I’ve never really appreciated just how much of a difference different thermal pastes can make. I’ve an old IBM / Lenovo ThinkPad x60. Over the years I’ve noticed it have a few thermal shutdowns during particularly CPU intensive tasks, for instance during a Linux kernel compilation or a TrueCrypt partition encryption. I’ve always just figured that “it was one of those bad designs” that could not be improved upon.
Anyway, last night it had a thermal event while applying Microsoft patches!! Hmm. I figured this was a bit much. Once before I did open it up and check the fan for lint and dust but it made no difference. I’m not sure why but last night’s shut-down annoyed me and I decided to try an improve the situation. So I decided to follow this blog post about stripping down an X60 to replace the fan. I wanted to double check the dust and lint situation and also replace the thermal paste on the CPU.
Once I undid all the screws and unattached various wires and ribbon cables I had access to the CPU cooling components. The thermal paste looked OK but it was to be replaced. I had some Arctic MX-4 on hand which I was going to use. I cleaned the CPU and heat sink as best I could using a lint free cloth and an isopropanol based cleaning liquid. I applied some new thermal paste and spread it quite thinly. I reassembled the laptop (not sure it will ever be the same!!) and somehow managed to have no screws left over!
I fired up the laptop and all looked OK. I started running Prime95 to generate load and the CPU temperature (as shown by CPUID HWMonitor) is hovering around 87 deg C. Last night I noticed it getting up to about 92-95 degrees. So, a new layer of thermal paste has dropped the CPU temperature by 5-8 degrees. More importantly, however, for me is that the laptop is running Prime95 without having any thermal events.
I know the over-clocking community has various recommendations for which thermal paste to use – but it seems it is relevant even for unmodified PCs!