Intel thermal pad or Arctic Silver?

[Zink]

Can anyone tell me out of his experience if it is worth using arctic silver
thermal paste on a boxed intel Pentium4 cpu ?

 

[Zink]

They seem to have a fairly good setup.
In short, leave the thermal pad. You don't need the arctic >silver for a

P4.

Yeah I think so too, the thermal pad is actually an alluminium sheet wich
should be pretty good.

I would have to wonder why you would want to >overclock anyways?

Tried that once, sucked. I dont understand why some people are getting all
excited about putting 50 decibel fans inside their pc. What you gain in
performance, you lose in stability and durability.

 

[kony]

Can anyone tell me out of his experience if it is worth using arctic silver
thermal paste on a boxed intel Pentium4 cpu ?

It'd make removing the 'sink later, a bit easier, except that AS
requires more effort to clean off.

Thermally there's not near as much different on a P4 due to it having
the heat-spreader. If you already have a syringe of AS you might go
ahead and use it but I wouldn't buy some just for a P4.


Dave

 

[Wheat Muncher]

w_tom enlightened us all with these words of wisdom:

With each medium change is a sharp increase in thermal
resistance. The best thermal interface is CPU direct to
heatsink. Therefore better heatsinks machine their CPU
interface surface. Cheaper heatsinks have found it easier
(and more profitable) to, instead, apply a thermal pad. CPU
to thermal pad to heatsink means inferior thermal
conductivity. But too many computers builder's never bother
demanding a 'degree C per watt' specification.

Numbers would demonstrate the point. The most critical
parameter for a CPU heatsink fan assembly is 'degree C per
watt'. Any heatsink assembly that does not provide this
critical parameter is suspect.

Thermal compound improves thermal conductivity by replacing
air in microscopic holes with 'the grease'. Thermal compound
is superior to thermal pads or thermal tape because CPU is
almost completely in contact with the heatsink (assumes too
much thermal compound not applied). Thermal compound causes
so little increase in thermal conductivity that CPU
temperature is only reduced by single digit degrees.

If thermal compound results in greater thermal conductivity,
then heatsink quality is suspect or something may be wrong
with how heatsink is held onto CPU..

Almost all heat is transferred from center of CPU to
heatsink. Therefore thermal compound that is effective is
only in the center one half of the CPU. If too much thermal
compound is applied, then thermal compound squeezes out to the
outer half of a CPU. Even worse, if thermal compound gets out
into CPU pins, then intermittent CPU failures can result.

Examples of CPU to heatsink interfaces are demonstrated in
experiments in:
http://www.dansdata.com/goop.htm

This author also makes an important point:

Arctic Silver is hype. Just about any thermal compound is
equivalent to the so expensive and overhyped Arctic Silver 3.
AS3 like any thermal compound is applied only for single digit
temperature decreases. Advantage by any thermal compound is
insignificant to a CPU used in normal operation if heatsink
was properly machined up front - therefore no thermal pad or
tape necessary. Thermal compound may result in a minor
temperature improvement maybe permitting an overclocker to run
a CPU a touch faster. But thermal compound is not the major
improvement factor hyped by myths.

If heatsink was properly machined, then thermal compound
only causes single digit temperature improvement. However
thermal compound can demonstrate quality of that heatsink
surface. Too much temperature reduction when using thermal
compound says the heatsink was defective by manufacturer.
Which begs the question - did that manufacturer even bother to
provide a 'degree C per watt' parameter? Just looked at seven
different heatsinks manufacturers. Only one provided that
spec. Many heatsinks are sold on hype and myth - especially
when they forget to include that 'degree C per watt'
parameter.

Ummm, nope. I disagree, yet again Einstein. I have used many, many
compounds and arctic silver is by far the most effective when used
properly, and single digit decreases is a load of hooey. I have dropped
temps by 15 degrees or more, just by using AS with a good copper
heatsink. As for you're properly machined heatsinks- they don't exist. If
you have ever tried to machine metal (alloy or not) you would know that
perfection is nest to impossible, even when machined using the latest
technologies. That is why compounds were produced. In order to rectify
the small, natural abnormalaties present. Try using a lathe sometime, see
how perfect you get that pole. You'll see what I mean.

 

[Zink]

I have now tried the thermal pad on a pentium2.6 800fsb, and the peak
temperature was 51-52 degrees. On a pentium 2.6 533fsb with arctic silver,
the temperatures never got over 50°. The 800fsb system had also higher
motherboard temperatures at full load, around 40 degrees. The 533 fsb MB was
always at 36°.
These differences might be given by the 800fsb chipset, or software bugs in
the temperature reporting utility used (asus probe).
I do get the feeling that with arctic silver the cpu cools down faster,
however note that the thermal pad is already installed and saves alot of
time and acetone.

 

[kony]

I have now tried the thermal pad on a pentium2.6 800fsb, and the peak
temperature was 51-52 degrees. On a pentium 2.6 533fsb with arctic silver,
the temperatures never got over 50°. The 800fsb system had also higher
motherboard temperatures at full load, around 40 degrees. The 533 fsb MB was
always at 36°.
These differences might be given by the 800fsb chipset, or software bugs in
the temperature reporting utility used (asus probe).
I do get the feeling that with arctic silver the cpu cools down faster,
however note that the thermal pad is already installed and saves alot of
time and acetone.

It is good to see this followup... too often you see inquiries, but
the final result is never mentioned.

IIRC, the P4 on a 800MHz FSB will use a very slight bit more power,
not enough to even consider except in a very controlled, precise
experiment. Asus' PCProbe can/does report inaccurate temps, I have an
Asus Athlon system here that's confirmed to report 6C over actual
temps, though that means little as to how accurate it is on your
board.

Right now I have a stock of generic (but fairly good) silicone-zinc
thermal compound (Circuitworks I "think"), AS Alumina, and AS3... I
would use the generic compound on a P4, it makes negligible difference
but is still easy to clean compared to AS3 or the stock pad.


Dave