Yousef, <Moving from 2.2V down to 1.8V will require
"major" changes to their FET "architecture" ...>
Really? Major? You must be a real top-notch Ph.D to
derive this. Let me, as a "screwdriver", tell you
something. A digital device either works on
specified workload in the specified temperature
range and voltage, or does not (some other
conditions apply). If it works at 10% below
that limit, it will work. Forever. Period.
If you can remember, old-fashion CMOS logical
devices worked at any voltage from 3V to 15V,
just the top frequency was specified somewhat
differently:)
Now go try an experiment.
Take a PC board with an analog secondary power supply,
attach a trimpot to a proper place, and use a small
screwdriver (don't try to use your brain power, it
will not work, use a screwdriver) to trim it, with a
voltmeter attached.
Got it? Now plug in any K6-266-2.2Vcore. You will be
amazed to find out that the device will work at
200MHz and 1.7V (even below that) very reliably:
all OS, benchmarks, burn-in, screen savers, etc.
Tell me now what "major" "architectural" changes
had happen during this experiment? Then get lost.
<If the demo K7 is at 1.8V, then AMD will claim
that this is their .18um process.>
What if not? Who cares what they would or would
not "claim"? Don't be ridiculous, you must
understand that the 0.25 or 0.18 "feature" size
is a purely marketing blurb, no serous technical
parameter stands behind this "feature". It is
very stretchable, and depends how you define it:)
Better give a straight answer, as in a deposition:
in your opinion, what variant of K7 processor AMD
is capable to demonstrate, in a simple
form <freq:voltage>. Actually, a range could also
suffice. No need to supplement your answer with
calculations of Leff and Idsat...
Please share with us your deep vision of current
technology. Don't be shy. All AMD investors are
looking at you, our technological leader! You
can make it!
|
