Slashdot Mirror
The Thermal Paste Revolution
arhines writes "ZZZ is running an article about an interesting new thermal paste which surpasses even solder in thermal conductance by 33 percent. If this paste makes it to the market sometime soon, we'll all surely be thinking about putting it in our boxes. In fact, if use of the paste becomes commonplace, it may even give the semiconductor industry a little speed boost."
They seem to have just duplicated the inventor's press release - the article doesn't contain any independent evaluation of the substance whatsoever.
of 0.004% that thermal paste actually makes a difference
bite my glorious golden ass.
I think you misunderstood the statement. They may have compared the thermal conductance of this material to solder, but they're not looking to replace solder. Solder is made up of metals, so it's naturally a good conductor. But metal doesn't spread very well over microscopic cracks -- and no, you can't fill in the cracks with solder, because the metal will contract when it cools and be useless as a thermal paste. You wouldn't want to use a thermal paste as a solder, because thermal paste typically takes a very, very long time to dry when not exposed (ie: between a cpu and a heatsink).
So it looks like CowboyNeal is saying if this new thermal paste can improve the effectiveness of a heatsink (and fan) by a reasonable amount, manufacturers will be able to push their clock speeds a little higher.
>"we'll all surely be thinking about putting it in our boxes"
;-)
I won't be doing this. Such a think could void warranties. Stability and reliability is more important than speed IMHO - especially since I don't play games.
If manufacturers start using this paste, and it doesn't deteriorate after 5 years, then that is different.
Obviously it won't be used in resistors - as conductance is not very good in such components
Mike
This seems a very odd comparison to me; when was the last time you soldered a heatsink to a hot component or used paste to assemble a circuit board? The use in chips seems a little suspect too..
:-)
I can just see the warnings now. "Do not mount vertically or internal circuits will drip out!!!"
Technologies like this are complementary to each other.
As someone previously pointed out, one of the major problems facing device manufacturers is transferring the heat out of the die. When improvements are made in this area, then the problem will still be transferring the heat away from the die where it can be dissipated by however (water cooled slug, air cooled sink, etc. etc.) once the heat can be effectively moved away from the die-face at 99.999% efficiency there's still the issue of radiating the heat away from the assembly altogether. Then there is still the ambient air temp (have you priced out a redundant 10+ ton HVAC system for a datacenter lately?)...
It's all a catch 22 but anytime a significant improvement is made in one of these areas is leads to higher performance and lower costs (tolerances can come down or stay at the same level for a bit while increasing performance), which by themselves may seem minor, but when they are all taken together make a significant improvement.
01:36AM up 426 days, 2:46, 1 user, load average: 0.14, 0.11, 0.05
The cooling technology has been around for decades. Take a look at old Cray supercomputers or IBM ECL mainframes. The problem is that it has to be designed into the system from the beginning, not tacked on to an existing design.
Mea navis aericumbens anguillis abundat
What's the point of buying exepnsive silver-based paste if 3C reduction is all that you will get?
i don't give a flying fsck what anyone uses.
i've got a sachet of generic cpu aluminium oxide stuff that came free with a heatsink that's enough to treat about 30 CPUs at a guess, that i've been using. arctic silver's easy to get hold of - it may be much more expensive than a tub of generic heat paste from an electrical components shop, but it's still only a couple of quid.
a milspec number might indicate quality, but the converse is not true. plenty of good things don't have a military spec number on them...
If possible, switch on the machine for a few minutes before removing the heatsink.
This will heat the CPU and it will usually melt the thermal paste. It should now be easier to remove the heatsink from the cpu.
You know, that sounds like a major pain in the ass to save a few degrees. Maybe manufacturers should just make cooler CPUs instead of us needing industrial strength copper heatsinks and 19mm fans. I really miss the days when a heatsink and fan were optional on a CPU. I think my 386 was the last of that kind though. My 486DX2/66 ran fine after the fan on the heatsink failed though so the heatsink was good enough. Even most Pentium systems were just fine with a big heat sink and no fan. Nowadays all my Athlon systems sound like I'm at an airport with the 5 or 6 fans I need in my system to keep it cool.
The big win with using a better thermal paste is that you can use a quieter fan to cool your CPU, since you don't have to supply as much air to specific spots. (Heat buildup in modern CPUs is quite local and varies with what you actually do). Better thermal conductivity will not do much for total heat dissipation.
However, reducing fan noise by 3 dB is worth a lot to me.