[MachineForth] Chuck's 25x and IBM's Blue Gene

[Waynesma <waynesma@xxxxxxxxxxxx>]

 --- Jeff Fox <fox@xxxxxxxxxxxxxxxxxxx> wrote: >
Waynesma wrote:
> > Still keep in mind:
> > That 400Mhz-1Ghz cheap edram.  It might be of
> interest
> > to investrors in a future design.
> 
> I have little idea any more what investors want
> they seem to want tried and true, proven, technology
> in wide use and don't want to risk being first at
> anything.

Well this technology is meant to replace the current
widely used technologies with many advantages.  Unless
something better upstages it I think they definitely
are going to succeed due to amrket demand.

> 
> They have 400Mhz in 0.15u and 600Mhz to 1G in 0.13u 
> while Chuck is saying 1G at 1.8u so he is still an
> order of magnitude ahead because of his improved
> transistor model and superior layout technique.
> That is transistor level.  At the chip level his
> approach and language ideas translate into more
> compact elegant code and much smaller nodes with
> much fewer transitors.

True, but it offers attreactive cheap high
density/performance memory options for high end
investors who want cheap large megabyte memory on
chip.  But until somebody with money and access to the
technologuy comes it wouldn't cost much as an option.

> 
> > IBM's 210Ghz silicon transistor process.
> > http://www.ibm.com/news/2001/06/25.phtml
> 
> To get 2400 MIPs Chuck must have transistors
> of at least 25G.  Going from 1.8u to 0.13u
> alone should yield 200G without any exotic

I'm just mentioning it in passing (as IBM won't even
use it commercially as common technology itself for a
few years), but it would be interesting to see what
they are doing and what Chuck could do with it
instead.

I still think that the x25, extremely good (of course
we may not even see this for a while).  A x25 with a
P64 (sorry still want monolithic Linux type OS) as a
front end processor (with Floating point/DSP/3D
Graphic routines in X25 ROMs) would give almost any
desktop a run for it's money.  Has Chuck ever
considered approaching some of the minor (hopefull)
players in the handheld video game industry, there has
been a number out there wanting to do systems?  An
x-25 with P32 front end would be great for a handheld,
and cost less money than a strong Arm.  Some 3D
systems and VOS's scale beutifully when extra
Parrallel processign power is added.  A module system
could be made to accept additional proccessors to
increase graphical performane, for 10 dollers the user
would stack x-25's.  Really, with the right functional
mix the X25 style chip array would make a greatly
competitive console that could emulate games machines
or run virtual code.  At the risk of promoting
competitors, the Toas based Amiga VOS is good and
should have much software shortly, and a port ot the
x25 would allow the x25 to tap into that software.  

> > Maybe he could become involved in some alternative
> > technology, say Optical computing.
> 
> He has talked with folks with exotic technology 
> with 1000000 Ghz transistors who wanted to see
> what his designs would do on that.  100M Mips per
> node with 3D fabrication and no heat issues?
> But Chuck is still working with commonly available
> silicon fabrication in 1.8u.

Thats a shame, so they aren't interested in hiring
him.  It would be a crowning acheivement in the
computer industry.

> 
> > > Chuck Moore's law seems different than
> > > Gordon Moore's law.
> > 
> > Out of curiosity, are they somehow distantly
> related ?
> 
> I don't think so. Chuck might have mentioned it
> or gotten better funding.  

:)

Thanks

Wayne.


_____________________________________________________________________________
http://messenger.yahoo.com.au - Yahoo! Messenger
- Voice chat, mail alerts, stock quotes and favourite news and lots more!
------------------------

To Unsubscribe from this list, send mail to Mdaemon@xxxxxxxxxxxxxxxxxx with:
unsubscribe MachineForth
as the first and only line within the message body
Problems   -   List-Admin@xxxxxxxxxxxxxxxxxx
Main Machine Forth site   -   http://www.