The Great Thermal Interface Material Con

Jokester · 23 Sep 2006 at 14:43

We've all heard stories about how expensive TIMs are a waste of money and the bog standard stuff is just as good, hell even toothpaste or marmite will do a job just as good.

But is there any truth to this?

Well I thought I would test them to find out, so I did a head to head between:-

Using the following test set-up:-

4800X2 AM2 socket
Asus M2N32-SLI MB
Thermaltake Big Typhoon

The setup was placed on the floor to minimise temperature fluctuations and the room window was fully open to ensure proper ventilation.

First test was at stock (2.4GHz) using 1.5V to generate a bit of extra heat to emphasise the effect of the TIM. First up the AS5 (as originally installed), a 15min S&M FPU test run produced the following peak load and idle temps:-

Load 49C
Idle 26C

HS off showing coverage of AS5:-

Next test was to use the toothpaste, it was surprisingly easy to apply and a even spread was easily achieved using the credit card method:-

Jokester

Jokester · 23 Sep 2006 at 15:03

Load 60C
Idle 29C

Using the toothpaste made a little difference to the idle temps but quite a significant difference to the load temps (11C).

Following this I thought about seeing how far I could overclock using the toothpaste. Knowing the max stable overclock achieved on this chip was 3GHz at 1.45V I restored these options and re-ran S&M:-

Load 62C
Idle 29C

CPU coverage:-

Test was re-run with AS5 using the same settings as a direct comparison:-

Load 52-53C
Idle 27C

Again a significant difference between the two of the order of 10C. Sadly this chip won't go over 3GHz regardless of cooling so I wasn't able to ascertain how much of a difference this would make to an overclock.

One thing I was pleasantly surprised with was that cleaning off the toothpaste left the CPU spotlessly clean:-

So in summary, paste does make a difference, if you're overclocking you might as well use the best you can get, if you're just using it at stock just use the free TIM that comes with it, or if you need to remount at some point, just use toothpaste!

Jokester

BillytheImpaler · 23 Sep 2006 at 15:13

Try it with no TIM, Jokester. I'm interested in seeing the difference between using a TIM, even a crazy one, and no TIM at all.

Mrchu_ · 23 Sep 2006 at 15:19

LMAO that is so funny

I suppose just in case you were righteously preeching the postive cooling effects toothpaste had.

{But actually youve never cooled anything before in your life,and missheard your mate saying he cleans a very important part of the cpu to make it shinier & cooler ;the TIM, with colgate.

That blew you away,and is the only bit of the conversation you remember}

MikeTimbers · 23 Sep 2006 at 15:56

Your numbers are interesting as they are different from those achieved by the infamous Dan who found little to no difference. You really need to do the same tests several times to remove the possibility of a bad mount affecting the results.

Jokester · 23 Sep 2006 at 16:05

MikeTimbers said:
Your numbers are interesting as they are different from those achieved by the infamous Dan who found little to no difference. You really need to do the same tests several times to remove the possibility of a bad mount affecting the results.

I've always considered his results as being bull. There's been numerous other test carried out using a variety of compounds (admitadly not toothpaste or marmite) and they seem to always place AS5 ahead by a degree or two in front of Ceramique which is a degree or two in front of the rest.

Jokester

Edit:- I'll re-run the test with the toothpaste at the same settings as the AS5 and no TIM that I've just done.

Jokester · 23 Sep 2006 at 17:17

Testing at 2.4GHz and 1.4V

AS5-

Load 42-42C

Toothpaste:-

Load 48C

None at all:-

Load 55C

And then rerun the 3GHz at 1.45V with the toothpaste to check repeatability.

Load 62C

Jokester

jpe20 · 23 Sep 2006 at 19:43

nice thread!! ... read in PC format that there is not really a huge difference (1-2C) between the various thermal compounds. I would go for one that is cheap and easy to remove. Having said that I still have a nearly full tube of AS5 to play with.

Cheapest solution I can think of is vegtable oil!!!! I dunno whether it is too thin to use but I am sure it would be more effective than toothpaste!! At 60p for a 1litre bottle veg oil maybe a nice cheap option. It is flamable I suppose though I expect so is thermal grease!!!!

gurusan · 23 Sep 2006 at 20:54

very interesting

VeNT · 23 Sep 2006 at 20:56

jpe20 said:
nice thread!! ... read in PC format that there is not really a huge difference (1-2C) between the various thermal compounds. I would go for one that is cheap and easy to remove. Having said that I still have a nearly full tube of AS5 to play with.

Cheapest solution I can think of is vegtable oil!!!! I dunno whether it is too thin to use but I am sure it would be more effective than toothpaste!! At 60p for a 1litre bottle veg oil maybe a nice cheap option. It is flamable I suppose though I expect so is thermal grease!!!!

oils a bad idea as if it gets too hot it will harm your components, also its not realy viscus enough for it to be applyed well

Hades · 23 Sep 2006 at 21:43

Quality. Have 5 stars

jpe20 · 23 Sep 2006 at 21:53

It was only a silly idea after all!!

wizardmaxx · 23 Sep 2006 at 22:13

Would silicone grease work. Just wondering cos its cheap and doesnt dry out. Dont know what thermal conductivity would be like.

Also copperslip mabye

w3bbo · 24 Sep 2006 at 08:35

Nice experiment Jokester!

I would like to see are more indepth test of a few known brands of thermal paste though as only the real hard up are going to use toothpaste imo when a cheap grease costs £2. Brands such as mx-01, akasa, zalman, liquid pro etc showing the idle load/ differences under a controlled environment (constant ambient temps)would be great. Toothpaste is all good and well but I would wager that it would dry out in no time leaving a thermal barrier rather than an interface after a week or so - maybe less. Whereas proper thermal grease will still be 'greasy' for months on end.

I trust forum reviews rather than website reviews as such websites tend to be biased to the products they sell

N1 Jokester.

Mitch007 · 24 Sep 2006 at 08:38

the thermal conductivity of the thermal compounds is way way lower that that of the metals they sit between (in this case heatsink and chip). The real contrubutor to the difference in performance would be application thickness. These pastes are gap fullers and should be applied wafer thin, and to ensure aon high points on the mating surface are in contact metal on metal... then you'll get best overall thermal performance.

Thnner is better, scrub heatsink in place before locking down with bolts or clips. The force of these clips alone is not good enough to push out excess heatsink compound.

I'll see if i can get my mate to run some thermal sims if you guys can give me the following info:-

Chip power dissipation.
Thermal conductivity of compounds you want to evaluate, (Arrctic silver, TIM supplied with heatsink)
Areaof contact

dark4orz · 24 Sep 2006 at 09:20

do you recommend using Colgate instead of AS5 ?

silentphoenix · 24 Sep 2006 at 10:04

dark4orz said:
do you recommend using Colgate instead of AS5 ?

For your teeth yes... for your PCU no

Jokester · 24 Sep 2006 at 13:34

Spent this morning testing the various compounds I had at my disposable.

Got some cheapo free stuff (out of a Waterchill kit I think)
Thermaltake paste (came with TT BT I think)
Geil Copper paste (not sure where I got it from)
Arctic Silver 5
Arctic Silver Ceramique

First up was the toothpaste from yesterdays application (I've added in chipset temps to get an idea of any changes in ambient):-

Load 53
Chipset 32

Next up the cheapo stuff:-

Load 53-54
Chipset 31

Very much similar to the toothpaste

Next up was the Geil Copper paste, this was really easy to apply and getting a smooth even thin layer was simple:-

Load 50-51
Chipset 31

Just marginally better than the toothpaste.

Jokester

Jokester · 24 Sep 2006 at 13:39

Next up was the AS5:-

Load 46
Chipset 31

Ceramique:-

Load 47
Chipset 32

Very close the AS5 probably nothing in it when considering the chipset topped out 1C more than in the AS5 case.

Thermaltake:-

Load 48
Chipset 31

Pleasantly surprised with this one, had the same appearance as Ceramique and was equally easy to spread so it is perhaps just rebadged.

Jokester

FunkyCowie · 24 Sep 2006 at 13:50

My wife would like to see the colgate tested for a month

The Great Thermal Interface Material Con

Man of Honour