Date:
Tuesday , April 07, 2009
Author:
Marc Adams
Editor:
Kyle Bennett
Google +1

Thermal Paste Shootout - Q209

We take 10 high profile thermal pastes used for mating heatsink surfaces to CPUs and other components and subject those to over 1200 hours of testing to find out which one reigns supreme. Drop or spread? And we tried American cheese too, just in case you find yourself in a pinch.

continued...

Temperature Results

While some of the thermal pastes tested do not state a required amount of time for curing to occur, some do and we want to test to see if any of the pastes perform better over time. Since this can get out of hand rather quickly we settled on a five day period over which we will test each paste. Fewer days would result in the possible skewing of the results for some pastes and more time would have caused this article to take an exorbitant amount of time to publish.

Each paste will be tested five minutes after being applied followed by a twenty-four hour, forty-eight hour, seventy-two hour, ninety-six hour and one hundred and twenty hour interval. During testing the PC will not be used for day to day tasks which would introduce inconsistencies. Rather, each day the PC will be started and allowed to idle for one hour and set to full load using Prime95 for thirty minutes. These heat cycles will allow the paste to cure if needed and remain consistent across all pastes tested.

Let’s first look at how the pastes perform under stock settings during the curing period.

Stock Settings

Article Image

Article Image

What is interesting about these scores is that even pastes like Noctua’s NT-H1 which is listed as requiring no curing time do in fact show some improvement over time even if only a minute amount.

Overclock Settings

Now let’s crank up the heat and see if we can’t thin out the herd.

For our overclock portion we turned the CPU up to 3666 MHz with 1.4v which is lower than what we use for heat sink testing. This is due to the fact that the CPU needed to last significantly longer and if any issues arose from too aggressive an overclock the whole article would have to be restarted or scrapped. Now this doesn’t mean that our other settings are unstable. This just reflects a bit of caution on my part when dealing with a $1000 CPU and a wide variety of thermal pastes and testing taking 1200 hours.

These temperatures were recorded directly after the one hundred and twenty hour mark.

Article Image

We really start to see who can handle the heat here and in no uncertain terms it is the Shin-Etsu 7783D paste that takes the top spot today. Of course when you step back you realize that all participants performed very well today (our cheese excluded) with a variance of four degrees Celsius between the top and bottom spot.