cooling computer with refrigerator

[shegeek72]

Yes, emphasis on "adequate", not run-it-in-a-freezer, cold.
The goal is not to try to get temp as low as possible,
rather to use the time tested and proven method(s) to keep
the temp from getting too high.

Thanks for the info. As a hacker, I like to explore alternate ways to
do things.

 

[Technomage Hawke]

BTDTGTTS - well, almost. Our disco amplifier was constantly tripping out
on over-temperature until, one day, we got the idea of shoving it (less
mains transformer) in a fish-tank and filling the latter up with
distilled water. Back-lit, with a few plastic fish to show the thermal
effects, it was a great success.

If you try this, be warned. The end result has about as much alcohol
tolerance as I do - a pint of beer and it ceases to operate in any
useful manner...

distilled water is "ok".
However, mineral oil is much better (and with the addition of a proper
dessicant and filter, even condensation can be removed).

 

[Ken Maltby]

No, there is not. CPUs are based upon clock rate. If they
were able to operate faster, they won't until that clock
rate changes.

Hmmm.... I agree with most of your posing here on a practical basis,
but superconducting can be used to enable extreme increases in
performance of CPUs (Haven't read of any collage students super
cooling a GPU, not even in Japan.). It's not something you can
keep up for very long, unless you are NASA. So technically once
you reach a temp. that allows the chip to superconduct there is a
real performance increase.

Also a lessor cooling can allow you to up the voltage to the chip
without damaging it. At higher voltages the chip can be overclocked
higher than it can at its rated voltage.

For any practical purpose though "kony" has it exactly right.

Luck;
Ken

 

[kony]

Hmmm.... I agree with most of your posing here on a practical basis,
but superconducting can be used to enable extreme increases in
performance of CPUs (Haven't read of any collage students super
cooling a GPU, not even in Japan.).

Reread what I wrote, the contingency is "clock rate".
Supercooling will not help the tiniest bit to improve
performance if the clock rate remains unchanged.

If we want to open up to the idea of changing clock rate,
that's an entirely different can of worms. Even so, we'd
need isolation between extreme CPU cooling, and other
subcircuits susceptible to temp change.

It's not something you can
keep up for very long, unless you are NASA. So technically once
you reach a temp. that allows the chip to superconduct there is a
real performance increase.

In overly simplistic terms, a CPU is designed in a segmented
fashion such that when it has done a clock-tick's worth of
work, it can't begin that next tick's worth until the time
has come.

Also a lessor cooling can allow you to up the voltage to the chip
without damaging it. At higher voltages the chip can be overclocked
higher than it can at its rated voltage.

For any practical purpose though "kony" has it exactly right.

Yes, overclocking is the opposite side of the coin. We
still might see it as the cooling following the need though,
rather than the other way around. To get substantial
enough benefit from cooling to make the endeavor worthwhile,
it'd need be more than just an AC ducted to the case.

 

[Mxsmanic]

Hmmm.... I agree with most of your posing here on a practical basis,
but superconducting can be used to enable extreme increases in
performance of CPUs (Haven't read of any collage students super
cooling a GPU, not even in Japan.). It's not something you can
keep up for very long, unless you are NASA. So technically once
you reach a temp. that allows the chip to superconduct there is a
real performance increase.

A chip that isn't designed to be superconducting will fail if it is
forced into a superconducting state.