I constantly hear about Arctic Silver on Forums.
It seems like every OEM and builder uses something different.
Some manufacturers tout their own stuff, like Cooler Master and their 'Nano-Fusion'.
So it seems like a great subject for discussion (and Testing!)...
What is the best thermal interface for CPUs and Heatsinks?
personally i use artic silver 5 never had a problem with it and i always recomend it compared to that crap intel sends with its boxed processors.
any name brand of thermal paste/grease will be good quality, just beware of the white sutff that looks like elmers glue in the unmarked plastic. it really is glue.
It largely depends on what HSF/WB you get and the surface finish of it.
You would choose a different grease/paste/gap-filler, a different thermal interface material, depending on the surface of the heat-sink? So, you'd choose one brand if the surface was lapped smooth, and another if it was rough?
I'm definately learning here - they don't cover this sort of thing in my computer classes.
Regarding the Arctic 'X' materials, they say that the efficiency of the product actually improves, as it cures - although this takes 300 hours...
That makes a certain amount of sense, although I wonder what percentage of improvement one would see.
It also would seem to make testing the full range a bit more time consuming. Let's see, 300, carry the 2... this would take 2 weeks of 24 hour operation, with a reasonable load, before it would be ready for the 'torture test'.
It would definately be an interesting article for Tom's.
I also read the article in Maximum PC, and the bottom line was... use any thermal grease or paste, since the temp diff is usually so minimal it doesn't matter. Personally, I stand by AS5, though it seems 50/50 between AS5 and AS Ceramique users...
So far I've only accumulated surface data on Artic range i,e. Silver5, Ceramique, Alumina and MX-1. All vary in performance depending on surface finish.
That must be a rather time-intensive study you're doing, Wusy. Testing not onloy the different thermal compounds, but gathering comparison data on how they perform with different contact surfaces. Perhaps you should publish?
Someplace, I recall seeing a poster recommending that one use AS5 between a GPU and waterblock, but stating that Ceramique was better for the RAM-chips in the same installation (Danger Den-type full-coverage waterblock). Struck me as odd, at the time, but if, as you're saying, different compunds work better with different contact surfaces (which does make a certain sense) then the issue seems much clearer. One would want to choose an array of thermal compounds to maximize the effect of each cooling instalation. Or, as seems most common, one would choose a superior compound for the critical interface (like a processor) and use it elsewhere as a 'good enough' solution.
Quote :
I also read the article in Maximum PC, and the bottom line was... use any thermal grease or paste, since the temp diff is usually so minimal it doesn't matter. Personally, I stand by AS5, though it seems 50/50 between AS5 and AS Ceramique users...
I didn't mean to question MrsD's assertion (or yours), simply that I didn't find the article online. So many PC mags, so little time. I generally only do a search on topics that I'm currently curious about and have 2-3 'general' PC news sites and forums that I watch for mention of interesting topics.
It seems clear that some sort of thermal interface compound is better than none at all.
I don't think Wusy was claiming that he had tested all of these, rather that he had read about and taken note of the data. I have seen (I have no idea where) data about the size of the particles in different Arctic Silver products and what surfaces each is best for. I know you can find information about particle sizes on the product pages on their site. I replaced factory thermal paste with AS5 and saw about 5 degrees improvement.
You're right, Fireheart, the bottom line is that the use of thermal paste is necessary, no matter what paste you can get your hands on. Aragorn is also right, 5 degrees celsius with AS5 IS IMPORTANT (to overclockers like me! ).
Well, I have only used Artic Sivler 5 when it comes to after market thermal paste and compares to other thermal compounds that comes with the product like cpu it is very good obviously. Also I have seen some reviews comparing different thermal interface and compounds and the Artic Silver 5 comes to the top, so eversince I've been using it for my cpu, chipsets and gpus and it has given some great cooling results so that I never look back to anything else.
I didn't mean to question MrsD's assertion (or yours), simply that I didn't find the article online. So many PC mags, so little time. I generally only do a search on topics that I'm currently curious about and have 2-3 'general' PC news sites and forums that I watch for mention of interesting topics.
you need to get the mag to get all their articles. The do not put everyting online as they are primarily a print release. That being said, I have read them since they were "boot" and while they are a tad slower on new release hardware tests then sites like Tom's, what/how they test and their conclusions are generally dead-on. Whatever hardware you want truly tortured they have or will do it. I have read articles where their tests resulted in speakers smoking in ruin and psu's sweating from heat. Good stuff.
Their conclusion in the article was that as5 and that liquid metal performed the best, w/ liquid metal being more expensive and harder to use. BUT, all compounds did well, w/ only ~3 degrees diff from all of them.
I have used Arctic Silver's various revisions on all of my systems, but if I "only" had another brand I would use it w/ no regrets.
interstingly mPC also had a blurb about how much paste apple used on their intel macs... so much that it was overheated and ran crappy until they re-applied it for testing... just amusing how apple assumes stupidity on their customers part and gives them crappy quality wrapped in swank design.
Actually, the AS5 - Liquid Metal article in Maximum PC came out a few months after the article on thermal paste comparisons... Don't think I'd spend double or triple more on a paste that gave 1-2C better cooling, if that. 4-6C? Yeah, that might fly...
on my 6800 ultra gpu i use to use as5 temps were around 69 - 74 at load, so i thought i should reapply some, so i took it off cleaned it, couldnt find the as5, so i tried zalmans paste that came with the zalman cu-77000 fan and now temps r 68 - 70. it could be that the as5 just lost its heat transfer after time, but this stuff works just as good for me
You're right, Fireheart, the bottom line is that the use of thermal paste is necessary, no matter what paste you can get your hands on. Aragorn is also right, 5 degrees celsius with AS5 IS IMPORTANT (to overclockers like me! ).
Heh, I'm not an overclocker... yet. I want to learn a little more, before I start cranking up the heat. My current computer is... technically, thoroughly obsolete (P4 1.6 Ghz, on a Abit TH7 motherboard, and a whopping 512 Mb of Rambus memory.) I want better, but the money isn't really there, unless I'm willing to buy another obsolete computer... And I know that just about as soon as you get your box built, it's already obsolete, unless you're willing and able to spend $K (kilo-bucks).
Sure, $500 at Dell or HP would get me a computer ten times better than what I have, but it would be a POS. Very careful shopping may net me something that's not completely up-to-date, but with Potential. Meanwhile, I keep learning.
So, I realize that 5 degrees is significant. It could be that difference between a dangerous Mod and one that would run stable.
I hadn't seen that article before but it was quite informative. Thankyou. I guess this means that if we have a really mirror smoothe surface we should be using AS Ceramique rather than AS5 and AS5 of other applications.
THG just did an article about an extreamly fast machine for $720. If you go to Air cooling you can still have about a 3.8 GHz machine for a little over $600.
THG just did an article about an extreamly fast machine for $720. If you go to Air cooling you can still have about a 3.8 GHz machine for a little over $600.
Yep, that article and the ones related to it have caused me to consider the 805, instead of sticking to AMD. It would require that I learn a lot more about overclocking - I realize it is primarily a matter of working the various ratios of voltage, clockspeed, & FSB to produce some stable ideal, but I just don't know enough to feel secure about diddling with the numbers.
Still, this would be a good solution to my plodding old system. I'm looking into it.
).
