[Skinnee Labs] The TIM Thread...

[Vapor]

Simply a central location on XS for all my TIM testing from 2011

Thread will/should be updated within a day or two of each new installment going live, as long as I remember to keep updating it Reviews will be done in batches of 2 or 3 TIMs, with each TIM taking a minimum of 7.5 days to test (so a little over 2-3 weeks every update).

I'll post the 'master' charts here, every TIM will be on every chart and charts will probably get pretty large when the number of TIMs grows (and hopefully not too difficult to read, but there's only so many colors in the rainbow). Each installment will have a smaller selection of TIMs for comparison so that readability will be easier.

Quick summary of how I test:

I use three different Koolance CPU-360 blocks with various modifications to alter the base's convexity and pair them with different mounting pressures as well. This gives me the ability to test over a spectrum of contact from "Poor" to "Moderate" to "Great." Most CPU blocks land somewhere around "Moderate" contact (some better, some worse); I've never seen a retail block approach the contact that the "Great" block does. Often GPU/NB/MOSFET blocks will be somewhere between "Poor" and "Moderate."

Each TIM gets tested with five mounts per contact setting, with each mount being 10 hours long. Temperatures are logged every second of those 10 hours per mount and from the constant monitoring I can analyze cure (or any performance change over time). The worst mount on each contact setting is excluded from the average.

Full how/what/why can be found in any and all of the installments--this thread is more about the results than how I got them, IMO.


Installments:

Arctic Cooling MX-2, Indigo Xtreme, and Spire SilverGrease SP-457

Arctic Silver Matrix, Cooler Master IceFusion, and Prolimatech PK-1

Coolink Chillaramic, Nanoxia Heat Buster, and Nexus TMP-1000

Arctic Cooling MX-3 and MX-4

Deep Cool Z3/Z5/Z9

Arctic Silver 5, Ceramique, and Ceramique 2

Thermalright Chill Factor, Chill Factor 2, and Chill Factor 3

IC Perihelion, Noctua NT-H1, and OCZ Freeze

Shin-Etsu G751, X23-7762, and X23-7783D

EVGA Frostbite, Phobya HeGrease, TIM Consultants Quantum

Tuniq TX-2 and TX-4

Alphacool Silver Grease, Dimastech HTX-EE, and XSPC K2 (in progress)

Zalman STG-1 and STG-2 (upcoming)

Feser H-Bridge, Scythe Thermal Elixer, and ZeroTherm ZT-100 (upcoming)

Antec Formula 5, Formula 6, and Formula 7 (upcoming)

Gelid GC-1, GC-2, and GC-Extreme (upcoming)

Cooler Master ThermalFusion 400 and TIMTronics Grey Ice 4200 (upcoming)

Rosewill Varieties (upcoming)

SIIG Varieties (upcoming)

Thermaltake Varieties (upcoming)

Xigmatek Varieties (upcoming)

Newer Shin-Etsu Varieties (hopefully)

IC Diamond, Coollaboratory Liquid Ultra, Liquid Pro, and Liquid Metal Pad (finale)


Overall and General Performance Charts:

Best Temperature (any contact):


Contact Dependence:


Cure Factor:


Poor Contact Performance:


Moderate Contact Peformance:


Great Contact Peformance:


Contact vs. Temperature:


Overall Performance Score:


Average Temperature Weighted for Overall Performance Impression:


Average Temperature Weighted for Typical CPU Mounting Quality:


“Overall Performance Impression” takes into account all three contact settings, “Typical CPU Mount” takes into account Moderate Contact, Great Contact, and Best Mount temperatures. ”Overall Performance Impression” is an average designed to describe the overall performance of a TIM where “Typical CPU Mount” is an average designed to describe expected performance with CPU blocks and heatsinks, where contact quality tends to be good.

Last Updated September 27, 2011

[Church]

It's interesting to see how close to liquid metal type TIMs pastes can get, even worse and cheaper ones .. if only they are applied in thin as possible layer. Many stock OEM coolers do bad not just because they are worse themselves, but also imho because of very thick preapplied paste layer.
Vapor: do you think that any of those mods to increase mounting pressure are easy enough and advisable to perform on blocks? If so, maybe worth to write up some article about different mods you thought of during all your LC & testing career (eg. silicone mods and such), how stock performance of waterblocks can be improved. Also interesting is how do waterblocks line up performance wise after mods are done on all of them. Some DIYers might appreciate for any hints how to squeeze any extra performance out of them.

[tiborrr]

I've tested many TIMs in the past weeks and I've got scatter as high as 1.7°C between mounts using the same TIM and applying the same mounting pressure! The problem lies in the CPU diode itself which has the lowest resolution of 1K. It's difficult to draw any conclusions afterwards that's why I'll be switching to a load tester with probe installed.

[antiacid]

nice to see IX still leading by a fair margin!

Do you think it would perform as well when used in a sub-zero system? Perhaps some chemical compounds used in other TIMs would not perform as well when very cold but I'd think the metal in IX would conduct nicely at any temperatures. It'd be interesting to check out if you ever had the time!

[Gimmpy224]

nice to see IX still leading by a fair margin!

Do you think it would perform as well when used in a sub-zero system? Perhaps some chemical compounds used in other TIMs would not perform as well when very cold but I'd think the metal in IX would conduct nicely at any temperatures. It'd be interesting to check out if you ever had the time!

Now that I would def be curious in seeing.

Either way I now have this post bookmarked and sub'd for this year to reference ^_^ TY VAPOR!

[Church]

tiborrr: hence guys at skinneelab do many mounts per each block (per each paste this time) drop best/worst results and average the rest, to lessen mounting errors and luck impact on results. Imho still best CPU simulator is .. CPU .

[vhaarr]

Aww man I suddenly have an urge to rip out all my blocks done with MX2, flush the tube down the toilet and run to the shops for some IX

[PiLsY]

Churchy - true when block testing but with TIM's less important. As long as you can do hot spot testing its all good with thermal goo. Thermal conductivity and pressure reliance are the only things you need to test. All this is usually done in a lab - iirc laser diffraction is used to measure the thermal conductivity of most standard tim's.

None are actually designed purely for cpu usage theyre just reformulated and rebadged depending on projected temperature curve. Indigo is repackaged for cpu and gpu use but I'm pretty sure it was originally formulated for other purposes. There's not enough money in performance pc tim's to design them from the ground up for pc use. Its all just licensed product .

[gr1p]

Will be interesting to see how mx-3 and mx-4 fare as I recently switched to mx-4 on all my builds.

So far I am liking it a lot.

[Vapor]

It's interesting to see how close to liquid metal type TIMs pastes can get, even worse and cheaper ones .. if only they are applied in thin as possible layer. Many stock OEM coolers do bad not just because they are worse themselves, but also imho because of very thick preapplied paste layer.
Vapor: do you think that any of those mods to increase mounting pressure are easy enough and advisable to perform on blocks? If so, maybe worth to write up some article about different mods you thought of during all your LC & testing career (eg. silicone mods and such), how stock performance of waterblocks can be improved. Also interesting is how do waterblocks line up performance wise after mods are done on all of them. Some DIYers might appreciate for any hints how to squeeze any extra performance out of them.

If I were to do publish any mods, they would be in a block review and I'd think the mod would have to be pretty idiot proof.

Block makers are getting better at not leaving behind any 'free' performance now, too. Which is a good thing overall, IMO.

nice to see IX still leading by a fair margin!

Do you think it would perform as well when used in a sub-zero system? Perhaps some chemical compounds used in other TIMs would not perform as well when very cold but I'd think the metal in IX would conduct nicely at any temperatures. It'd be interesting to check out if you ever had the time!

All TIMs behave very differently sub-zero...most have a specific temperature where they change phase (freeze) and then perform magnitudes worse.

IX has a similar issue...it's CTE (thermal expansion coefficient...basically how much it changes shape as temperature changes) is different than that of the IHS/pot...so below a certain temperature it's bond breaks (at the microscopic level) and its performance is permanently hampered (significantly). I'm not sure what temperature it happens at, but I know if your CPU doesn't have a cold bug and you use LN2, IX isn't your best choice.

Will be interesting to see how mx-3 and mx-4 fare as I recently switched to mx-4 on all my builds.

So far I am liking it a lot.

I'm really interested in MX-4 as well. I really disliked MX-3; even though I just do a pea/dot method application, I found it hard to use. I never felt like I used the right amount and contact was always pretty bad, visually. I'm glad MX-3 was EOL'd and MX-2 was brought back from the dead, to be honest. MX-2 is so foolproof, universal, and inexpensive.

[Utnorris]

I just started using MX-4 over Artic Ceramique, while I have been pleased with the results, it will be interesting to see how good it really is in comparison to the other TIM's out there. I have a set of IX that I never used since I was swapping out CPU's almost once a month, but now that I have decided to go back to x58, it might be worth cracking the box open and using. It's a shame it isn't really reusable, but I guess if you don't swap out CPU's often then it's all good.

[gr1p]

I just started using MX-4 over Artic Ceramique, while I have been pleased with the results, it will be interesting to see how good it really is in comparison to the other TIM's out there. I have a set of IX that I never used since I was swapping out CPU's almost once a month, but now that I have decided to go back to x58, it might be worth cracking the box open and using. It's a shame it isn't really reusable, but I guess if you don't swap out CPU's often then it's all good.

Yeah thats the same reason why I haven't used it yet either.

In the future when I spend the money for a monster multi-core socket R 2011, I'll spring for an IX set.

[Martinm210]

Awesome work as always, thanks for putting so much time and detail into this!

The cure time graphs are really cool..

Martin

[LagunaX]

IX will stay king BUT it will be interesting to see how MX-4 performs.
Add Diamond IC7 and Shin Etsu X23-7783D too and your review will be complete.
IX will be top but for the vast majority of us it will come down to MX-4, IC7, or X23-7783D.

[Vinas]

Thanks for testing the Prolimatech PK1 Vapor. I remember asking you about that a while ago and am glad to see your results matched what I have experienced.

[Vapor]

IX will stay king BUT it will be interesting to see how MX-4 performs.
Add Diamond IC7 and Shin Etsu X23-7783D too and your review will be complete.
IX will be top but for the vast majority of us it will come down to MX-4, IC7, or X23-7783D.

I will never use IC Diamond again. First time I tested it was just underwhelming considering the hype it gets. Now that I'm also aware of its polishing effect...I tried a few mounts before any of this 2011 Roundup testing started and tried a different removal procedure that had zero wiping (wiping = friction = polishing.....process was citrus oil + dabbing a paper towel to remove it, extremely slow process). Despite the new removal procedure, after just 3 mounts, it had polished my IHS clean (polishing of IHS = permanent altering of the surface = inability to maintain testing continuity). I just don't see the appeal or evidence of hyped performance.

X23-7783D will definitely be interesting though, especially how it compares to PK-1. X23-7783D will be in a batch with X23-7762 and G751 later down the road.

Coolaboratories Liquid Pro/Ultra should at least match IX with Moderate and Great contact (though I wouldn't be surprised by a complete failure with Poor contact), unfortunately I won't test either of those until I'm satisfied every other TIM I'd want to test has been tested (diffusion into IHS/block = permanent altering of the surface = inability to maintain testing continuity). I have 30+ TIMs on the docket right now...will be a long time before I can get to CLP/CLU, but I hope I can.

Of all the TIMs I've never used before, MX-4 has me the most excited. PK-1 was a great surprise though, have a hard time not recommending it for someone who doesn't want to use IX/CLP/CLU.

[thegcpu]

I also get good results from Indigo Xtreme, but just be ready for a chilling reflow experience.

Here is a video of me experience.

http://www.youtube.com/user/thegcpu#p/u/7/rdVYTgqKUVg

[Hondacity]

will the gcx be included? i got some tubes unused..i can email one tube

[Callsign_Vega]

I thought Coollaboratory Liquid Ultra is the new king, even better than Indigo Extreme.

[meanmoe]

Must do shin etsu tims

[m0r7if3r]

I'd like to see AS5, just cause so many other reviews have been done with it, so it adds a pretty huge amount of universality to the testing.

[gr1p]

I also get good results from Indigo Xtreme, but just be ready for a chilling reflow experience.

Here is a video of me experience.

http://www.youtube.com/user/thegcpu#p/u/7/rdVYTgqKUVg

I just watched your video, and I have to say its way different watching it than hearing about it.

I think I'll stick to the next best thing instead of going through that reflow.

[Martinm210]

I just watched your video, and I have to say its way different watching it than hearing about it.

I think I'll stick to the next best thing instead of going through that reflow.

+1, Great video.

[gmat]

I'm curious about the latest CPU IHS performance too (specifically i7 9xx and 2600K series). Should we lap them or not for waterblock usage, using liquid metal (like CoolLaboratory thing) what change of performace does lapping gives (worse or better) ?
I'm still pondering about lapping that 990x that i'll use in my next build.

[Church]

I wouldn't lap cpu if some liquid metal or phase alloy like IX is used, though i wouldn't use liquid metal either. Don't like it diffusing into IHS & WB, even though i know it won't affect performance. In my eyes IX is liquid metal effectiveness without it's drawbacks.