Hey, I was testing the thermals on a passively-cooled Atom chip.
I forget which thread I mentioned that I had this thing in and that I'd see if the heatsink could be removed for you, but it's a Supermicro A1SRi-2578F. The Atom chip is rated as a 20W part.
I was able to remove the CPU cooler and was having strange issues ever since I did. I tried reapplying paste, but my rubbing alcohol didn't seem to faze it so I put it back on like that. My CPU was consistently going all the way down to a x12 core clock multiplier (running at 1.2Ghz vs its stock 2.4Ghz)... So I reapplied thermal paste which took a lot of elbow grease (it seriously took me about 40 minutes to get all the thermal paste off and it involved me carefully scraping at it to get it off) and booted everything up. My clock has stayed at its x24 multiplier for a few hours now and I even just got done running a stress test for an hour.
Here's where we get back to passive cooling being an issue. I was stress testing this chip (and paste and heatsink) to see if things were back to working order. They are, but over the course of an hour, a 20W chip being stress tested got all the way up to 70C on a passive cooler.
Similar to what Dimitri did in that HC video, I plugged a case fan into one of my fan headers and put it over the 70C passive heatsink. Over the course of time it took me to write this post, my CPU has cooled by 23C, with the fan being slowed down by PWM.
Siffice to say, having something to circulate heat out is a tremendous boost to the effectiveness of any sort of radiator. It might be possible that you could create a large enough heatsink to where it could actually "conduct" room-temperature air to circulate, but this thing made it all the way up to ~150F and I can assure you that's nearly double the temperature in the room which outght to be between 72F and 78F.
I forget which thread I mentioned that I had this thing in and that I'd see if the heatsink could be removed for you, but it's a Supermicro A1SRi-2578F. The Atom chip is rated as a 20W part.
I was able to remove the CPU cooler and was having strange issues ever since I did. I tried reapplying paste, but my rubbing alcohol didn't seem to faze it so I put it back on like that. My CPU was consistently going all the way down to a x12 core clock multiplier (running at 1.2Ghz vs its stock 2.4Ghz)... So I reapplied thermal paste which took a lot of elbow grease (it seriously took me about 40 minutes to get all the thermal paste off and it involved me carefully scraping at it to get it off) and booted everything up. My clock has stayed at its x24 multiplier for a few hours now and I even just got done running a stress test for an hour.
Here's where we get back to passive cooling being an issue. I was stress testing this chip (and paste and heatsink) to see if things were back to working order. They are, but over the course of an hour, a 20W chip being stress tested got all the way up to 70C on a passive cooler.
Similar to what Dimitri did in that HC video, I plugged a case fan into one of my fan headers and put it over the 70C passive heatsink. Over the course of time it took me to write this post, my CPU has cooled by 23C, with the fan being slowed down by PWM.
Siffice to say, having something to circulate heat out is a tremendous boost to the effectiveness of any sort of radiator. It might be possible that you could create a large enough heatsink to where it could actually "conduct" room-temperature air to circulate, but this thing made it all the way up to ~150F and I can assure you that's nearly double the temperature in the room which outght to be between 72F and 78F.
