[sdiy] Parallel processing clock design question

[John L Marshall]

Magnus,

Take a look at what n-Cube is doing today. They used to build hypercubes
with 9000 chickens in parallel for DARPA.. www.ncube.com , serious
bandwidth.

Take care,
John
------------------------------------------------------------------
ROBERT RICH will speak at this meeting.

Pacific Northwest Synthesizer Meeting

DATE CHANGE: August 2, 2003

www.sound-photo.com
------------------------------------------------------------------

----- Original Message ----- 
From: "Magnus Danielson" <cfmd at swipnet.se>
To: <Stromeko at compuserve.de>
Cc: <synth-diy at dropmix.xs4all.nl>
Sent: Monday, July 14, 2003 2:40 PM
Subject: Re: [sdiy] Parallel processing clock design question


> From: ASSI <Stromeko at compuserve.de>
> Subject: Re: [sdiy] Parallel processing clock design question
> Date: Sun, 13 Jul 2003 23:11:02 +0200
>
> > On Saturday 12 July 2003 23:18, Glen wrote:
> > > 12,000?  With 12,000 uP chips, even the uP chips from my old Atari
> > > 800 could be used to make a fast computer.
> >
> > ... that could compute almost nothing of general interest. Would you
> > rather have four horses or 1024 chicken pulling your plow?
>
> A deer... a John Deer... or any other suitable tractor.
>
> But the 1024 chickens really seems as an interesting concept ;O)
>
> > > Hmmm...  I what would 12,000 of the new AMD 64 bit CPU's would be
> > > able to do?
> >
> > For starters, they'd pull about 1MA peak current from your 1.2V supply.
> > For more details, check project "Red Storm" (Sandia Labs and Cray,
> > 10368 Opterons, <2MW for power supply).
>
> A computer that isn't hooked up to a dedicated nuclear plant isn't worthy
of
> much attention ;O) Naturally DOE has the resources... ;O)
>
> Maybe it's time to rebuild R1 again so we can have a fatter computer at
the
> university...
>
> Cheers,
> Magnus - who have been having a party in the uni's former reactor hall