Testing

Methodology

We’ll be using both the ZEROtherm Nirvana NV120 Premium heatsink and the AMD stock HSF. The reasoning behind this is that we discovered that the clip system on the ZEROtherm cooler would prevent one side of the interface pad from making perfect contact with the core, reducing the effectiveness of the paste. The stock cooler however always makes perfect contact with the core although its cooling ability is lower than the Nirvana. The Nirvana NV120 was run without the fan speed limiter meaning it was running at full throttle hence the disparity between the earlier results for that cooler.

The temperatures are taken from the built in diode on the AM2 6000+ Dual Core processor. Idle results are recorded from letting the computer idle for 30 minutes directly after cold boot and taking the average temperature. Load results are recorded by running OCCT for 30 minutes and taking the average temperature.

Ambient temperature is 19C.

Results

The results are very close, close enough that it could have been reading error etc. causing the temperature to change by a small fraction. As a result, each test was repeated three times, which means the coolers were cleaned and the CPU IHS each time for a reapplication to remove the chance that a single application of paste could sway the results. After cleaning 18 times, and using nearly all of our kitchen roll in the process, we got these results.

For the ZEROtherm cooled processor, the Tuniq TX-2 manages to pull a victory out of the bag, with the Noctua paste a close degree behind, while the Arctic Silver 5 is held at arms length lagging behind with a temperature of 37 C. All of the pastes managed to keep the core at a low 24 C for the idle testing.

It should be noted that the ZEROtherm cooler didn’t always make full contact with the core, with varying amounts of contact (between 50% and 75%) meaning that there were other factors involved. As a result, I also tested with the stock cooler which gave me these figures:

This time, the Noctua heatpaste seems to be the winner by a tiny fraction, beating the other two contenders by a degree at both idle and full load.

While Arctic Silver 5 in these tests appears to be the worst of the lot, it does require a burn in time that the other two don’t meaning that these nearly instant tests are putting it at a disadvantage. As a result, we tested the Artic Silver 5 after it had had 200 hours of drying time, and the difference wasn’t repeatable (multiple runs gave varying results) and the difference was 1-2 degrees at best.

These results are quite inconclusive as to which is the greatest product to smear on your CPU, but the Noctua NT-H1 and Tuniq TX-2 seem to pull ahead when the cooler on top is more capable at removing core heat. If you are running a quad core overclocked chip, you are probably going to see the differences rise between the pastes with these two products pulling further ahead of the old dog Arctic Silver.