Re: [Cfrg] intel license (was: Re: RG Last Call - draft-irtf-cfrg-ocb-00)

["Blumenthal, Uri - 0558 - MITLL" <[e-mail]>]

Of course Intel has a ton of patents around their CPU and instruction set,
granted - including this one for doing those weird multiplications. But
who in the right mind can imagine Intel suing a software developer (not a
chip cloner!) for using an instruction from the CPU instruction set? Let's
not be ridiculous, OK?
 
The _license_ (not a copyright) quoted by David is both clear enough and
permissive enough. FUD aside, it seems painfully obvious that anybody can
write code using Intel AESNI instructions (and other IA64 instructions
too, if it matters) without legal repercussions. On top of that, Intel
allows unrestricted use of their ("highly optimized") assembly code that
implements GCM and uses AESNI instructions.

AESNI-assisted GCM is fast enough for at least some of us, and free enough
for all of us to use. OCB has legal restrictions. Call me back in 2021?

P.S. If you plan to start a company (say, Outtel :) and clone an Intel CPU
chip - then you would probably hear from Intel lawyers, and this patent
#8,340,280 could be one of the many thrown at you. :-)
--
Regards,
Uri Blumenthal
<Disclaimer>




On 2/13/13 10:55 , "Ted Krovetz" <[e-mail]> wrote:

>
>> Intel has released a GCM implementation that uses PCLMULDQ and granted
>>its
>> use under a BSD-style license
>
>I believe this style of license is saying that you have permission to use
>the code -- thus you are not violating copyright by using it. But there
>is no explicit patent grant from Intel in the license.
>
>If you want to use a patented algorithm, you need both source code to
>execute and a grant to use the invention. This BSD-style license gives
>access to Intel's code, but is there a patent grant somewhere allowing us
>to use the invention too?
>
>Phil's patent grant is giving lots and lots of people free use of his
>invention, and my OCB source code has a BSD-style license attached,
>allowing free use of an implementation. You really need both to use
>something patented.
>
>-Ted
>_______________________________________________
>Cfrg mailing list
>[e-mail]
>http://www.irtf.org/mailman/listinfo/cfrg