Thermal Paste Comparison, Part Two: 39 Products Get Tested

[Microgadgethacker]

Great article and very helpful! I had a gaming laptop with an overheating GPU, and appreciated the tips on how to apply the compound. For anyone in this situation, CHECK your heatsink's FLATNESS!. I held a metal ruler against the bottom of mine and it had a good 2mm bow to it, ruining the contact with the GPU. Since it was 2mm copper extending from the heat pipes, I was able to carefully remove the bowing. Good paste and this fix together dropped GPU temp by more than 20C in hard use.

 

[Dr_b_]

Did you let the Arctic Silver 5 cure or bake in before taking those temp measurements?

 

[PyroYuy]

I delidded my 3770k, and used liquid metal between the die and heat spreader. Dropped my temps by 4 degrees.

 

[Paul Jacobsen]

What about IC Diamond or the Antec products?

 

[Rock71]

Excellent article!. I was wondering, what was the CPU frequencies and voltages to do stress testing for this results??

 

[FormatC]

What about IC Diamond or the Antec products?

Take a look at the charts! IC Diamond is listed with two products...
And Antec? Antec is a label, not a manufacturer. I can only test what I get and I wrote to all my contacts to send me samples. If not, it means not. I'm not able to force someone.

 

[kindtmj]

Thanks for this and all the great info. Reworked my NVIDIA 285 ftw graphics card with Coollabs metalpad from arctic silver 5 and redid the pads to fujipoly. Went from 72c to 65c. I am so pleased. Really cool to watch the temp drop at burnin (a one time deal).

 

[MessalineApghar]

I'd like to see IC7 Diamond and Antec formulas 6 & 7 added, as well as Tuniq's selections. their TX4 was a longtime favorite of mine.

 

[jawnTEM]

Very well written, and most informative for enthusiast and beginner alike. This was also, a lot of work, so I applaud your dedicated efforts. Such articles as these are a very much needed in the community.

I have been using Indigo Extreme for a number of years and was hoping to see it included in your coverage, and, by someone that obviously knows their way around the CP block.

Well done!

 

[JackNaylorPE]

can use this to augment the data in the article

80-way TIM Comparison

http://archive.benchmarkreviews.com/index.php?option=com_content&task=view&id=150&Itemid=62&limit=1&limitstart=12

 

[jawnTEM]

can use this to augment the data in the article

80-way TIM Comparison

http://archive.benchmarkreviews.com/index.php?option=com_content&task=view&id=150&Itemid=62&limit=1&limitstart=12

Thanks, Jack, for the extra info and link. An increase in knowledge is always welcome.

 

[JackNaylorPE]

I have learned a lot about coolers, TIM, application methods on that site. Most importantly, that due tot he different heat signatures, performance on one socket does not necessarily transfer to another socket.

 

[jawnTEM]

I have learned a lot about coolers, TIM, application methods on that site. Most importantly, that due to the different heat signatures, performance on one socket does not necessarily transfer to another socket.

Well..., That is quite a logical concept and development of train of thought. Lol! Too bad that others, (Speaking about me mainly) haven't come to that conclusion as well. Certainly makes sense though, and most glad that you brought it to my attention.

I'm wondering though, just how critical the degree difference might be. Especially, when you consider that most TIM's, aren't really as effective as most people imagine or hope that they are.

 

[JackNaylorPE]

IIRC, you will see about a 4.5C difference between the best and the worst....

As for the socket think, that's why I have no use for sites that use a heat block instead of real CPUs. For example....the EK Supremacy rules on 1155 but the Koolance 380 takes the top spot on LGA 2011

http://www.xtremesystems.org/forums/showthread.php?285753-CPU-water-blocks-roundup

 

[Jason Elder]

What alot of people don't realize is that CoolLaboratory's products are usually used for people that delid their CPU where it can be applied in between the CPU die and metal CPU heat spreader. That is where people see the best benefit. Pair that with AS-5, Arctic MX or Gelid thermal paste between the CPU heat spreader and your heat sing you'll see much better performance.

 

[Timeking]

Why aren't "mainstream" thermal compounds like Dow Corning 5026 or GC Electronics Z9 paste included in these tests? Truth be known, I'd bet that some of these proprietary compounds are just mainstream electronics paste re-packaged. Yes?

 

[Calculatron]

Why aren't "mainstream" thermal compounds like Dow Corning 5026 or GC Electronics Z9 paste included in these tests? Truth be known, I'd bet that some of these proprietary compounds are just mainstream electronics paste re-packaged. Yes?

Yep. Be Quiet!'s stock compound, not the DC1, is Dow Corning 5121, and I do believe Akaska's 5022 is Dow Corning 5022.

Since all of these TIM were listed in the complete, updated list, it goes to show how much a product can vary from either lot to lot production, or from just mounting your cooling solution ever-so-differently.

 

[Asmyldof]

About the burn in of Liquid Metal Pad:
Of course, you could do the burn-in with the CPU/GPU itself, but consider that a chip has a much higher storage temperature than its advised core operating temperature. You can, for nearly all CPU/GPU models just "bake" the heatsink+processor in an oven that can do 80deg Celcius (or, better, 100deg with an accuracy of 10degrees at worst). Most household combination microwaves and hot-air ovens can generate a temperature between burn-in and maximum storage temperature.

Be very careful with the electro-static discharge though, before putting it down, or picking it up, hold the oven or tray with the other hand (after baking: let it cool first, of course!).

In fact, as a Freelance electrical engineer my workshop houses a normal house-hold hot air oven of $ 120, in stead of the professional alternative of $1200, with no useful extra options and it's used extensively for burn-in's and such.

 

[tlhudson69]

The only other test I would like to see is if you "forgot" to put ANY paste on at all!! How does it compare to the "just for fun". The toothpaste could actually act as an insulator and not a conductor....Just curious!!