Thermal Paste Shootout - Q209 @ [H]

Gelid GC Extreme seems the best for me as it has a thinner feel like Arctic Silver 5 making application easy as pie (rare in the top performers), is safe for my mobo if I spill any, has no curing times and is clearly a top performer in the reviews.
 
i know these can probably be exhausting seeing as how you have to test each for 72 hours, but why no diamond thermal paste? It's been around for quite a while and people have made their own and claim better than anything you can by. So can it be added :D
 
I've always wondered what a superthin piece of alumimum foil (like what you find at a dollar store, not the quality thick stuff) cut into the size of a quarter would perform like. I'd normally assume it would be worse - because there is now an extra air gap but then again a very thin piece of aluminum will deform much easier to fill in gaps much more than a block of aluminum does (Think squishing alumimum foil between your fingers too see your fingerprints) Besides the fact Anodized aluminum is many orders of hardness compared to foil.

I've often thought the same about ultrathin sheet lead (like used in lead sinkers, not very environmentally friendly, but interesting still) or maybe even gold leaf if we are talking about $10 a gram anyhow, why not.

On a lot of processors - there seems to be some concavity and it seems that filling the void with something - anything (like cheese too) is better than leaving the airgap.
 
Last edited:
I've tried diamond pastes before. I've even made several different batches with varying grades of diamond powder. Its actually quite complex to explain, but diamond paste isn't where its at. I never got any of them to perform better than AS5.
The Coollaboratory Liquid Metal Pro is a whole nother story. This is the TIM of the future. It has worked better than anything I have ever seen. The entire material itself is a thermal conductor, unlike your typical pastes that still contain binders and such substrates that dont conduct heat very well at all.
The key to using the stuff is that both mating surfaces must be nearly perfect. This means you will be lapping your HSF AND the IHS on your CPU. Oh, and don't use it with anything aluminum.
Posted via [H] Mobile Device
 
@ICDiamond: I've tried IC as well. Doesn't make any noticeable difference vs. my OCZ Freeze.
 
Whoa, cheese actually works? Would it rot or mold? It looks like it could work for non-overclocked machines.
 
You can buy extremely thin, silver foil. I wonder how that would work?
 
Last edited:
all the high end TIM's are SO CLOSE.

Exactly. And so much depends on ambient temperatures that it hardly makes a difference. As long as you select one of the top 5 thermal compounds, all is well.
 
Sounds like a contact problem or another anomaly. Test show that much deviation between TIMS is not the case.
 
Sounds like a contact problem or another anomaly. Test show that much deviation between TIMS is not the case.

Until you compare the thermal transfer capabilites of Liquid Metal Pro against even the leading thermal paste. It is quite significant. People just won't give it a try for some reason.

Here is a nice big comparison:

http://www.hwreviewlabs.com/index.p...-up&catid=32:round-ups&Itemid=47&limitstart=3

tic_comparo.jpg
 
Last edited:
You're own table shows a 1.5C difference between AS5 and Liquid Pro. Like I said before, you had something wrong when you applied AS5. But I should have said that all HIGH PERFORMANCE TIM is not ever a 10C difference.

You should probably read other reviews, not just the ones that tell you what you want to hear:

http://benchmarkreviews.com/index.p...k=view&id=150&Itemid=62&limit=1&limitstart=11

Liquid Pro didn't even make the top contenders here.
 
Last edited:
You're own table shows a 1.5C difference between AS5 and Liquid Pro. Like I said before, you had something wrong when you applied AS5. But I should have said that all HIGH PERFORMANCE TIM is not ever a 10C difference.

You should probably read other reviews, not just the ones that tell you what you want to hear:

http://benchmarkreviews.com/index.p...k=view&id=150&Itemid=62&limit=1&limitstart=11

Liquid Pro didn't even make the top contenders here.

That isn't my table. That is from the link in my post. I've been doing this for a LONG time. I know how to apply thermal pastes.

Like I've told people before, in order to get THE MOST performance out of the LMP, you HAVE to spend quality time lapping both the IHS and HSF. If not, you won't get much better performance than high end pastes. LMP will actually bond to bare copper, creating a much better thermal interface. Leaving the nickel plating on the IHS and HSF surfaces can prohibit a good bond. I'm willing to bet the people doing comparisons on thermal pastes don't have a lapped IHS or HSF, and don't allow enough time for each paste to set and level out thermally.

I also like the OCZ Freeze. Since LMP doesn't like bare naked dies, you must use a paste (a lot of GPU dies are naked).
If you don't want to go through the work (and voiding warrantee) of lapping, definitely pick a different TIC like OCZ Freeze.
 
I don't think you mentioned lapping and LMP in the same post. I cannot comment on lapping and then using LMP. Is there a review that actually laps first, and then tests?

I didn't mean you put the AS5 on wrong. It just sounded like there was something wrong to use another TIM and get a 10C decrease.
 
I don't think you mentioned lapping and LMP in the same post. I cannot comment on lapping and then using LMP. Is there a review that actually laps first, and then tests?

I didn't mean you put the AS5 on wrong. It just sounded like there was something wrong to use another TIM and get a 10C decrease.

I had mentioned it in one of my posts much earlier on, but not in my previous post.

I do wish that I could find a review where the person testing had lapped their IHS and HSF, and did a light lapping with high grit paper between thermal compounds. This would rule out any cross-contamination as well as deviations in the 'flatness' of the IHS and HSF.
 
That was a pretty interesting article.

I'm surprised how well American Cheese did as a TIM :p
 
Well if Mr. Baz is correct, lapping both the spreader and sink and using Liquid Metal makes a 10C difference, and that's no joke.

As far as not lapping goes, I just used Shin Etzu G751 and Thermalright The Chill Factor and found no real difference.
 
Been using AS5 for the 5+ years I have been into computers. Still plan to continue using it just mainly due to the fact Radioshack has it and I don't have to pay or wait to get it to me. :p
 
Still got a big tube of AC Ceramique, love it also it's non capacitive
 
Did a new build over Black Friday. Athlon II Quadcore 2.8Ghz (Propus 630) with stock cooler. Had a thin layer of dark gray compound applied to the heatsink. Was in more of a hurry to get everything tested, so I used it as is.

Temps ranged from 74f to 113f peak. Average idle temp was 74 to 79f, average load temp (not forcing full load but under my uses) was from 109 to 111f.

When I swapped my parts into the new case I got for everything, decided to apply the compound I bought on sale along with everything else. Arctic Cooling MX-2. Cooling setup, number and types of fans, arrangement, etc... is pretty much identical. It was just a better looking case.

2 weeks now. Any possible curing time would have elapsed. Saw no real differences between just-applied and a few days after. Temps have ranged from 72f to 107f peak. Average idle temp has been 74 to 76f, average load temp was 103f to 105f. I could probably raise that ceiling easily... but under my general usage, that's a nice drop.

Interesting. No overclocking here. CPU fan speed is constant, so no changes under idle versus load. Was surprised how thick the MX-2 was. Had a hard time getting it appropriately and consistently thin. Ended up having to use the brush on my empty bottle of Zalman compound to get it spread out evenly.

An improvement, certainly, but not as wide as I expected it to be. Stock coolers and the compound applied to it have come a long way since having a plastic squeeze bag of white grease included with it, that's for sure.
 
i'm surprised to see the ceramique doing so well. i use it on xboxes and customer pc's because it's cheap and comes in huge tubes but it's nice to see it actually does quite well.

i'm of the opinion that as long as you're well below a processor's max operating temp, you're not going to damage anything. the measures people take to go from a 45c load temp to a 40c load temp seems almost ridiculous to me but i guess that's why they call it a hobby, heh.
 
Just went to the Egg and bought the Shin-Etsu 751 2nd place stuff. On sale so I bought two so the shipping didn't scare me. I only build three or four units a year so this will last for a while.
Thanks H!
 
i havent seen an article as light hearted as this one in a long time now.....well, since April 7 2009 to be exact.....cheese....LOL

i think it's time for a reprieve !!

and yes i necro'd this bitch, laugh a little
 
My Artic Silver 5 has been in my drawer for years, and now its liquid has separated from the solids. How do I fix this problem? .... :(
 
My Artic Silver 5 has been in my drawer for years, and now its liquid has separated from the solids. How do I fix this problem? .... :(

mix it up....probably should just wait until next time you need it, squeeze some out on a piece of paper, mix it up good with a toothpick or such and then use it right away, toss the rest, get a new tube
 
Just an update on the MX3. I did use this and it's excellent. I don't think you'd need to look further for a paste. This paste also has an 8 year shelf life. I tried 3 different pastes, including Arctic Silver 5?, the Shin Etzu, and one other top brand, but this was the best of all. It is a very thick paste. My temperatures were all about the same under load, but the winning feature of this paste is the 8 year shelf life.

OCed Core i7 920 at 3.8Ghz / Prime 95 Torture Test:

Ambient: 65F/18.3C
Max Temp: 74-76C
Idle: 38C with fans on medium. Fans on full around 35C

Playing Crysis for hours: never over about 58C.

My motherboard has a lot to do with this, however. I installed the Gigabyte UD5 rev2 and for some reason, the max vCore never goes over 1.264. With rev1 I had to push the vCore to 1.34 to get a stable OC.

HardOCP.jpg
 
Back
Top