Re: [oc] i386 legally

[Niclas Hedhman <niclas@hedhman.org>]

On Thursday 22 May 2003 04:05 pm, John Dalton wrote:

> If building, you will have to check whether Intel
> (and others?) have patented aspects of the i386 and these
> patents have not yet expired.  Perhaps try and find
> out whether AMD licensed Intel technology or
> did their own?  

AMD did license the 80386 technology, as did many others at the time. AMD 
claimed (I think it went to court), that the 80386 licensing agreement had 
extension clauses into next generation families. Intel did not grant them.
And AMD started their own.

80386 was released ~1987, so patents would be covered for another 8-10 years. 
Exactly what patents there are on the 80386 is hard to know without a search.

IANALE, but worst case scenario of a 'clean-room' implementation of a CPU, is 
that Intel has to go to court and prove that infringement of a patent is in 
place. And if they win, they can request money OR that the production stops. 
If you choose the latter, Intel receives no money, and only have cost on 
lawyers. I.e. It is fair to believe that Intel will not pursue this avenue, 
as long as Intel perceives that they don't loose revenue, or you are very 
successful and they think they should have a share of that success.

If you are interested, send an email to Intel and ask which patents are 
covering the 80386 design. Some or even all may have expired.

Niclas

P.S. I know that the instruction set mnemonics are always protected by Intel. 
Zilog had to change the syntax (but machine code compatible) for their Z80, 
which basically was an 8080 clone with an enhanced instruction set.


--
To unsubscribe from cores mailing list please visit http://www.opencores.org/mailinglists.shtml