[Arm-netbook] FSF-Endorseable Processor

[jm]

On Wed, 2012-12-05 at 08:23 +0000, Simon Kenyon wrote:
> On 12/05/12 08:14, luke.leighton wrote:
> > On Wed, Dec 5, 2012 at 7:28 AM, Jacky Lau<i90091e at gmail.com>  wrote:
> >> As far as I know, the first MVP based SoC, ICube IC1 has been taped out in
> >> May last year. And ICube claimed that the first end-user product will
> >> present in market this summer. But I have not seen this product. Do you know
> >> something about the end-user product based on IC1?
> >
> >   not that i can tell, i'm under NDA on this aspect.  you'd need to get
> > icubecorp to answer, directly.  or i get permission to answer, from
> > them.
> >
> >> In ICube's roadmap, the
> >> next SoC IC2 will has 4 cores 16 threads, and employ 40nm technology. What's
> >> the difference between ICube IC2 and your SoC?
> >
> >   shush!  don't tell everyone it's icubecorp!  ok, basically, both we
> > and ICubeCorp have pretty much the same goal - i.e. their roadmap
> > happens to be coincidentally near-identical so we've been working with
> > them and steering them towards a specific set of I/O, if you know what
> > i mean.
> 
> you want $10m to give to icube to do what they really ought to be doing 
> anyway or might already be planning to do?
> 
> what is the business logic?
> 
> for this money do you get the right to buy the chip? or do you get a 
> certain quantity (like say $10m worth)? or do you get sole rights to it?


It would be good if ICube released their VHDL to show their intentions
are made of gold.

Open Cores have done it http://opencores.org/shop,items

Their OpenRISC processor SoC runs Linux.

[I bought their FPGA board to run the VHDL - but was not smart enough
to spot I needed the USB JTAG debugger to go with it so I can't comment
yet about the power of that SoC.]

It seems to me they got everything there needed including huge amounts
of open sourced IP needed to build SoCs. (They did
a failed kickstarter project, but still their boards got made and
nearly always they are out of stock these days.)

For sure, this CPU is not an arm, and so its not encumbered with
ARM IP; which makes it a lot easier to adapt and expand.

Also the software headers I was talking about earlier can be
more easily managed because you would be starting out with
a new CPU and not fighting the sea of 20 billion+ ARM chips
out there without header files and names for register and bit fields.