Skip to content

Martin's Look at Processor Cooling

Hello, all! Martin here and welcome to our most recent edition of Martin's Corner. It's been quite a while since I last made an update, since that time we have completed our move to the Austin, TX area. And everyone has taken the time to get settled into our new home (many of us have also relocated the places where we sometimes park our cars and sleep as well). :-)

This month I wanted to talk about processor cooling. In the lab here, we like to use a product called Artic Silver. (consider this a free plug to them) On their website, they have some very detailed instructions regarding the application of their product. We realized that many of our readers may not know the in's and out's of processor cooling, and might find this interesting. There is a lot more going on here than one might think when it comes to processor cooling and it definitely will impact the performance of your system!.

First of all, the temperature reading of the chip is not always the most dependable method of determining a cooling compound's level of effectiveness.
Depending on the situation, the temperature your chip reports may be HIGHER when your thermal compound is WORKING BETTER.

Sounds backwards huh? Here's the explanation: Intel typically positions their thermal couplers (the part of the chip that detects the temperature) in the same area as the memory cache... which is not the hottest part of the chip. This means that when a thermal compound is working best, it is making the heat spread and flow away from the warmer areas. This means that instead of acting as an insulator, (keeping hot areas hot, and cool areas cool) which might result in a lower temperature READING, a good compound will act as a thermal conductor, conducting or transmitting the heat throughout the compound. (and into the heat sink, and into the air). That's why the chip may report a higher temperature (as much as 5C higher) when in fact, it is cooling more effectively. By spreading that heat around, (and away from the source) we're more evenly distributing it over the surface of the chip (and over the surface of the heat sink), in the process however the heat is also spread over the chip's thermal coupler, which can result in a higher temperature reading.

So then how do you know if your compound and heat spreader are actually working vs. another compound that may result in a higher temperature reading because of poor performance? Well, a much more effective measurement of a cooling compound performance would be to measure the amount of heat being conducted into and radiated from a particular heat sink. In two given machines with equal specifications, and identical heat spreader assemblies, if one heat spreader is warmer than another, that would indicate that the warmer one is pulling heat away from the chip more effectively in that instance…which is a good thing.
Most people think that the application of the 'grease' or thermal compound to the chip is not a very complicated thing. "Just goop some on, clamp down the processor heat sink/fan, and you're good to go". That would not seem to be the case at all. As far as the quantity you apply… for this product at least, "less is more". Too much, and you create a layer of insulation, that can't conduct the heat away fast enough. Too little, and well, you will have too little to be effective, and once again, you are insulating the chip (though this time, with air pockets), instead of properly filling those microscopic cracks and crevices with the compound, so it can conduct the heat away from the source.

Take this example of a dual core chip. They only suggest a very thin line of compound, down the center of the chip, and even caution to apply the compound in the correct direction. The purpose is to allow the compound to be positioned directly over the source of the heat, conducting as much of the heat as possible over the entire surface area of the chip, maximizing the surface area in contact with the heat sink/fan assembly.

In this example with a quad core chip, the suggestion is to apply the compound in a different direction, running across the hottest parts of the chip assembly.

They also caution that if you are using a heat sink/fan that already has used thermal compound on the surface, that it should be removed first, ideally with alcohol and a soft tissue or cloth. Not all compounds will work well together, and dry compounds in particular are not compatible with grease.

Remember, the idea is to conduct heat away from the chip. If you mix two different compounds you will very likely just increase the insulation characteristics of the compound, making it less effective.
There is a lot more science to this than one might think, and that's just for the application. We've not even touched on the characteristics of the thermal compound itself!

The application instructions for the various compounds may be found at this link: http://www.arcticsilver.com/instructions.htm

Remember, the application techniques we've discussed here, will change for the various different types of thermal compounds. Different thermal compounds whether from Artic Silver, or from other manufacturers will each be applied differently. Never assume the instructions for one product will be applicable to others.

Well, I hope this has been enlightening for you. Happy Cooling!