Re: Shrinking the IBM Roadrunner Supercomputer to desktop form factor - Wafer Scale Integration
- From: Quadibloc <jsavard@xxxxxxxxx>
- Date: Sat, 21 Jun 2008 21:19:42 -0700 (PDT)
On Jun 20, 6:06 am, Quadibloc <jsav...@xxxxxxxxx> wrote:
Of course, there are other ideas too: IBM recently announced progress
in forced-water cooling that would enable 3D stacking of chips, for
example.
One thing I've wanted to do was put the news item of a while back
about a 600 GHz transistor made from Indium Phosphide and Indium
Gallium Arsenide into perspective.
A 600 GHz transistor doesn't mean a 600 GHz clock cycle in a
microprocessor.
Transistors made with Silicon Germanium technology were demonstrated
by IBM which handled 200 GHz.
There was excitement over Indium Antimonide, because transistors made
from it could be placed on normal silicon chips; those transistors
would only be somewhat faster than normal silicon transistors, but
they would use much less power.
As I recall, current microprocessors use strained silicon technology,
which was claimed to be almost as good as silicon-germanium.
Speaking of Indium, dots of Indium used as solder allow direct bonding
of chips, turned face down, to other chips - this is called the "flip
chip" technique, which, of course, brings to mind DEC's trademark for
their transistor logic modules. So one could build a microprocessor
out of multiple Indium Phosphide/Indium Gallium Arsenide chips with a
few thousand transistors each, directly connected, without the
overhead of Sun's technique.
But it would be quite expensive, particularly if the best you could
get from it is a microprocessor with, say, a 12 GHz clock - three
times faster, because that's the speed advantage over the transistors
in conventional microprocessors.
John Savard
.
- Follow-Ups:
- Re: Shrinking the IBM Roadrunner Supercomputer to desktop form factor - Wafer Scale Integration
- From: Quadibloc
- Re: Shrinking the IBM Roadrunner Supercomputer to desktop form factor - Wafer Scale Integration
- From: Quadibloc
- Re: Shrinking the IBM Roadrunner Supercomputer to desktop form factor - Wafer Scale Integration
- From: Rob Warnock
- Re: Shrinking the IBM Roadrunner Supercomputer to desktop form factor - Wafer Scale Integration
- References:
- Shrinking the IBM Roadrunner Supercomputer to desktop form factor - Wafer Scale Integration
- From: iajzenszmi
- Re: Shrinking the IBM Roadrunner Supercomputer to desktop form factor - Wafer Scale Integration
- From: Quadibloc
- Re: Shrinking the IBM Roadrunner Supercomputer to desktop form factor - Wafer Scale Integration
- From: Stephen Fuld
- Re: Shrinking the IBM Roadrunner Supercomputer to desktop form factor - Wafer Scale Integration
- From: Quadibloc
- Re: Shrinking the IBM Roadrunner Supercomputer to desktop form factor - Wafer Scale Integration
- From: Stephen Fuld
- Re: Shrinking the IBM Roadrunner Supercomputer to desktop form factor - Wafer Scale Integration
- From: Quadibloc
- Re: Shrinking the IBM Roadrunner Supercomputer to desktop form factor - Wafer Scale Integration
- From: Stephen Fuld
- Re: Shrinking the IBM Roadrunner Supercomputer to desktop form factor - Wafer Scale Integration
- From: Andrew Reilly
- Re: Shrinking the IBM Roadrunner Supercomputer to desktop form factor - Wafer Scale Integration
- From: Quadibloc
- Shrinking the IBM Roadrunner Supercomputer to desktop form factor - Wafer Scale Integration
- Prev by Date: Re: Shrinking the IBM Roadrunner Supercomputer to desktop form factor - Wafer Scale Integration
- Next by Date: Re: Shrinking the IBM Roadrunner Supercomputer to desktop form factor - Wafer Scale Integration
- Previous by thread: Re: Shrinking the IBM Roadrunner Supercomputer to desktop form factor - Wafer Scale Integration
- Next by thread: Re: Shrinking the IBM Roadrunner Supercomputer to desktop form factor - Wafer Scale Integration
- Index(es):
Relevant Pages
|