Heatsinks

[clocker]

After seemingly endless fan configuration changes in my new case, and , believe me, I changed just about everything possible-number, location, blowing in/blowing out, I had hit a brick wall concerning my chip temp.
Basically nothing made much of a difference.
Using a Zalman 7000cu Flower heatsink with a 92mm fan, my temps stayed very consistent at 40-42 degrees C.
The good news was that the temp variation between idle and load was minimal-approx. 2 degrees.
The bad news was that I couldn't get any lower.

While trying to straighten out my workbench, I happened upon my original OEM AMD heatsink.
Not terribly sexy, just a aluminum sink, not too many fins, a dinky 60mm fan, I had discarded this piece when I got my new motherboard as not being "good enough" for my new , cool setup.
I also just happened to have a 60 to 80mm fan adaptor laying about and several unused fans.
Hmmmm, I said.

It's now installed and the first run of burn-in/ stability tests are over.
These results are very preliminary...I understand that Arctic Silver thermal paste takes some time to set up and reach it's optimum results, but....

The AMD actually works better than the high dollar Zalman.
So far, by two degrees.
I will need to run much longer and monitor the results over a period of time, but I feel very safe in saying that even if it isn't substantially better, the stock AMD HS with a 80mm fan is at least as good as the aftermarket Zalman.

I am surprised ( and somewhat dismayed ).

Until I have more test results to verify my initial findings I hestitate to make a steadfast assertion in this matter, but for all of you thinking of a new HSF combo, right now my experience is that the AMD is not a shoddy performer at all, and it's hard to beat the price.

I was rereading several reviews last night and I was struck by the fact that many of the sites that review HSFs do so by artificial test methods. They use a regulated heat source ( which must give them very repeatable test parameters) and the set up is not in a case.
I thought that I had done my homework prior to buying the Zalman, it was getting great reviews and was supposedly very quiet.
My real world results were different.
My Asus board has a featured they call SmartFan which theoretically varies the fan speed in accordace to temp loads.
With the Zalman I can't really say that it ever slowed below 2500rpm. With it's 92mm fan this made the Zalman , while not obnoxious, not exactly quiet, either.
If I disconnected the fan from the board and hooked to a 5v power lead the noise went away, but the temps jumped almost to 50 degC.
Not acceptable.

I haven't tried this with the new setup yet.
The noise level seems to be about the same, it's hard to really tell cause the pitch is somewhat different.
I also have two other brands of fan that I can try.

I can't say about Intel heatsinks, but I would advise you to think twice before you chuck your AMD unit for something cooler looking and certainly more expensive.

You may be disappointed.
I was.

[Kunal]

Clocker i have to disgree with you here slighly.

With the stock heatsink and a decent 80mm, on my 2000+, i was getting temps about 60c! and with the aero 7 lite i had a few hours ago on my setup, i was getting temps around 44c and now i have the volcano 12 my temps are now around 40c, but of course all of these fans run at a similar rpm (not much more than 2500rpm as the noise level starts to increase)

I do however say copper heatsinks are not as good as they seem. i havent had much time to do some fiddling with my new setup yet, but do beleive the heatsink on my aero 7 lite (high quality alu heatsink with copper insert) is better than the high density 66 fin one on my volcano, the only thing that makes a big difference is the fan, the aero ( a case of fashion over function) fan disapointed me lots, i could get similar temps with a standard 80mm on the same heatsink.

I will tell you my full results soon, just let my coursework load come down.

[clocker]

Kunal,

My results are only directly applicable to my case/fan combination and the Zalman heatsink.
I completely expect that your results ( and other folks, too) may vary.

That said, how much research did you do, how much data have you gathered concerning your original HS and the new ones?
How many burn in tests?
Were conditions the same for both setups?

I am currently monitoring temps for CPU. mainboard, HDD, PSU and ambient internal case.

Still, it will be interesting to see what you discover.
Please post results.

[Virtualbody1234]

Hey clocker, I'm surprised by your results with the Zalman. I'm mostly surprised by the fact that you say it was noisy.

while not obnoxious, not exactly quiet, either.
I would have thought that was Zalman's specialty (quietness).

As you already know, I have a stock 'all aluminium' AMD 'microfin' heatsink. I also have my fan adaptor to fit an Antec 80mm - 2600 RPM fan. I'm very happy with it at about 45C and very quiet. I didn't expect that my setup could come even close to that Zalman but I am surprised. :blink:

Microfin: http://www.frostytech.com/articleimages/oczdomin_detail2.jpg http://www.frostytech.com/articleimages/oczdomin_nofan.jpg

[clocker]

Well VB, as I said , the Zalman was silent, but not very effective at 5v. That worked out to be right around 1400 rpm.

My AMD HS is not the "microfin" version.
It is much cruder appearing, as if carved out by beavers.

Maybe I should look for one like yours.

I too, am surprised by the Zalman/AMD comparison.
Even more surprising, as I took my afternoon nap, the temps dropped to 38 degC.
Only now, after 30-45 minutes of (light) use have they risen up to 40c.
This is a behaviour NEVER displayed by the Zalman...it would stay constant, but not drop.

Weird.

[clocker]

VBod,

If I may, what chip are you using?
What is the frequency and the vcore voltage?

Just trying to make a comparison here.

Edit:
Does your fan blow into the heatsink or pull air out?
Have you ever tried the reverse of what you are running now?

[lynx]

I'm using a XP2400+ (thoroughbred core) standard clock.
HSF is Akasa 824Cu (http://www.cclcomputers.co.uk/specs/cooler/akasa/ak824cu-bl/spec.htm) (I'll get round to removing the silly led's eventually) and the copper base and alu top seems to work a treat.
Average fan speed is about 1900rpm (1600rpm idle) and cpu temp never goes above 37C.

It seems near ideal so I'm not going to mess about with it, but perhaps it gives you some pointers.

[Virtualbody1234]

I have the 2400+ thoroughbred.

266MHz FSB.
Running at stock 2000 MHz
Core voltage is 1.65 V.

The fan is blowing towards the heatsink.

Yes, I have tried to run it the other way. The temps were higher. I have been intending to try it reversed and adding a duct to direct the air right out the rear of the case. My theory being to get rid of the heat directly from the CPU and out the of the case, lowering the case temp and therefore all temps.

http://server4.uploadit.org/files/191103-hms.jpg

[clocker]

Lynx, those are impressive numbers.
I'll have to look into that HS.

Virtual,
I like your idea. Is your case wide enough to accomodate a 90 degree bend from the hs fan? With my current 60-80mm adaptor, I don't.

[Virtualbody1234]

Is your case wide enough to accomodate a 90 degree bend from the hs fan?
Yes, but I would have to build the duct myself. I need to make it wider than the 80mm but not as thick depth wise.

[clocker]

Lynx,

I have been reading up on your Akasa HS (here and (http://www.hexaplague-hardware.com/reviews/akasa/ak824cu/akasa-ak824cu-page1.php) here are typical (http://www.thinkcomputers.org/beta/home.php?x=reviews&id=99&page=1)).

No offense, but nothing that I've seen would lead me to expect the very good results that you are reporting.
Have you made any other modifications that may be partially responsible?

Never exceeding 37 degC is a goal I would be very happy to achieve.

[lynx]

It seems strange that in the two links you gave, the first has results similar to mine and all that's been done is to remove the thermal pad, the second has poor results but he's taken all that trouble lapping the base etc.

My previous cpu (XP1700+) ran a lot hotter (around 45C) and I haven't made any other changes except possibly to improve the airflow through the case by tidying the cables a little.

Ambient temp is about 20C, and the way I've got my case positioned means that exhaust air can only go up the wall behind the desk and therefore can't be drawn back in.

Obviously there are differences between temperature sensors, so the absolute temp shown is a little subjective without verification from another source, so 37C shown on my pc could well be a few degrees higher on another machine, but it should not be dramatically different.

[clocker]

Lynx,

If 37c is your max observed temp, is your idle temp significantly lower?

I know what you mean about the veracity of case temp monitors...I'm sure that mine is not accurate, but I only use it to make relative comparisons , so it's fine...

[lynx]

I'm using speedfan to control the fan, so my idle temp is about 34C, with my desired temp set at 35C. I've set it so that minimum duty cycle for the fan is 5% which gives me a min fan speed of about 1800rpm. If I run the fan at full speed the idle temp drops to about 31C.

[SciManAl]

hmm clocker i am actually not impressed with the zalman (sp) it seemed to be in effiecent (my opinion), the fins were far away from the actual die, it seemed as though the heat had quite a ways way to travel... however ther are defintly more hard core HS (without fan) with really low TR (thermal resistance) this is what i think you should aim in the future is to try a really low TR HS and see what happens my water set up has a .24 TR and if you could get that or better in a HS then you would be really cool i would think...

the better the TR is the better the heat is actually gotten rid of... (how much the thing obsorbs versus passing along into the air...)

examples:

a bit more expensive brand that has very low TR (http://www.frozencpu.com/cgi-bin/frozencpu/cpu-swi-08.html)
Same one a very intresting idea... (http://www.frozencpu.com/cgi-bin/frozencpu/cpu-swi-06.html)
good example of micro fins (http://www.frozencpu.com/cgi-bin/frozencpu/cpu-tri-06.html)

just a few to show, i think you won't get that much better results though with air cooling, i think you are reaching perfection...
(no joke i think you are getting to your peak with Air cooling, and the result is not bad at all)

[3RA1N1AC]

Originally posted by clocker@19 November 2003 - 13:30
The AMD actually works better than the high dollar Zalman.
So far, by two degrees.

...

I am surprised ( and somewhat dismayed ).
quite a while back, someone was asking on this board about which athlon heatsink to buy, and i advised that the stock HSF that comes with the retail chips is perfectly fine and, combined with a good thermal paste properly applied, it'll keep the CPU in the lower 40 degrees C. maybe not quite as cool or quiet as some other aftermarket ones, but it's definitely not a slouch. the guy just didn't want to believe it. :huh:

as for whether it really is better or worse than that Zalman HSF, there's always the random affect of each individual application of thermal paste. sometimes you get it right, sometimes you get it "less right." but i'd say the very high 30s or low 40s is about as low as any HSF combo is going to go with athlon XP chips, without some other extreme solution appended to it (like an air-conditioned case).