Thermal Paste can Squeeze Out

Actually, AMD said it because over time with the vibrations thermal paste can squeeze out from under the HSF and leave less thermal conductivity. By over time, of course, they mean 6-12 months of 24/7 usage. Pads hold their shape much better over that same period, and can be used for 3-4 years without replacing. In addition, they only explicitly recomend it on their Athlon MP systems, where vibration and hence the excess spreading is that much worse (two fans, more HDDs, sometimes 15k RPM SCSI drives, and often in a rack where there’s still more vibration). As a side note, there are pads that approach the thermal conductivity of AS3, although they don’t reach it. They’re also phase change pads.

Personally, I use AS and just reapply it every semester (the ooze is noticable since I have alot of vibration), but on my server I just used the Intel thermal pad on the HSF.

There is a difference between thermal pads and thermal tape. Pads aren’t adhesive, tape is. I have both hanging around here, and yes, I use tape only on RAMsinks (where I buy industrial thermal transfer tape direct from Minnesota Mining and Manufacturing, none of the weak stuff normally used). As for checking and cleaning their system once a year, that is a bit excessive for the average user. My servers havn’t been checked since they were built for any problems under the HSF, and I don’t expect any. Why? Because I use pads on the CPUs so I don’t have to worry about it.

Now, when I did the mobo and CPU shuffle down to server for one of the Celeron 400A systems I had (BP6 based), I did remount the motherboard so I did replace the cooling after 12 months as a workstation. One of the CPUs had the stock Intel HSF with thermal paste, the other had some generic HSF with a pad. Believe it or not, the paste, AS original actually stripped the metal coating on the IHS and left me with the bare copper underneath, and I could see green dissolved copper all over the HSF and throughout the AS. It was pretty sickening. The bare dies are not prone to this, but I tested it on a Pentium III with IHS and was able to get to bare copper within 30 seconds of rubbing with AS.

The pad I use on that CPU is made by thermaltake and works just fine, btw. It comes attached to their Silver Orbs (after discovering that, I quickly swapped my HSFs for Silver Orbs and the problem hasn’t come back).