> Nvidia have 2 options:
> 1) Take an existing ARM core and tweak it to run faster.
> 2) Get an Architecture license and design their own core.

I think it has already been established from the start that nVidia is going for option #2 with "Denver". What isn't clear yet is what still to be revealed ARM ISA version they'll base the chip on.

> Hope that answers that one.

Depends on what (you think) the question was about, actually ;-)

> At best they'll be able to look at Linux code

Or even better from a legal point of view: *BSD code :-)

> For an ARM port they can get the docs for the ARM architecture
> by going to ARM, filling in a form and downloading them. That
> won't give them the full docs to whatever chip they target but it's
> better than reverse engineering.

How is having an ARM ISA version's documentation and trying to port an OS to an ARM SoC with a core based on that ARM ISA version any better than having a Power ISA version's documentation and trying to port an OS to a machine with a discrete PPC CPU based on that Power ISA version? You won't even need to fill in a form to get whatever Power ISA version's docs you want:

http://www.power.org/resources/downloads/

So I don't think there's much reverse engineering involved in porting MorphOS to the PPC970 CPU.

> As for 64 bit, scroll to the bottom of this page.

Thanks for this insightful link. To quote the relevant part for lazy bums: "Hitherto, we've decided it's not been sensible to have 64-bit programs. Extended memory addressing at 40 bits is in the latest Cortex-A15 ... but we haven't had the need for a 64-bit [arithmetic logic unit]."

Seeing how the tenses have been used in this statement I believe it can't be ruled out that a yet to be revealed ARM ISA (which "Denver" is supposed to be based on) will be 64-bit. Any objections? Not being a native English speaker I could have interpreted the specific usage of tenses in a wrong way.