SocketA Cooler Roundup
11-May-2001 - Article by: Wolfy

Overclockers Australia have assembled and tested 21 heatsinks rated for use on an AMD Socket A platform - many were tested with different fans, so in total we have 27 sets of results.

This article is intended to be a review and comparison, testing heatsinks as you buy them and how you would expect them to perform in a real system. I have not set up a fake test bench or modified my computer in any special way for this review. The coolers were all tested in the same machine, under identical conditions and the results recorded. This also means that the motherboard's in-socket thermal sensor was used to measure the 'CPU temperatures'.

I believe that using these methods will produce a fair and valid set of results, results that can easily be replicated in anyone's system. Also in line with this, each of the heatsinks were tested as they arrived in their 'stock' forms - that is - exactly how they come out of the box when you buy them from a shop. The only modifications that were made was the removal of any thermal paste, or thermal pads that were pre-applied.

These results are indicative of the coolers in stock form. Obviously using more powerful fans or modifying the cooler in other ways will produce better results … but more on that another time.

Testing Hardware
The test machine consists of the following hardware:
AMD Duron 600 CPU - running at 1.9V to increase the heat output
Asus A7V motherboard
256M Ram module, mounted in the last RAM slot.
Leadtek TNT2 S320 II AGP Video Card
SoundBlaster Live!
Adaptec AIC 7850 SCSI Card
Intel Pro/100+ Management Adapter Network Card
Seagate 27Gb ATA66 HDD
2 x IBM 4.5G SCSI HDD
Teac 532S SCSI CD drive
Teac R55S SCSI CD burner
Mitsubishi FDD


The test system.

The test machine is housed in a modified HX08 full tower case. The case has an 80mm intake fan mounted in front of the drive bays under, and a 120mm intake fan mounted at the lower front. Another 120mm fan is mounted below the power supply to draw air out of the case, and the position of the powersupply has been modified so that it sits above the rear fan. There is also an 80mm exhaust fan at the top of the case. The case has fairly good airflow, due to the number of fans, but the cables are a little messy at the moment.


Modified HX08 test machine.

Testing Methods
While I do have a Senfu thermal probe, I decided not to use it for measuring the resulting temperatures for each of the coolers tested. I noticed that the thermal probe can be quite touchy with its results, and it would be hard to mount in a consistent way without impacting adversely on the results, as each cooler would require a different mounting method. The Senfu probe did make an excellent thermometer to measure the ambient room temperature, which was always between 24.5 and 25.5 Degrees Celsius when testing was conducted. If the ambient temperature was out of this range testing was stopped.

The test results were taken from the onboard sensor located under the CPU, and while this is not the actual CPU temperature it is a good indication of the performance of each of the coolers tested. And because this is a comparison of the different coolers, then any bias of results will be consistent between all of the test units. Between each test, the machine was turned off and the heatsink installed. The machine was then turned on and the idle temperature measured and recorded after 15 minutes. Seti @ Home and Prime 95 where then both run simultaneously to apply 'full' load to the CPU, and the results were measured and recorded after 1 hour. The results were recorded using MotherBoardMonitor 4.17 and logged to the hard drive.


The tortured Duron 600 and CPU socket after testing.

Each cooler was tested with a small amount of Arctic Silver thermal paste (for small contact areas) applied between the CPU and the heatsink. The various 'Orb' fans were run directly from the motherboard headers, but all other fans used a 3-pin-to-Molex connector and took power directly from the power supply.

Each of the heatsinks and the CPU were cleaned with De-Solv-it citrus "sticky spot and stain remover".


Cleans Arctic Silver no worries.

I'd also like to take a moment to thank the various vendors and manufacturers who provided coolers for this roundup. I remember when custom coolers were an extreme novelty item, but now there are many many places where you can order modern high-performance coolers online and get them delivered in Australia. Thanks to (alphabetical order):

Below-0 - www.below-0.com
Computer Alliance - www.computeralliance.com.au
CoolPC - www.coolpc.com.au
EYO Technologies - www.eyo.com.au
KingMax Australia - www.kingmax.com.au
TRIG Engineering - www.trigonline.com

Vantec - www.vantecusa.com

Thanks also to fellow review site OverWear.net who sent over some of their cooler stash. :)

NEXT Page - Thermaltake Orbs