rctic Silver ArcticClean
was used to clean both the IHS and heatsink base between tests. I know of no other cleaning and preparation product on the market which works effectively as ArctiClean. Prepping your thermal transfer surfaces is just as important as the TIM you’re applying and how you apply it. On a side-note Arctic Silver also hosts what must be the most in-depth instruction pages of any TIM maker. Below I've illustrated a photo to show their recommended "line" across the IHS application for Quad Core. The water block pictured above with the phenomenal finish is Koolance CPU330
. (Background photo borrowed from TheRegister, Bits 'n' Chips)
Intel Test System
|CPU ||Intel Quad Core Q6600 Retail 2.40GHz (1.285V) Socket-775|
|Mainboards ||Gigabyte GA-P35C-DS3R|
|Memory ||OCZ Technology DDR3-1066 (2x1MB)|
|Graphics ||BFG 8800GTX|
|Power Supply ||NZXT Precise 1200W|
|Storage / Optical ||Seagate Barracuda ST30815AS / Plextor 760SA|
|Cooling ||Intel Stock Cooler used through out|
|Operating System ||Windows XP|
In order to simulate a 100% LOAD to all four cores I chose Prime95 v25.5a
. This and Orthos X are the only two programs which currently support Quad core testing. Core temps were recorded using Core Temp 0.95
and an average was calculated from the values. To increase wattage beyond the processor's TDP our Q6600 was overclocked from 2.4GHz to 3.0GHz (Vcore 1.28V / 9x334FSB) for a baseline 128W at IDLE. I recorded every result with a crop of Prime95 and Core Temp 0.95. An example can be found in the thumbnail below (temps are exceptionally low in this example as it represents H20 temps under LOAD)...
In every test I applied a specifc paste then cycled between IDLE and LOAD as produced by Prime95. After 7-days I removed the heatsink and photographed the IHS and heatsink base. For each paste I repeated the process above three times for each paste and for each application method to ensure I was getting solid contact and spread. I chose the lowest temp out of these multiple mounts for use in our results. In theory this should have given us the best contact (heat transfer). Throughout these tests I made sure the Ambient temp remained at a steadfast 21°C.
The charts are separated by application method.Conclusive Thoughts
With the exception of Intel's factory installed thermal interface material, the temp difference between pastes was relatively close. Clearly synthetic blend pastes are now dominating the market and while silver infused products have great potential they are not the competition killers they once were.
By the criteria laid out on the first page:Thermal conductivity of the material
Electrical conductivity of the material
Spreading characteristics of the material
Long-term stability and reliability of the material
Ease of application
In the list below in order of the best performer first, reflects the criteria above as best as possible:
1. Tuniq TX2
2. Arctic Silver AS5
3. Arctic Cooling MX-2
4. Arctic Silver Ceramique
5. Nanotherm PCM+
6. Intel stock thermal interface material
The number of pastes tested today doesn’t come close to what’s out there. One thermal interface material in particular I wanted to test was Innovation Cooling's IC Diamond 7 Carat Thermal Compound which in theory has the ideal formula, and CooLaboratory LiquidMetal Pad different from their Liquid Metal Pro. After the holidays I will be repeating these tests including the products mentioned on both a simulated-die and on the Q6600.
If you’re in the market for a new tube of thermal paste, I would strongly recommend Tuniq TX-2 as your next choice. It out performed the competition tested here, it’s much easier to apply and remove, and requires no spreading. Initially I mentioned price, but when you think about the role your TIM plays in the scheme on things their all invaluable and whether one paste is a few dollars more or less pales in comparison.
I would like to thank all the manufacturers fro submitting samples.