articles
FORUMS!
links
contact

PC Database
([an error occurred while processing this directive] entries)

Folding Team

SETI@HOME Team
RC5 Team
Genome Team

OCAU chat

Performance Comparison
The testbed was as follows:

MSI 8809 3D AGPhantom, a GeForce SDR card
Asus A7V socketA motherboard
AMD Duron 700MHz @ default speed and voltage
Enermax 350W PSU (with extra intake fan above CPU)
Miscellaneous other goodies wrapped up in a generic case with 80mm case-front fan

All heatsinks that came with mounting hardware to fit the 2 mounting holes near the GPU were tested using that hardware with Arctic Silver Thermal Paste from CoolPC. All heatsinks that came with a thermal pad or tape were tested with that. This means some units were tested twice.

The test consisted of 2 runs through MadOnion's 3DMark2000 v1.1 at default resolution, using the Hardware T&L of the GeForce to make sure the GPU was kept busy. These runs were taken after 15 minutes of playing NFS5: Porsche Unlimited, a 3D driving game, to heat the case, CPU and video card up - this was also to ensure the thermal pads had a little time to "bed in". The testing was conducted over 2 days but the ambient temperature during the testing was 27C - I confirmed this wasn't an issue by running some of the tests on both days and got identical results.

Temperature was recorded using an external thermal probe mounted on the back of the video card (facing up towards the CPU), reading from the PCB directly behind the GPU core. The recorded results are the temperature shown as 3DMark2000 dropped back to the results screen ("Final Temp C") and the peak temperature registered during the second 3DMark2000 run ("Peak Temp C"). On all units, the highest temperature always registered during the medium-detail "3D Game - Adventure" section, and the temp would slowly drop during the remainder of the run.

As usual, the results speak for themselves. The retail unit proves itself as woeful as you'd expect - a peak temperature of nearly 80C is frightening, especially considering it was tested with Arctic Silver thermal paste. The Tennmax is a little dissapointing but still a huge improvement when used with silver paste. The Blue Orb and FAA20 prove quite effective but the real surprise is the MOS Xtreme Cu. Despite the fact it's only using a thermal pad, it triumphs with the lowest peak and final temperatures. For a comparison, look at the 10.1C peak difference between the Blue Orb and the MOS Extreme Cu when both are using the provided thermal tape/pad.

I guess for sheer convenience the Blue Orb retail pack wins, being a pretty good cooler and having everything you could need included in the package, but in terms of sheer cooling grunt the MOS Cu Xtreme triumphs. Considering this is a cooler you could slap onto pretty much any component it fits over, thanks to the sticky pad - the ICS on the Asus P3V4X springs immediately to mind - and given you could drill some holes in the roomy heatsink base to mount it a bit more firmly (and hence cool more effectively) using the motherboard/video card holes, the MOS Xtreme Cu is an impressive unit. Further, I now realise that the base is quite tarnished or oxidised, beneath that dull surface lurks a more conventional shiny copper and with some lapping - the technique I describe here would be ideal - I'm sure even better performance could be unleashed.