[Arm-netbook] [review] SoC proposal

[lkcl luke]

On Thu, Feb 9, 2012 at 6:03 PM, Vladimir Pantelic <vladoman at gmail.com> wrote:

> right, so none of these ended up in a consumer facing device as an
> application processor

 correct.

>- which is what you are proposing and and for which
> you want to port/adapt all the existing (linux) software.

 yyup.

> I am quite sure
> that a mobile phone basestation has tremendous benefit from 200 cores,

 ah you've misunderstood.  that's a highly application-specific
example that massively benefits from massive parallelism.  i mentioned
it purely as one example.

> but what about OpenOffice or Firefox?

 that's why i specified a limit of 8 cores.

 i've spoken to the CEO of tensilica and he was both excited and
confident that their RISC core technology could not only do over 1.2
ghz in 28nm but also that it would do a decent job.  and they have the
SMP capabilities as well.

 he also stressed that it wasn't their core area of business i.e. it's
not something that they actively go and shout about.  it was damn hard
to even find out that they have SMP capabilities.

 but yes, they make enough money from their other business areas to
not _need_ to do a general-purpose 8-core SMP 28nm SoC.  it just so
happens that they're the absolute world leader in application-specific
versatile and adaptable RISC technology and DSP technology, such that
i have confidence that it'll make an absolute corker of a
general-purpose CPU.

 l.