Do you have a question? Post it now! No Registration Necessary. Now with pictures!
June 6, 2005, 5:49 am
rate this thread
I'm interested in a Thermalright SP-94 to cool my CPU a bit more than
the standard socket 478 HSF. I see from Thermalright's website that the
recommended 92mm fan is 55dBA and pushes 110.18CFM. It does seem a bit
over the top, considering that the recommended 92mm fan for the SI-97
is 30dBA and pushes 48CFM.
Now I know that these are two completely different heatsink designs, and
on two completely different sockets, but is a 55dBA fan really necessary
to cool the average socket 478 Pentium? I don't intend on overclocking
the hell out of it, perhaps taking it from 2.6GHz to 2.8GHz at the most.
Do you think it would be fine to use the 30dBA 48CFM fan?
Re: Thermalright SP-94
Yes it's a bit extreme. However, you mention "cool... a bit
more", and merely swapping a 'sink but not using a
higher-flow fan, will of course have lesser benefit.
If on the other hand, you were content to have not much
better cooling but at lower noise levels, that's a different
Either way, you "could" get that fan (or something similar)
and undervolt it. Perhaps your board has the ability to do
this already? Either way, a 92mm x 32mm thick fan is a good
start for good flow/noise ratio, doing somthing as simple as
placing a 2W, 47Ohm resistor in series on the fan's power
lead may be (rough estimate of) something to try. The
higher powered the fan starts out, the smaller the resistor
value you'd want, generally speaking.
Then again, the average P4 stays cool enough with the stock
'sink, even with the modest o'c you mention below (2.6 to
2.8). Such a high-end heatsink is usually chosen for
1) Extreme overclocking
2) Noise reduction over what you presently use.
Of course there IS a 3rd option,
lowest noise possible with slight overclock, which fits what
you're doing nicely.
- Do you need to increase the voltage for your overclock?
That is, have you already tried it, and was it
- What is the present temp running at full load, stressed
by something like Prime95's Torture Test after it has ran
continuously for at least 30 minutes?
In general, most people could use the slower fan you
mentioned. Some people, myself included, would shoot for
even lower than 30db, would get the panaflo fan and
undervolt that a little too if/when possible. Certainly
some testing would be necessary to fine-tune it, but I'd
shoot for about 2200 RPM and see if the CPU stays stable
(running the Prime95 test mentioned above).
If you're willing to go the inline-resistor route, you can
always choose a fan faster than you think you'd need, and
simply throttle it back to an appropriate level. Same goes
for using a ready-made fan controller that allows fine
adjustments (typically those with the rotary knob and
potentiometer type construction).