"KWW" said:
Running an AMD XP+ 2600 (Barton, I believe) no my son's A7N8X-E, using
Arctic Silver for the heat Xfr paste, and wanted to switch from an
aftermarket (all copper), low-RPM fam cooler to the aluminum (with copper
contact point) cooler that came with the CPU because:
a) It feels to be about 1/3 the weight.
b) The mobo is on its side, so the cooler is hanging in space rather than
bearing down on the chip.
c) Neither cooler has tiedowns... they both rely on the socket for support.
d) My son is going to college 1/2 way across the country and I won't be
there to fix it if it breaks.
Problem is, according to ASUS probe, the CPU temperature with the (better)
heavier cooler is 44C Max. The temperature with the stock cooler is 47C
Max. This is in an A/C'd room, case closed, not the coolest one in the
house (ambient 87F), playing videos/checking email, running apps
simultaneously. (He is not a heavy gamer.)
So are we borrowing trouble with the less capable cooler or will the weight
difference probably make the reliability better in the long run?
TIA
KWW
47C is not a bad temperature. 65C is a relatively bad
temperature.
Performance is based on "deltas". In your example, ambient is
87F. This is equal to 31C. AMD says a well ventilated case will
have a case air (motherboard temperature) rise of no more than 7C.
That would bring the air to no more than 38C. So, that is how
you tell if the case has enough fans, by the delta from room
to case air.
Your processor is 54 watts max. For the processor, the power
times the thermal resistance, gives the temp rise or delta,
between case air and die temperature. I.e. A heatsink with a
thermal resistance of 0.25C/W would results in a CPU temp of
38C + 0.25*54 = 51.5C die temp
First off, for thermal benchmarking, it would help to have an app
that pushes the machine to 100% load. Prime95 (in "torture test"
mode) is one application people use to heat up the processor and
test for calculation stability (mersenne.org). CPUBurn was another
I've used in the past. (
http://users.ev1.net/~redelm but site
appears to be gone.) They'll give you an idea of what a good
computing load will do.
I can see a couple of approaches. One approach is to build a
battleship. This is an example of a product with a backing plate
and is screw mounted:
http://www.thermalright.com/a_page/main_product_index.htm
http://www.thermalright.com/a_page/main_product_sp97.htm (585 grams)
A couple of the products on the Thermalright page use the screw
holes in the board. There are four holes around the socket (or
rather, they are shown in the manual, but check the actual board,
as there have been Asus boards in the past, where not all board
revisions had their holes).
Using the screw holes means the clip won't get busted off in
transit. You can see the substantial nipples and spring loaded
screws at the top of this page:
http://silentpcreview.com/article132-page4.html
The only question I have about this technique, is when the PC is
being shipped, would a twisting or torquing action on the case,
cause increased pressure on the die. Any processor has a max
pressure that can be applied (the "normal force" or force normal
to the surface). While in one advert, they claimed they dropped
a PC equipped with a screw hole heatsink from a second story
window, and the heatsink stayed on the board, the pressure applied
to the die could be another matter. In the end, is there much
difference between a computer that arrives with a crushed die, or
one that arrives with the clips busted off ? I don't think I've
read an article that addresses the effects of shipping.
Rather than build a battleship, the other approach you could take,
is coaching your son on heatsink installation. For risk reduction,
the screw down approach might be safer than the clip approach,
depending on what the clips look like. You could ship the PC
without the heatsink fully installed, and he could complete the
job when the PC is uncrated at the destination.
This one also screws down, but it weighs a bit less. To use
this one, you need to check the compatibility list, plus as
well, if the processor socket is near the upper edge of the
motherboard, you'll need about 10mm clearance from the top
edge of the motherboard, to the PSU metal case. Not all cases
have enough room to meet the 10mm requirement.
http://www.zalman.co.kr/eng/product/view.asp?idx=54&code=005
http://www.zalman.co.kr/Upload/product/CNPS7000A(S)_en3010.pdf
This one uses spring loaded screws, but still uses the tabs on
the socket for support. This one would be best installed at the
destination.
http://www.swiftnets.com/products/mcx462-V.asp
(Note: The products that don't come with a fan, you may be able
to recycle a fan from one of the HSFs you already have.
Otherwise, a Panaflo FBA08A12L1A is quoted as a quiet fan - but
it doesn't have a tacho signal. Model number FBA08A12M1BX
runs at higher speed and is louder, but has a tacho signal.
A Fanmate or similar device can be used to adjust the speed of
the fan, to reduce the noise. Digikey carries Panaflo...
http://dkc3.digikey.com/PDF/T042/1225.pdf )
So, I'd repeat your measurements, with Prime95, or some other
100% load application. Record your measurements as "deltas",
for comparison with the thermal performance when the computer
arrives, or at some later date. Then decide which heatsink
you like, whether purchasing another one is required, and
whether you son has the dexterity to install at the
destination.
If the computer is travelling sitting on the back seat of
a car, that might not be so bad. Probably any product, clip
or screw mounted would be safe like that. I drove a computer
1500km that way, and arrived with no damage. A car seat is
soft enough, that shock is pretty mild. If shipping by
courier I'd be more concerned about it, and would strip
some of the more sensitive guts from it before shipping.
(Maybe I'm just paranoid, or I hate UPS, maybe both
HTH,
Paul
