By: Mark Roulo (nothanks.delete@this.xxx.com), August 15, 2013 3:31 pm
Room: Moderated Discussions
David Hess (davidwhess.delete@this.gmail.com) on August 15, 2013 2:38 pm wrote:
> Patrick Chase (patrickjchase.delete@this.gmail.com) on August 15, 2013 12:17 pm wrote:
> >
> > As I understand it (I am not a lawyer) there is plenty of case law establishing that while you can
> > copyright a brand name like "Pentium" or "IA64" or "MIPS", you can't copyright an architecture per
> > se. The only things that truly prevent somebody from implementing a given architecture (as opposed
> > to advertising the fact that they implemented it) are the patents. This legal principle recently
> > came up in the litigation over a certain well-known Java-like VM. Recall that in that case the judge
> > ruled that the architecture wasn't subject to copyright and set aside the jury verdict.
>
> A brand name would be protected by a trademark and not copyright. Intel used "Pentium" instead of
> some variation of 586 because their trademark applications for 486 and 586 (but not i586) were denied.
> Some numbers *are* trademarked like Levis 501 (jeans) and Boeing 747 but apparently that depends on
> them being rendered or used in a unique way for which Intel's chip numbers did not qualify.
>
> > With that in mind, MIPS is a very old architecture, enough so that many of the critical
> > patents have expired. This gives a would-be implementor like ICT/Godson 2 choices:
> >
> > 1. Implement an older, unencumbered version of the architecture and don't brand it as "MIPS".
> > This is what Godson originally did, and is why the first couple versions were missing some MIPS
> > instructions (DSP extensions if memory serves, though I can't find a reference offhand).
>
> I remember when this came up but the web articles discussing the details are gone
> or behind paywalls now. I think the patent in question had to do with instructions
> that handle misaligned loads and stores and affected 4 instructions:
>
> https://www.google.com/patents/US4814976
The Wikipedia article on Lexra talks about "four unaligned load and store (lwl, lwr, swl, swr) instructions" for the MIPS instruction set that were patent encumbered. I'd expect that the Godson/Longson folks would have run into the same trouble.
> Patrick Chase (patrickjchase.delete@this.gmail.com) on August 15, 2013 12:17 pm wrote:
> >
> > As I understand it (I am not a lawyer) there is plenty of case law establishing that while you can
> > copyright a brand name like "Pentium" or "IA64" or "MIPS", you can't copyright an architecture per
> > se. The only things that truly prevent somebody from implementing a given architecture (as opposed
> > to advertising the fact that they implemented it) are the patents. This legal principle recently
> > came up in the litigation over a certain well-known Java-like VM. Recall that in that case the judge
> > ruled that the architecture wasn't subject to copyright and set aside the jury verdict.
>
> A brand name would be protected by a trademark and not copyright. Intel used "Pentium" instead of
> some variation of 586 because their trademark applications for 486 and 586 (but not i586) were denied.
> Some numbers *are* trademarked like Levis 501 (jeans) and Boeing 747 but apparently that depends on
> them being rendered or used in a unique way for which Intel's chip numbers did not qualify.
>
> > With that in mind, MIPS is a very old architecture, enough so that many of the critical
> > patents have expired. This gives a would-be implementor like ICT/Godson 2 choices:
> >
> > 1. Implement an older, unencumbered version of the architecture and don't brand it as "MIPS".
> > This is what Godson originally did, and is why the first couple versions were missing some MIPS
> > instructions (DSP extensions if memory serves, though I can't find a reference offhand).
>
> I remember when this came up but the web articles discussing the details are gone
> or behind paywalls now. I think the patent in question had to do with instructions
> that handle misaligned loads and stores and affected 4 instructions:
>
> https://www.google.com/patents/US4814976
The Wikipedia article on Lexra talks about "four unaligned load and store (lwl, lwr, swl, swr) instructions" for the MIPS instruction set that were patent encumbered. I'd expect that the Godson/Longson folks would have run into the same trouble.