On Wed, 23 Apr 2008 16:05:02 +0100, Franklin
<never.he@rd.of.it> wrote:
I would have to disagree with this, nowadays more than ever
CPU 'sink fans are larger and lower RPM, it used to be, only
a single-digit # of years ago, an 80x25mm fan was considered
quite large for a heatsink. Today 92x25mm fans are found on
the low-end budget heatsinks and 120mm, sometimes even a
pair of them on higher end heatsinks - these larger fans not
turning very fast at all, modern decent heatsinks with heat
pipes can keep a 100W CPU cool enough without such fans even
spinning at 1000 RPM.
It is better for what electronics? Fans are not
specifically designed only to run at 12.0V, they are merely
rated at that voltage because it's a common one to provide a
context for other ratings such as current or RPM. A full
proper fan spec sheet lists the voltage range any particular
model should run at, typically a range for a fan briefly
listed as a 12.0V fan would be about 6V-14V. The range gets
even larger for 24V fans, and of course these ranges are
only what the fan manufacturer guarantees, many people find
the actual range a bit larger.
A machine baking is usually due to fan failure or not
cleaning out dust periodically. We can't control the user's
environment or lack of proper maintenance when it comes to
cleaning, but when it comes to fan failure the slower a fan
spins, instead of at full speed the whole time, the longer
it's lifespan. Further the slower it spins the slower the
dust accumulation.
Throttling back fans helps prevent a situation where one
bakes their machine, provided they do as any other situation
requires - pay attention to the details in implementation
and check the result instead of just assuming one can slow
down fans as much as possible without bothering to check the
resultant temperatures.
Indeed, though there is one other risk - relying on software
to control system cooling also relies on a weaker link, the
operating system on which it runs. Pure (discrete)
hardware solutions aren't vulnerable to a system instable
from motherboard failure, CPU errors crashing the program,
windows itself crashing from one of many causes ranging from
OS bug to malware to driver bugs to ...
The safest strategy is a cooling solution that doesn't rely
on anything else working properly (except of course power
getting to it within a safe voltage range), a self contained
solution that would increase fan speed even if the rest of
the system is totally locked up and inable to do anything
but create heat.