With the Athlon CPUs getting ever faster and hotter a good cooling solution is more important today than ever. To give an overview about different designs this article will discuss 23 CPU coolers.
A few of these coolers are not meant for the Athlon 1.33 Ghz CPU used. They were included nonetheless to give you a picture of what happens if you either choose a undersized cooling solution, or transcend its limits by overclocking.
A Novelty in this test
Cooler tests are something you can find at every corner of the
internet. This is, as far as I know, the first test using the AMD
Athlon Palomino's integrated temperature measurement diode for rating
the performance of coolers. This has been done for Intel's Pentium
CPUs since they included on-die temperature measuring. With AMD
CPUs this was not possible up to now. But, with the availability
of the first mainboards by Fujitsu-Siemens
supporting this measurement and using a Palomino 1.33 Ghz CPU, we
now have the best measurement method for the actual die temperature.
Other measurements, be it measuring the heatsinks's temperature
or measuring at the sides of the CPU die, have to be inaccurate
due to the physical distance to the center of the die.
Keep this, plus the ambient temperature of 35°C in mind when comparing absolute temperatures in this test with the ones of other cooler reviews. The temperatures there inevitably will be a significant amount lower, but the ones given by this test are closer to the reality inside a PC housing.
Check out other cooler test nonetheless. Over at Tom's Hardware there is quite a good one, effectively comparing over 40 models. Some of the coolers tested there are reviewed here, too. Together with other tests around the internet this should give you a good idea of what designs are good - and which simply suck.
Why these coolers and not others? Why is XYZ's Supercooler not included ?
People asked that quite a few times. Well, because these were all the coolers available. As you might have noticed, there are NO advertisments on this site. This is purely a hobby and a "service" to the internet community, almost no money is to be invested. But if any distributors / retailers / manufacturers of all those fancy expensive high performance heatsinks would provide samples, these could be tested, too :-)
Let's see, as interest in this test is quite good (20.000 hits today) perhaps there will be a follow up, soon ...
CPU Cooler Design
What AMD has to say on this issue can be found at
document available from their tech
Note that AMD specifies the maximum allowable weight
at 300g. Some heatsinks exceed this ...
This is the standard design for heatsinks. Most coolers, cheap as well as more expensive ones, are variations on this theme. All the manufacturers do is take a reasonable sized aluminium block, machine it into a base with fins and fix some fan on top of this. Add a steel clip to hold it to the CPU base and voila - you have got a mass-produceable heatsink of the most unimaginative sort. How these products fare against more sophisticated ones we'll see ...
Radial Fins ("Orb" Design)
Thermaltake was one of the first to use this cool-locking design. Unfortunately, if you look at aluminium heatsinks like the one you see at the right, they only look cool - yet they don't keep your cpu very cool.
Folded Sheet Metal Fins
Several coolers for the Intel Pentium IV use this design. A sheet of metal is folded in a zig-zag pattern and either soldered or welded to the base.
This design seems to be quite hard to produce, as many heatsinks manufactured that way look slightly "crappy".
The advantage is that you get very thin "fins" and thus quite a big surface area. The disadvantage is that the fins and the base are not from one piece and thus have a bit of thermal resistance between each other, degrading performance.
This technique is very new. A single block of metal is shaped, presumably by sawing, to form a base with about 60 very thin fins. This yields a very big surface area, and thus good heat dissipation.
Questions ? Discuss this article in the Forum
Next Topic : Test Setup