Amazing differences between AS5 & CLP! [Archive]

Katzenschleuder

07-28-2006, 06:17 AM

Hi,

I am really impressed by the improvements I got only by switching to Coollaboratory Liquid Pro thermal paste:

Opteron 144 @ 2.96 GHz (1.5 V)
Ambient temperature for all tests: 24 °C
The results are measured with full CPU load
Each combination was tested two times in order to make sure that there is an optimal contact between the CPU and the cooler.

with IHS & AS5: 43 °C
with IHS & CLP: 36 °C
without IHS & AS5: 41 °C
without IHS & CLP: 34 °C

This is a difference of 7 K, almost for free!
I was able to improve the clockrate to 3.12 GHz (same voltage) because of this.

Very impressive in my opinion.
I find it surprising that many people over here still use the old AS5.

creidiki

07-28-2006, 06:22 AM

I use ceramique. I'm not putting that stuff in my pc.

hint: have you seen what it does to alu? want to take a guess at what the pins on the SIMD resistors on your CPU are made of?

Katzenschleuder

07-28-2006, 06:37 AM

I use ceramique. I'm not putting that stuff in my pc.

hint: have you seen what it does to alu? want to take a guess at what the pins on the SIMD resistors on your CPU are made of?This isn't aluminium but a tin-alloy.
But anyway! I am no noob who spreads thermal paste all over the CPU package.

There are absolutely zero problems involved if you know how to apply thermal paste.
You have to coat the die with a microscopical thin layer using a razor.

Ceramique has a much worse thermal conductivity than CLP, or even AS5.

creidiki

07-28-2006, 06:41 AM

And yet my load temps are unchanged from when I used AS5.

buff

07-28-2006, 06:42 AM

where can this stuff be bought?
Couldn't hurt to try.....

Ominous Gamer

07-28-2006, 06:44 AM

I use ceramique. I'm not putting that stuff in my pc.

hint: have you seen what it does to alu? want to take a guess at what the pins on the SIMD resistors on your CPU are made of?

Since when you put thermal paste on anything but the core? CLP has been proven to give great results. The fact that it eats up weak metals isn't an issue to take up againist CLP. Its the fact that people are using weak metals as their coolers.
I however still want to see someone use this stuff in some vapor systems.

Nonetheless, this thread seems like a well placed advertisement.:rolleyes:

where can this stuff be bought?
Couldn't hurt to try.....
http://www.xoxide.com/coollaboratory-liquidmetal-thermal-interface-material.html $13.99

[XC]melymel

07-28-2006, 06:47 AM

What heatsink was used and what are load and idle temps? also what are you using to meausre temps?

I really can't see a 7C difference being possible just for swapping paste over...

:toast:

Katzenschleuder

07-28-2006, 06:47 AM

I don't want to persuade here anyone.
I just wanted to share the results, which made me very happy.

Marci

07-28-2006, 06:56 AM

The main point to make here, is that there hasn't been a single test where the CLAMPING FORCE OF THE HEATSINK OR WATERBLOCK has been measured and 100% guaranteed replicated at every mount. Without that control factor, ALL results measured are USELESS.

FACT.

Each combination was tested two times in order to make sure that there is an optimal contact between the CPU and the cooler

For empirical testing, you need to replicate the mount at least 20 times to elliminate variances, then look at the averages to work out how perfect the mount is across the range of results for each compound tested. Twice isn't enough to allow one to make any conclusive / beleivable statements calling one better than the other. This was covered at ProCooling years ago...

If you must, the simplest way to produce remotely beleivable results... head down to yer local toolshop and get a bit for a torque wrench that matches your waterblock's mounting nuts. Do them up with the torque wrench so that each spring / nut is identical. Take your test results. Unmount the block, clean off the compound, reapply the same compound, remount the block, retest. Do this around 5 times. Check consistency of results. Repeat for the "comparison" compound.

Ideally you also need some very good resolution weighing scales or an alternative way of measuring the amount of compound used to ensure the same amount is used every time. IE: Weigh the tube before you start. Make the first application. Weigh the tube again. Work out the difference. On all subsequent applications, go slowly, weighing frequently, until the same weight difference is achieved, thus at least ensuring a roughly identical AMOUNT of paste is used.

If someone (Radiical??) is prepared to test to THAT extent, then a statement can be made with regards performance that can be taken as gospel until proven otherwise. But either way, the testing methodology above is still VERY inaccurate and does not take into account all factors that can influence the results.

Katzenschleuder

07-28-2006, 07:06 AM

What heatsink was used and what are load and idle temps? also what are you using to meausre temps?Sorry, I didn't measure the idle temperatures (I don't care about them).
I am using a Cuplex Evo 1.1.

I really can't see a 7C difference being possible just for swapping paste over...

:toast:Well, it's the fantastic reality. :)
There are many other tests out there with similar results.

Thrilla

07-28-2006, 07:09 AM

Read my sig if you want to see what it does to aluminium :P (I do remember someone saying even liquid metal evaporates, ie mercury, so the long term effect might not stay the same)

Liquid metal is a good stuff, I've been using my home made for a long, long time now, the stuff does work better than AS5, in my case as much as 8'C on load. Liquid metal is also acts very fast, the point it that you want to fill microscopic valleys not to put a layer of insulation between the core and the HS, so idealy as little TIM as possible is the best (under the condition that it covers the entire heatload).

Edit: hold up it's no longer in my sig i'll find it
anyway here it goes
http://www.xtremesystems.org/forums/showthread.php?t=100168&highlight=war+liquid+metal

buff

07-28-2006, 07:13 AM

I might be able to grab a torque wrench and test it, but with aircooling......
Gotta get the CLP first

Grinch

07-28-2006, 07:24 AM

I'll stick with AS5 and be happy...:toast: :woot:

Katzenschleuder

07-28-2006, 07:24 AM

The main point to make here, is that there hasn't been a single test where the CLAMPING FORCE OF THE HEATSINK OR WATERBLOCK has been measured and 100% guaranteed replicated at every mount. Without that control factor, ALL results measured are USELESS.

FACT.

For empirical testing, you need to replicate the mount at least 20 times to elliminate variances, then look at the averages to work out how perfect the mount is across the range of results for each compound tested. Twice isn't enough to allow one to make any conclusive / beleivable statements calling one better than the other. This was covered at ProCooling years ago...

If you must, the simplest way to produce remotely beleivable results... head down to yer local toolshop and get a bit for a torque wrench that matches your waterblock's mounting nuts. Do them up with the torque wrench so that each spring / nut is identical. Take your test results. Unmount the block, clean off the compound, reapply the same compound, remount the block, retest. Do this around 5 times. Check consistency of results. Repeat for the "comparison" compound.

Ideally you also need some very good resolution weighing scales or an alternative way of measuring the amount of compound used to ensure the same amount is used every time. IE: Weigh the tube before you start. Make the first application. Weigh the tube again. Work out the difference. On all subsequent applications, go slowly, weighing frequently, until the same weight difference is achieved, thus at least ensuring a roughly identical AMOUNT of paste is used.

If someone (Radiical??) is prepared to test to THAT extent, then a statement can be made with regards performance that can be taken as gospel until proven otherwise. But either way, the testing methodology above is still VERY inaccurate and does not take into account all factors that can influence the results.Dude, I don't have the time for such a bull:banana::banana::banana::banana:.
All I wanted to share here is the experience I made. I don't care what you think would be neccesary for a evidence or whatever!

Fact (or as you say: FACT) is that I saw amazing improvements.
Fact is that other users saw amazing improvements.
Fact is that professional hardware reviewer saw amazing improvements.

creidiki

07-28-2006, 07:31 AM

uncotrolled tests != facts.

I can take AS5, put my block on weakly, and test @ 28c ambeint, then take :banana::banana::banana::banana:ty thermal paste, mount tight, test @ 20c ambeint and shout "amazing results!"

we're all glad that youre happy with it, but facts they are not.

lets not even start on the quality and professionalism of 99% of thermal product reviews.

Katzenschleuder

07-28-2006, 07:46 AM

uncotrolled tests != facts.What an idiotic quarreler!
The fact is that I/we saw these results!
Don't accept it, whatever! I don't care!

Unbelievable what smartasses are around here.
I will never waste my time here again!

Butcher_

07-28-2006, 07:51 AM

I use ceramique. I'm not putting that stuff in my pc.

hint: have you seen what it does to alu? want to take a guess at what the pins on the SIMD resistors on your CPU are made of?
Plated copper like as not - that's what most pins / leads on components are.

Torin

07-28-2006, 07:51 AM

lets not even start on the quality and professionalism of 99% of thermal product reviews.
Key point.

We're all glad you see an improvement. Let's leave it at that, because no one can prove empyrically one way or another, much less whether a 7C difference is accurate. Take off the CLP, reapply AS5, and see if you get the same results as your original AS5 results. I'd put money on it that you don't.

There are many factors that affect CPU temps when it comes to mounting a heatsink, and this is a biased test regardless of how you perform it, probably for reasons you don't even realize.

EnJoY

07-28-2006, 07:56 AM

What an idiotic quarreler!
The fact is that I/we saw these results!
Don't accept it, whatever! I don't care!

Unbelievable what smartasses are around here.
I will never waste my time here again!

And we don't need people who can't take criticism (especially as constructive as it was) on their results or posts. Allow me to be the first to say, don't let the door hit ya on your way out... :rolleyes:

xXxDieselxXx

07-28-2006, 08:01 AM

Guys don't trash this guy.. .he is simply excited as all of us are when we see some progress... also he obviusly don't have the time or will to do such extensive study. If I was him with those temps I won't even touch that sink anymore :D

In my view the facts are:

1- AS5 Has been used for quite some time and by many users. The majority happy with it. Tons of reviews have been made.

2- Ceramique has also been popular and performs almost like AS5 does and picked by many for VGU cooling for the "non conductive" properties.

3- AC MX1 has also shown it's great paste and some people swear its better than AS5.

So basically where are the facts? I can find you reviews where this paste is better than this one and so on.

My 2 cents here is, maybe those who feel like trying something new should give it a shot to the CLP and log more results.. so it can be known better by the cooling community.. keeping in mind the metal corrosion issue.

Guys this is the cooling area so be cool ...:toast:

Butcher_

07-28-2006, 08:07 AM

Hi,

I am really impressed by the improvements I got only by switching to Dihydrogen Monoxide (DHMO) thermal paste:

Opteron 144 @ 2.96 GHz (1.5 V)
Ambient temperature for all tests: 24 °C
The results are measured with full CPU load
Each combination was tested two times in order to make sure that there is an optimal contact between the CPU and the cooler.

with IHS & AS5: 43 °C
with IHS & DHMO: 28 °C
without IHS & AS5: 41 °C
without IHS & DHMO: 26 °C

This is a difference of 11 K, almost for free!
I was able to improve the clockrate to 3.12 GHz (same voltage) because of this.

Very impressive in my opinion.
I find it surprising that many people over here still use the old AS5, especially given how easy DHMO is to obtain!

For more info on this miraculous substance: http://www.dansdata.com/goop.htm

Katzenschleuder

07-28-2006, 08:11 AM

And we don't need people who can't take criticism (especially as constructive as it was) ROFL!
Yeah! Very constructive to tell me that my post is "USELESS" and would have to measure the clamping force, mount everything 20 times and whatever!

VERY constructive!

Bye bye...

Torin

07-28-2006, 08:19 AM

Bye.

Remember.... not everyone on the Internet is going to agree with you.

Ominous Gamer

07-28-2006, 08:20 AM

Marci

07-28-2006, 08:21 AM

Anyone who posts on this subject is gonna get exactly the same response (and many already have) until PROOF is provided. And the only way to provide sufficient proof that the change to CLP alone is the cause of the temp drop is to test it properly. Simple as.

"Constructive" is viewed as - advising how to do what you've done PROPERLY due to flaws in your testing methodology. Anything else is just wafting a flame / ego. My post was perfectly constructive. Your assessment of it is just plain destructive to the nature of product reviewing / comparison and promotes the heresy that abounds all around the net on review sites - hence they are ignored and folks turn to forums containing knowledgeable people.

When someone brings on proof, all will beleive. Til then, it's assumption, and everyone knows the rule that comes alongside assumption...

Fact is that professional hardware reviewer saw amazing improvements.

Where? A professional reviewer lists full methodology, provides full results, and controls the variables that can influence those results. I'm waiting to see this review by a professional... got a link?

EnJoY

07-28-2006, 08:23 AM

ROFL!
Yeah! Very constructive to tell me that my post is "USELESS" and would have to measure the clamping force, mount everything 20 times and whatever!

VERY constructive!

Bye bye...

Lol, I wonder how you act when don't do a homework assignment properly and your teacher corrects you. Bet you've been tossed out of many a class.

Cheers to your ignorance and your poor attitude towards reversing it. :toast:

septim

07-28-2006, 08:49 AM

"Sorry, I didn't measure the idle temperatures (I don't care about them)."

not really a nice way to put it.
not all of us live in cold climate areas,
ambient room temp matters, case temp matters
tsk tsk...

buff

07-28-2006, 11:42 AM

Dihydrogen monoxide?
you mean water?

hehehe

eXa

07-28-2006, 12:25 PM

hehe. i dont think that was an serious post... ;)

lv_dicedealer

07-28-2006, 12:42 PM

Wow, talk about someone who can't take criticism.

You would have thought that someone who was reviewing a thermal paste and was in the 'industry' would have known who Marci was..... I guess you can find the next wonder product on flea-bay ;)

Butcher_

07-28-2006, 05:05 PM

Where? A professional reviewer lists full methodology, provides full results, and controls the variables that can influence those results. I'm waiting to see this review by a professional... got a link?
I wasn't aware there were any professional reviewers out there. :p

Butcher_

07-28-2006, 05:05 PM

Dihydrogen monoxide?
you mean water?

hehehe
That's the common name of it, yes. ;)

Sneaky

07-28-2006, 05:34 PM

don't care how much better CLP is than any other TIM out there... wouldn't touch the stuff with a 3ft pole

creidiki

07-28-2006, 05:36 PM

Enough arguing. I declare shenanigans on the whole thing.

Now give me cake ^_^

SiGfever

07-28-2006, 07:03 PM

Enough arguing. I declare shenanigans on the whole thing.

Now give me cake ^_^
Hey, can I get some ice cream with that cake?

Fairydust

07-28-2006, 07:45 PM

I agree partly that without proper testing procedures, the results are not as meaningfull as proper reviews from trustworthy sources. However in the absense of those, these results are the only indicators of performance available and and not completely useless (almost but not quite).

Dunno how many threads on the subject we need though, my initial thread (http://www.xtremesystems.org/forums/showthread.php?t=72814) was looking more torwards the corrosive quallity.

septim

07-28-2006, 08:06 PM

vote to close thiis thread...

dinos22

07-28-2006, 08:30 PM

i would like to see marci test this one :p:

@thread creator........take it easy champ ppl here don't take anything for granted..........especially when someone shows up with such a massive difference in temps

Vapor

07-28-2006, 09:05 PM

I'd like to see Marci test this one as well...I use CLP for above-ambient cooling and Lumiere for sub-ambient....won't touch AS5 anymore (and have never touched Ceramique...except maybe once now that I think about it).

I have very similar results as the thread starter, just with my XTX and with three mountings of CLP and two of AS5. Denny also had great results with CLP. Though none of us have had truly empirical testing, we all have had consistnently acheivable improvement. I'm sorry some of you can't trust a user's experience....yes the whole story isn't told (well, it may be....but the whole story is rarely told because the whole story is rarely noted/remembered) and the tests weren't scientific in any way, but this isn't the first time that CLP has shown a great improvement.

BTW, due to the nature of CLP, the mount has to be PERFECT for it to work. AS5/Ceramique/Lumiere/AS3, not so much due to the goopiness/filling properties....this can lead to worse temps on those 4 while still being functional, so the concerns are pressing. I know with my two AS5, they were visually perfect. (and with the MCW-60 rev2, kinda hard not to be consistent :-/)...ATi was permanently stamped into the block on the first mount too.

I know we're all rooting against CLP due to its dangerous nature (I mean, c'mon, it eats aluminum like a fat kid eats cake!), but no one has yet to show anything BUT a consistent improvement.

NickS

07-28-2006, 09:06 PM

I use ceramique. I'm not putting that stuff in my pc.

hint: have you seen what it does to alu? want to take a guess at what the pins on the SIMD resistors on your CPU are made of?

Yeah, I have a nice 22G of Ceramique :D

Timmay

07-28-2006, 10:13 PM

Nonetheless, this thread seems like a well placed advertisement.:rolleyes:

http://www.xoxide.com/coollaboratory-liquidmetal-thermal-interface-material.html $13.99

Yet you just quoted a store on which to purchase it from:nono:

Ominous Gamer

07-28-2006, 10:32 PM

Yet you just quoted a store on which to purchase it from:nono:
Only after someone asked:slap:

Radical_53

07-29-2006, 01:17 AM

Well, maybe there is a test:

http://watercoolplanet.de/index.php?open=2&show=115

100W heat load, 24kg clamping force.

http://watercoolplanet.de/gfx/artikel/coollaboratory/Coollaboratory%20vs%20AS5%20-%202.jpg

They gained ~1.5K when switching from AS5 to CLP. With pure Gallium, there were even further gains, at the expense that this stuff isn't liquid at room temp :)

http://watercoolplanet.de/index.php?open=2&show=118

xXxDieselxXx

07-29-2006, 04:32 AM

Interesting.... I don't understand German but I take it as you don't have to cover the whole block correct? That thing will expand more I'm guessing.

:banana: :banana: One more vote for Marci to test it :banana: :banana:

http://watercoolplanet.de/gfx/artikel/gallium/gallium4.JPG

Radical_53

07-29-2006, 05:19 AM

Well, you don't put your normal paste on the whole block, right? :) Just the IHS/die...

septim

07-29-2006, 09:52 AM

is that like putting too much could would short out the cpu...

eXa

07-29-2006, 09:59 AM

if your putting paste on the entire block and\or put on so much that you actually short out the cpu, then you should consider buying a complete machine from dell or alienware or something...

Radical_53

07-29-2006, 10:39 AM

Well, you can put it on the block plus the chip. But you normally don't have to. I normally put it on the chip, take off the paste that's too much with the syringe, put the block on, maybe take if off afterwards and wipe the part that's on the block with a paper towel. That spreads it over the contact area of the block and pre-fills small gaps.
It's quite different to handle compared to normal paste, but once you know how it's really easy.

SiGfever

07-29-2006, 10:41 AM

It was like chasing a BB around until I got the hang of spreading it. :D

Clay

07-29-2006, 10:46 AM

Very interesting thread. The above pic looks like they used a template for spreading tim,of some sort. Coolermaster heatsinks used to ship a template-spreader simular to the pic above.Except thiers would cover most of the bottom of the sink if you bothered to use it. Probaby a very good idea for testing like that. Nice test rig!
Edit: No I see now they used a paint brush someone has a nice touch.
Interesting results. still I'll be stickin' with my ol' as5 til' more test are in and safer compounds are released. Not that as5 is really all that safe.(if anyones ever got it in the bridges of an organic bodied sckt462 they'll know what I mean LOL)

Fuzzy_3D

07-29-2006, 12:20 PM

if your putting paste on the entire block and\or put on so much that you actually short out the cpu, then you should consider buying a complete machine from dell or alienware or something...

Its worse than that, you'd need to step to a Mac.

Drow

07-29-2006, 12:50 PM

:off:
Apple is good stuff. :fact: :worship: :up:

creidiki

07-29-2006, 02:39 PM

I'd buy a mac right away if I had the spare cash...

[XC]melymel

07-29-2006, 02:49 PM

maybe he's referring to the mac guide of putting TIM on like ice cream when he made the mac comment :lol: there was a thread about it recently. :toast:

Butcher_

07-30-2006, 03:21 PM

Not sure I'd call watercoolplanet.de a reliable source... 1.5K seems a bit more like the difference I'd expect though.

Radical_53

07-31-2006, 01:01 AM

Well, so far their results were quite reliable. You might see other differences "at home" between those two pastes, but that's quite typical.

Marci

07-31-2006, 03:21 AM

Heh - can all vote for me to test it, but I don't have the gear to do so... so it ain't gonna happen. Back to the age old problem... define & quantify the heatsource... for a true test, various heatloads need to be tested and a curve produced to discover whether the performance gain is a constant, or whether it performs better (over an alternative product) at higher heatloads than lower heatloads etc...

I can see I'm gonna have to learn to read German rapidly judging on the past week or so...!

Radical_53

07-31-2006, 10:32 AM

Yes, tests with different blocks, chips (referring to heat and size) and pastes would be good.
Overall, as from what I've seen myself and on those tests, CLP was superior every time. Within margins of 2-5°, better than AS5 or other good paste (differs as far as I know from block to block which standard paste is best).
Downside to a real test would be application, right? It's quite easy to spread a thin layer of CLP (as far as I used it up to now, you can't put a thick layer of this stuff on a chip, it always gets thinned out some time), but it's always hard to get a really thin (and equally thin) layer of normal goo.

tw33ter

07-31-2006, 11:34 AM

What's the difference between spreading it out beforehand and just putting a little dab on it and letting the heatsink spread it out itself? Seems like you'd be more prone to get pockets of air if you prespread it.

Torin

07-31-2006, 12:26 PM

Butcher_

07-31-2006, 02:03 PM

Also scenario 2 you hope that the mounting pressure is high enough to spread the goop evenly. That's often not the case.

creidiki

07-31-2006, 02:09 PM

And yet, with AS5/Ceramique it often gives you better temps because you dont get trapped air.

buff

07-31-2006, 04:03 PM

I just got some CLP.......it seems to act like mercury.......

anyways, I applies real easy and spreads very thin.
On my X1800xt, normal idle temps with:
AS5-- 57-60
CLP-- 54-56

seems to help.
I havn't had a chance to do some load tests yet

tw33ter

07-31-2006, 05:09 PM

Also scenario 2 you hope that the mounting pressure is high enough to spread the goop evenly. That's often not the case.

but couldn't one also say that you hope there's enough mounting pressure to contact the prespreaded TIM if you spread it too thin?

[XC]melymel

07-31-2006, 05:21 PM

but couldn't one also say that you hope there's enough mounting pressure to contact the prespreaded TIM if you spread it too thin?

kinda yes but mainly No :p:, providing you spread it correctly you only put enough TIM to fill in the cracks and make it a level surface to give the best heat transfer. If it is correctly spread the surface will be level and you know then that as long as the Hetsink is making some contact it is making even contact so you don't need to put too much pressure in, just enough to assure the heatsink is making enough contact with the IHS by feeling slight resistance. However if you are using the heatsink to spread the TIM you don't know how much pressure you need and you can't be sure of a good mount just by feeling slight resitance. You may have put a lot of pressure in and feel some resistance but still not spread the TIM well enough and it will be sitting slightly raised on the TIM in one place and not making contact with the IHS in another. :toast:

creidiki

07-31-2006, 05:23 PM

Thats assuming that the IHS is flat (its not) or that dies are flat (theyre not).

Vapor

07-31-2006, 05:24 PM

What about lapped surfaces? How would the different consistencies and spread techniques be on those?

clokker

07-31-2006, 05:52 PM

What about lapped surfaces? How would the different consistencies and spread techniques be on those?
Define "lapped".
On truly "lapped" surfaces you surface both mating parts to each other and no compound would be necessary at all.
As generally used on tech fora, "lapping" really only means "making semi-flat" so an intermediary bridging material (i.e., AS5) is required to achieve acceptable surface contact.

tw33ter

07-31-2006, 05:56 PM

What about lapped surfaces? How would the different consistencies and spread techniques be on those?

I dont remember where I saw it, maybe someone can help me out here, but 2 perfectly flat surfaces, when pressed together will bond permanantly.

creidiki

07-31-2006, 06:01 PM

Theyd have to prefectly flat on a quasi-atomic level and of metarials that will form atomic bonds with each other, be they ionic or covalent or whatnot.

Most metals (certainly copper & alu & steel) form a layer of oxide as soon as they come into contact with air, so thats out of the question in the real world, as the oxides are pretty much chemically inert.

Smalltimer

07-31-2006, 08:28 PM

Well I for one have also had a good experience with the liquid metal.

I can't say for sure that its "that much better" than the regular leading brand, however I think much of the success is attributed to the fact that its easier to apply and yields better thermal transfers whereas the leading brands require more care and preparation.

I was using AS5 prior to liquid metal and I mounted my water block at least 7 times with consistent temp results(yes it set) however when the liquid metal arrived I lost 4c under load right off the get go. So in my case it was a better overall thermal paste solution.

I also read several people stating that it can ruin components due to its corrosive/reactive nature. To this I say rubbish! water cooling is equally dangerous and there are hordes of people running liquid in there cases. The stuff is perfectly fine as long as it's done with care and a little common sense.

So in my case its two thumbs up for the liquid metal thermal compound.