Testing:
Testing consisted of monitoring the CPU's thermal output while at idle and under a heavy load. Idle conditions were established by allowing the test system to sit at the CentOS 4.3 x64 desktop for a period of no less than one hour. The load conditions were generated by running two instances of Folding@Home, configured to take advantage of the dual cores on the Intel D-840 processor, for a period of 2 hours.
A Spire Digipanel temperature monitor was used to monitor the processor and ambient temperatures, while the System Monitor in CentOS 4.3 x64 was used to confirm that both cores were either at idle or at full load. Idle conditions saw the cores bounce between 0 and 2% activity, while load conditions had both cores pegged at 100% for the duration.
For comparison purposes, the Zalman CNPS8000 was tested head-to-head with the stock cooler that shipped with the Intel D-840, as well as a handful of other heat pipe coolers; the Sunbeamtech Tuniq Tower 120, the Scythe Mine and the Gigabyte G-Power Pro.
The image below shows just how compact the CNPS8000 is in comparison to some of these other heatpipe coolers...
The chart below summarizes all of the results recorded while the ambient temperature was maintained between 19 and 20 degrees Celsius. The Zalman CNPS8000, the Gigabyte G-Power Pro, and the Tuniq Tower 120 both come with fan speed controllers, so each was tested at the maximum and minimum speeds. The Scythe Mine and Intel stock cooler do not come with fan controllers so they were run at their only speed.
As you can see by the data above, running the CNPS8000 at its lowest speed is not a good idea. 59 C is just too high for my comfort, and I think the Zalman Fanmate 2 could lead to problems with people who do not monitor their temperatures closely. They may like the low noise output provided at the lowest speed without knowing that their processor was getting dangerously hot and perhaps entering the throttling stage (if applicable).
Running the CNPS8000 at its maximum speed provided much better results, and it was able to provide temperatures comparable with some of the other coolers, and eight degrees less than the stock Intel cooler while fully loaded. Considering the size difference with many of the other heatpipe contenders, the CNPS8000 did well.
Even at full speed, the noise output was not that bad. A full speed G-Power Pro is unbearable, a full speed Tuniq Tower is noticeable but not too bad, and a full speed Zalman CNPS8000 is the quietest of the three. The Scythe Mine only runs at one speed, and it was still quieter than the CNPS8000 at either minimum or maximum speed, without the extreme spike in load temperatures.
As I feared, with the processor heated up and the fan blowing, the body odor stench was unreal and gave me a crushing headache. It immediately filled my office with one of the worst smells possible, and even a full 10 days since it was removed and sealed in a ZipLoc bag there is still a faint hint of the odor. I had been running cooler tests for 24 hours up until this one, but two hours at either speed was all I could handle. I assume this is an isolated incident, but it was highly unpleasant.
|
|
