me to acquire an MPEG LLA
> > license *before* I can get the code from them, at least last time
> > that I tried.

Which is exactly in conformance with the MPEG-LA patent license agreement as
available from http://www.mpegla.com/m2/m2-agreement.cfm

VIA produces an 'intermediate product' and is not responsible for paying the
royalties.  It is the entity that 'makes a consumer product' which is 'sold to
an end user'.

So if you are the final system integrator who turns VIA's chip plus some
software into something that actually performs operations that are covered by
the patents held by the MPEG-LA, you need to sign the license agreement with
MPEG-LA and pay the royalties as per schedule.  In this case VIA can be sure
that you pay the licenses and everything is fine.

> > Perhaps we'll see a shift away from this policy (a la Intel) with the
> >  new open-source releasing that is happening now.

Well, maybe Intel thinks it has sufficient patents itself so that the MPEG-LA
would not sue them, or if they did, Intel has the leverage to countersue.

But as long as you accept that patents are a business reality, and assume that
at least most or many of the MPEG-LA covered patents are real and enforcible,
then anyone, including VIA will have to bow to the license agreement for those
patents.

What I personally believe is the best way to solve the problem is to disclose
documentation on the codec acceleration hardware.  This documentation is
clearly not a consumer product and not sold to an end user.

Then the FOSS community or anyone else can write their hardware accelerated
decoder, and decide by themselves if they want to expose themselves to the
legal risk of violating those patents.

> I would not expect the new VIA driver to get any mpeg acceleration
> capacities in the near future. And they will probably not release any
> documentation on the Unichrome Pro II mpeg engine anytime soon, both
> because of the MPEG LA license. It is still unclear to me (and VIA) who
> should be paying this license fee. 

I think it is pretty clear who should be paying it.  But it just doesn't work
with software that is freely distributable/copyable, and for which nobody knows
how many copies are floating around and running on some systems.

> I do believe the silicon maker should pay, but VIA rather want to have the
> driver provider pay as it "enables" the feature. 

It is not so much about what any of us believes, but about what the MPEG-LA
licensing conditions say.  I would also prefer if all silicon makers would be
responsible for it.  But the industry standard practise (not only for mpeg
related patents) is different.   If you buy a GSM chipset and license its
firmware from TI (like Openmoko does), then you are still responsible of
licensing all the GSM related patents by yourself from the respective holders.

> At any rate, I don't think releasing the hardware documentation would
> infringe any license/patent, but VIA wants to stay on the safest possible
> side.

I agree with you, and I'm trying my best to make this change inside VIA.

-- 
- Harald Welte <laforge at gnumonks.org>           http://laforge.gnumonks.org/
============================================================================
"Privacy in residential applications is a desirable marketing option."
                                                  (ETSI EN 300 175-7 Ch. A6)
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://wiki.openchrome.org/pipermail/openchrome-users/attachments/20080911/7a347c37/attachment-0001.bin