X-bit labs

Battle of the Titans: Super Coolers from Scythe, Thermaltake and Zalman Face to Face (page 2)

Category: Coolers

by Doors4ever

[ 10/21/2005 | 11:44 AM ]

Testbed and Methods

Before checking the coolers in action we must first choose the testbed configuration and the testing methodology. There can’t be any argument about the CPU – Intel’s Prescott-core processors are “unrivalled” in terms of power consumption and heat dissipation. Our tests show that an Intel Pentium 4 processor overclocked to 4GHz without increasing its default voltage consumes as much as 130 watts under load (for details see our article called FSP BlueStorm AX500-A Power Supply Unit Review)! This number will grow up much more at overclocking and at a higher voltage since power consumption is in a squared relationship with voltage.

So, I took an Intel Pentium 4 521 (2.8GHz, 1MB L2 cache, Prescott E0 core) for the tests – this processor can work at 4.06GHz at its default 1.35V voltage and at 4.2GHz at 1.425V voltage. The choice of the processor necessitated the choice of the rest of the testbed configuration which included:

• Asus P5WD2 Premium (i955X) rev. 1.02 mainboard, BIOS 0422
• Intel Pentium 4 521 CPU (overclocked from its default 2.8GHz to 4.06GHz frequency)
• Corsair TWIN2X1024-8000UL memory
• Matrox Millennium PCI graphics card
• Western Digital Raptor WD740GD hard disk drive
• SilverStone Zeus ST65ZF power supply (650W)
• Zalman thermal paste
• Windows XP with Service Pack 2

I took a Matrox Millennium graphics card not only because I’m not going to run any graphical benchmarks but because it doesn’t carry an active cooling system. This graphics card is absolutely silent, giving me the opportunity to better estimate noise produced by the tested CPU coolers.

In fact, the SilverStone Zeus ST65ZF power supply proved to be the loudest component of the testbed. This powerful PSU is targeted at modern top-end computers that include a voracious processor and two top-end graphics cards at once. Unfortunately, the fan of this power supply becomes rather loud soon after you turn it on, even if the system includes a non-overclocked processor and a weak graphics card like Matrox Millennium. This is no doubt a first-rate power supply that ideally suits for tests and experiments and for setting various records, but I wouldn’t put it into my own home computer.

The next problem to be discussed is the method of testing coolers. There are two popular approaches, in an open testbed or in a closed system case, and each approach has its pros and cons. Tests in a system case seem to be closer to reality and more objective as concerns comparing different coolers. True, few people use their computers in a dismantled state while most users usually put the components into a closed system case. This fact is considered by the cooler developers when they design new models. For example, coolers with a vertically positioned fan which have become popular exactly after the transition to heat pipes, drive the stream of hot air right towards the exhaust fan on the system case’s rear panel. This simple trick helps to reduce the temperature in the case and improve the cooling efficiency. It means that the full potential of such coolers as Thermaltake Sonic Tower will only be obvious in a closed system case.

This orientation of the cooler has its downside, though. It’s no secret that high power consumption of current processors makes the mainboard’s components in the CPU power circuit heat up much. I know cases when the MOSFETs literally melted on the mainboard due to overheat. The leading mainboard manufacturers long noticed this problem and came up with various methods of passive and active cooling of the MOSFETs (Active MOS from MSI, OTES from Abit, or ASUS Fanless Design). From this point of view, coolers with the traditional horizontal positioning of the fan, like Thermaltake Big Typhoon, look preferable since they also blow at the mainboard’s elements near the CPU socket.