[Review] Indigo Xtreme vs. AS5, MX-2, IC Diamond, Shin-Etsu X23-7783D

[Vapor]

Guess what's already starting?

Intel Round 2!

7 mounts per TIM, 25+ TIMs, 12 hours per mount. This is gonna take awhile, but it's worth it to get some great data on TIM

Oh, and I've added another wrinkle: test it all and then repeat again with poor contact. So yeah....it all looks like this: (7 mounts per TIM, 25+ TIMs, 12 hours per mount)*2

I have a second testbed dedicated to TIM testing and know it's going to take freaking forever to get done (so I'll be releasing data in chunks of 4-7 TIMs), but it's all set up and running right now, first results should be out in a few weeks

As for the inclusion of a "Poor Contact" test, I've been hand-picking parts so that I get GREAT contact on basically every mount, but that's not necessarily representative of everybody's install. It also minimizes TIM's resistance, so all the TIMs get really similar results. This will break that open. Great, great TIMs will continue to succeed when contact is bad, while others might just fail completely. I think it's really interesting to think about....and somewhere between the "Ideal Contact" and "Poor Contact" results will be the typical contact (which I don't actually planned to test).

So how am I manipulating contact but also maintaining consistency? The same way I get great contact right now: mechanical deformation + great mounting system. For my great contact tests, I use a CPU-360 (and its wonderful mounting system) with a slightly-too-thick midplate and the result is tremendous contact. I swap the midplate for one that's slightly-too-thin and suddenly contact is very poor, but still very consistent. When the plate is slightly too thin, the pressure between the block and the IHS doesn't come at the center of both objects, but more along the edges of the IHS. This creates a slight bow (inverted to the good kind) that creates a consistent gap between the two surfaces. The gap is slight, but it is consistent and it causes the bond line thickness to increase noticeably. If a TIM can fill the space and still maintain great performance, then it means it's capable of appropriately filling gaps and its bulk thermal resistance is low. It's a worst case scenario for TIM and the inept TIMs will fail, the bad TIMs will perform worse, and the great ones will be ahead of the pack. There should be a lot more variation between great/good/mediocre/bad with this test. But it's not representative of every scenario either. But by getting numbers from Ideal Contact and Poor Contact situations, we should be able to ascertain which TIMs are worth your attention and $$ and can be used in any situation.

It's gonna take awhile to get the picture painted though

At 100% efficiency (0 time to clean and remount and boot....no missed ends....every second of the day, the test is running), it will take 25 weeks, or about 6 months. I'm expecting to run at ~85% efficiency, so about 7 months from today till the end. Lots of small answers will be answered along the way, data will be released in chunks every few weeks, etc. so it won't be the last you hear from me for the next 6 months, as I also have CPU block tests to do

Anyway, hope you enjoy what's about to come

[Martinm210]

Awesome...any teasers on the new contestants?...

[Vapor]

First batch I hope to be: AS5, ASMatrix, TIM-Consultant Matrix 2, OCZ Freeze, Indigo Xtreme, and CM IceFusion

Second batch I think will be: TX-2, TX-3, TX-4, MX-2, MX-3, MX-4

Third batch I'm still deciding on

EDIT: overall, if it's a marketed product and seems popular, I'll be including it Obscure stuff, probably not.

[Sadasius]

Can we include a Mythbuster type of product in there such as.....I heard from a friend of a friend that toothpaste is the bomb for TIM.

[Alexandr0s]

Lol, you really want to kill Vapor's CPU, don't you Sadasius?

[Church]

Sadasius: it might be good joke IF tests would be done with flawed methodology like many other reviewers unfortunately do. With proper testing like Vapor does having to waste needlessly lot of time just for joke is just too much. (remember 7 mounts, few hours on each of them)

[Vapor]

Now that I have a dedicated testbed and it won't interfere with CPU block testing, I'll consider it. But a few conditions: just one non-TIM, it'll be tested after all the other TIMs, must be some evidence/myth of aptitude, and it can't be water based. You guys have ~7 months to figure out what you want tested

Thinking of re-arranging the first group....OCZ Freeze, Prolimatech PK-1, IC Diamond, Noctua NT-H1, Indigo Xtreme, CM IceFusion.

I want to do all the Arctic Silver products (Lumiere, Ceramique, AS5, ASM, TIM-Consultants Matrix 2) in its own batch I think, it can be batch 3.

Batch 1: OCZ Freeze, Prolimatech PK-1, IC Diamond, Noctua NT-H1, Indigo Xtreme, CM IceFusion.

Batch 2: MX-2, MX-3, MX-4 (hopefully available by then), TX-2, TX-3, TX-4

Batch 3: AS Ceramique, AS Lumiere, AS5, AS Matrix, TIM-Consultants Matrix 2

Some groupings of the other pastes out there (Shin-Etsus, Zalmans, Gelids, Scythes, Thermalrights, etc)

Batch final: Coolaboratory Liquid Pro, Coolaboratory Liquid Ultra, Coolaboratory Liquid MetalPad, 'Mythbuster'

[Sadasius]

Lol, you really want to kill Vapor's CPU, don't you Sadasius?

Nope, I was actually thinking of a 'Mythbuster' type of challenge for those who want to participate. Makes it much more interesting when you have broad audience participation with a study. Plus it teaches about some of the elements of today's TIM to see if/perhaps someone here can develop something better with what they have learned.

Sadasius: it might be good joke IF tests would be done with flawed methodology like many other reviewers unfortunately do. With proper testing like Vapor does having to waste needlessly lot of time just for joke is just too much. (remember 7 mounts, few hours on each of them)

No joke. I know I joke around an awful lot but this was actually serious as stated above. After all we are XtremeSystems and if anyone can make or do something better it is us right here!

Now that I have a dedicated testbed and it won't interfere with CPU block testing, I'll consider it. But a few conditions: just one non-TIM, it'll be tested after all the other TIMs, must be some evidence/myth of aptitude, and it can't be water based. You guys have ~7 months to figure out what you want tested

Thinking of re-arranging the first group....OCZ Freeze, Prolimatech PK-1, IC Diamond, Noctua NT-H1, Indigo Xtreme, CM IceFusion.

I want to do all the Arctic Silver products (Lumiere, Ceramique, AS5, ASM, TIM-Consultants Matrix 2) in its own batch I think, it can be batch 3.

Batch 1: OCZ Freeze, Prolimatech PK-1, IC Diamond, Noctua NT-H1, Indigo Xtreme, CM IceFusion.

Batch 2: MX-2, MX-3, MX-4 (hopefully available by then), TX-2, TX-3, TX-4

Batch 3: AS Ceramique, AS Lumiere, AS5, AS Matrix, TIM-Consultants Matrix 2

Some groupings of the other pastes out there (Shin-Etsus, Zalmans, Gelids, Scythes, Thermalrights, etc)

Batch final: Coolaboratory Liquid Pro, Coolaboratory Liquid Ultra, Coolaboratory Liquid MetalPad, 'Mythbuster'

7 months....SWEET! I will be working hard on this one! Thanks Vapor!

[Alexandr0s]

7 Months huh? *starts thinking*

[Sadasius]

Well after reading up on some of the materials used I have found my simple TIM for a 'Mythbuster' material.....Silver leaf. It is easily obtained in art stores and is thin enough to keep things real close plus pliable enough to get into the small imperfections of the metals to conduct the heat from one end to the other. If a square piece is cut and applied as is and then forced down with enough pressure I am sure it will be an excellent TIM on the fly or even better then the stuff we buy in tubes. This is my first submission anyway and I am actually going out to buy some and try it out.

[Church]

Even though thin but another metal layer between surfaces? I doubt it being flexible enough to fill mini-imperfections on WB/IHS, rather it might even worsen thermal interface as now you get not one but two surfaces (sides of silver leaf) where each of them is not perfect.
Imho more interesting would be test without any TIM at all, as in "can lesser temps of liquid cooling outweight worse temps of no TIM used"?

[skinnee]

You try it first and let us know how long your system holds stability, both the foil and bare.

[Biffa]

Can we include a Mythbuster type of product in there such as.....I heard from a friend of a friend that toothpaste is the bomb for TIM.

Did your friend read it here?

[Sadasius]

You try it first and let us know how long your system holds stability, both the foil and bare.

Yeah yeah I know what happens with metal on metal....Geez I am not retarded or something.

@churchy... No you don't want to try that. The CPU will heat up pretty quick and holds horrible temps. I know a reviewer who tried that and even with a waterblock as well for Bjorn3d.

Just wondering if silver leaf can actually be squished into the tiny crevices. Never tried it and never heard of anyone trying it. I have my waterblock hard mounted so it has more pressure on the CPU. If not then I seen a sample size(.5 grams) of 8nm sized graphene nanoparticles for $25. Would just need to mix it with something.

[Vapor]

Silver leaf would be interesting...I think it would be better than no TIM, but that leaves a wide range of where it could land of course

I have run no-TIM before and it's inadequate. Air cooling + TIM is way better than water cooling without TIM.

I'd ideally like to test something super inexpensive (not for my own sake, after all what's a few dollars for another 'TIM' after hundreds on real TIMs) or a commodity, since that way it could be recommended even as a super-budget option or a contingency 'free' option to someone who has it laying around. Something you'd find in a lot of garages/kitchens/bathrooms/whatever.

[Vapor]

Just did a quick one mount test of a potential future market product that makes me think silver foil could work

Anyway, OCZ Freeze has all 7 ideal mounts done, onto PK-1. Then maybe a mount or two of IceFusion (great as a reference of generic TIM), then off to Taipei for a few days. Get back, finish IceFusion, move to IC Diamond, then Noctua NT-H1 and IX, publish results in a new thread, then rinse and repeat with poor contact

[Church]

Vapor: Try russian made KPT-8 (КПТ-8) TIM paste. Once it was on par with other old pastes available then, but these days of course it's performance is not as best pastes arround. But it's still very popular in ex soviet union and suggested by some veterans and made/sold by several firms, simply because it's dirt cheap. You can even buy that in litres, not grams . Imho no other paste can best in price/performancy. Not the worst choice for some cheap OEM PC assembling where every $ counts.

[Grinchy]

Really impressed by the amount of work you did...i too noticed as well that anyway the measured variance between obtained results is very high...
Cani ask you one thing? Arctic silver is supposed to have a >200 hours curing time, don't you think that you didn't let it have the time to express its full potential with just few hours of cure time?
I have read another review where arctic silver 5 behaves slightly better than Shin-Etsu X23-7783, when tested with the right curing time.In this review i also found that Shin-Etsu MicroSi G751 performs better than x23-7783d and has the upside of not being as thick ,being so more multipurpose use than its thicker brother...

http://benchmarkreviews.com/index.ph...&limitstart=12

[mlee49]

Has there been any mention of using Gelid's Extreme thermal paste?

GELID Solutions GC-Extreme Thermal Compound

http://www.newegg.com/Product/Produc...-020-_-Product

[Grinchy]

Has there been any mention of using Gelid's Extreme thermal paste?

GELID Solutions GC-Extreme Thermal Compound

http://www.newegg.com/Product/Produc...-020-_-Product

that's been reviewed here together with other 80 tims:
http://benchmarkreviews.com/index.ph...1&limitstart=1

[Vapor]

Really impressed by the amount of work you did...i too noticed as well that anyway the measured variance between obtained results is very high...
Cani ask you one thing? Arctic silver is supposed to have a >200 hours curing time, don't you think that you didn't let it have the time to express its full potential with just few hours of cure time?
I have read another review where arctic silver 5 behaves slightly better than Shin-Etsu X23-7783, when tested with the right curing time.In this review i also found that Shin-Etsu MicroSi G751 performs better than x23-7783d and has the upside of not being as thick ,being so more multipurpose use than its thicker brother...

http://benchmarkreviews.com/index.ph...&limitstart=12

Arctic Silver's pastes are interesting...all their retail pastes 'require' 36-300 hours of break-in before final performance is found, but pressure and temperature seem to shorten it significantly. I'm interested in the long-term performance of the Arctic Silver pastes, but I'm not about to move the goalposts for them nor am I convinced that the numbers I see at the end of a 12 hour load aren't indicative of nearly-final performance.

As for other reviews out there, if I thought there was already good data out there, I wouldn't spend this much time on testing it myself.

G751, 7783D and Gelid Extreme are all on the docket (those in particular are already in house)

Basically, if it's a TIM that's for sale at one of the major stores that I have access to, I'm going to try to include it

[Grinchy]

Arctic Silver's pastes are interesting...all their retail pastes 'require' 36-300 hours of break-in before final performance is found, but pressure and temperature seem to shorten it significantly. I'm interested in the long-term performance of the Arctic Silver pastes, but I'm not about to move the goalposts for them nor am I convinced that the numbers I see at the end of a 12 hour load aren't indicative of nearly-final performance.

As for other reviews out there, if I thought there was already good data out there, I wouldn't spend this much time on testing it myself.

G751, 7783D and Gelid Extreme are all on the docket (those in particular are already in house)

Basically, if it's a TIM that's for sale at one of the major stores that I have access to, I'm going to try to include it

Yes it is true that there's good data out there, but your testing method that keeps in consideration more mounts is great and it would have been nice to have confirmation of those datas... you can't spend 200 hours just for as 5 but maybe you can give it some more time than other tims to see if indeed its temps stabilize or not...
There are many clones of AS5, same tim just with rebranded name.Antec produces one(Antec 77063 Formula 5 silver grease) and does not reccomend the curing amount of time reccomended by ARctic .SO testers tested it without the right amount of curing time and the result is that it performed much more poorly than AS5 with same testing conditions(besides curing time)... In your datas it is clear that the AS5 curve for temps/curing time is still going down while all the others are stabilized (maybe only ic diamond is still going down a little even if not like AS5).
So i think that to be fair toward a product it should be tested like advised by the manufacturer.
Another thing that i have to say is that the single drop or the single line of tim are not the best way to test a tim. Usually the best way is two parallel lines one third of cpu width as is tested and described in that article that i quoted. ANyway i am eagerly looking forward to updates It is going to be a monumental work .Again compliments

[jalyst]

So aside from Liquid Pro -which obviously has it's own usability issues...
Is IX still regarded as the best performing TIM, or have other OEM's now copied Enerdyne?

@Vapor
Have you still got a new round-up planned?
If it's done I don't suppose you could link to it in this thread?

*edit*
Doesn't even look like IX exists now, according to Enerdyne's site
http://www.enerdynesolutions.com/prod_indigo.html
Has IX been re-branded to Indigo 2? (I2)

[Vapor]

The various metal TIMs (Indigo Xtreme, Liquid Pro, Liquid Ultra, and possibly Liquid MetalPad) are surely in a class of their own still, and I'm not sure we'll see pastes match their performance for a long time, if ever.

New round-up is already in progress but it's hard to say when it'll all be done (probably May or later), but the first chunk of the round-up will be done within 4-5 weeks, I think. Indigo Xtreme, Prolimatech PK-1, Noctua NT-H1, OCZ Freeze, IC Diamond, Spire SilverGrease, and Cooler Master IceFusion are in the first batch. They'll all be tested with three different contact 'settings' (great contact, moderate contact, poor contact) and 5 mounts per contact config (so fifteen 12 hour mounts per TIM). Testing is about 30% complete on this batch.

IX still exists...look around at various retailers or at their website: http://indigo-xtreme.com. Indigo 2 is their industrial custom-order application, AFAIK. Indigo Xtreme is their enthusiast product.

[jalyst]

Thanks for confirming IX is still the best if one doesn't want the fuss of Liquid Pro/Ultra...
And thanks for clearing-up the difference between Indigo 2 and IX, I have since ordered some IX from a local retailer.

Wow it sounds like you're going into even greater depth this time round....
I look forward to hearing about your new results in this thread in May 2011!

I'm not sure what the community would do without users like you....
Users whom are prepared to spend their time doing such detailed comparative analysis, for no financial gain.

I wish you the best of luck and good health!