While testing TIM materials can be tricky, margin of error high and the tendency to draw to far-fetching conclusions from any single results equally fanciful, I though I'd post this round up.

I've only included fairly big roundups with fairly consistent - even if not faultless - test methodology:

HWReviewlabs 2-2009 Top 5 (all within 1.5C of each other, 230W peltier load)
1. Coollab Liquid Pro
2. IC Diamond
3. Noctua NT-H1
4. OCZ Freeze Extreme
5. Coolermaster Thermalfusion 400

Benchmarkreviews 6-2009 Top 5 (all within <1C of each other, Q9450 @ 3.2GHz load)
- Tuniq TX-3
- Gelid GC-Extreme
- Thermaltake Grease A2150
- Arctic Silver 5 Polysynthetic Thermal compound
- Shin-Etsu MicroSi G751

Lab 501 7-2009 Top 5 (all within 2C of each other, Q6600 @ 3.5GHz 1.4V load)
1. Nanoxia Nano TF-1000
2. Coollab Liquid Pro
3. Prolimatech PK-1
4. IC Diamond
5. Shin Etsu MicroSi

Now some personal caveats:
- not all of the same TIMs were tested in each round up, so a winner of one test may be completely missing in another test
- application to application error can be several degrees, unless controlled for
- ranking can change based on test round methodology (load time, load used, cooling used, etc)
- thermal conductivity alone is not be-all-end-all. Ease of application, corrosion, ease of removal, electrical conductivity, environmental issues, etc factor in. Please check specifics of all pastes, before buying blindly. This applies especially to Coollaboratory Liquid Pro or Nanoxia Nano TF-1000, which should never be used on aluminium surfaces.

Regardless, my personal conclusions are:
- the difference between the worst and the best paste is roughly 10C in almost every methodologically done round up
- The overall top 10 pastes almost always perform within the top 10 in each individual test and within 3-4C of each other.
- For a high performance silent enthusiast, who tries to gain every last inch of cooling performance without resorting to noise, selecting a paste within the top 5 may actually bring benefits
- Dan's Data may be funny, but I will not use hamster oil as my TIM :)

As an added bonus, take a look at TIM application notes from Lab 501.

A lot of new pastes are still coming out and have not been tested. This is just a snapshot in time. Still, I think even this field has progressed in the past 15 years I've watched it.

I hope you find this useful.

I wonder how Pepsodent (tooth paste ) does against their top 5?
But still. For regular user there isn't much to be gained in few degree's IMHO. Only real benefit for not using regular bulk paste is that retail paste's won't harm your precious coolers as some bulk pastes corrode copper during the time.

AS5 is still venerable TIM but I also like IC Diamond.

Performance T.I.M. gives psychological comfort to the user that parts aren't getting fried. 1-3 degrees Celsius temperature increase, unlike in a human being, is nothing to worry about, unless your life depends on pwn'g others with temperature charts. It is a fun game that many will eventually retire from and choose a reasonably good performing and priced T.I.M. for a few dollars cost. 1-3 degrees really is within margin for error and doesn't prove superiority of one material over another. However, 5 degrees and higher is enough to convince buying a better performing T.I.M., as well as to psychologically give one peace of mind.

Good comments.

However, I must add that one thing that people overlook often is as the weakest link in high output thermal dissipation.

If one has a very high thermal output (say i7 @ ~ 4GHz) with a well performing heatsink and a SPCR style fan, then the TIM can actually make a difference of 10C at loads.

Now, people with systems like this are far and few between, and mostly don't care about silence. But the remaining bunch are here, in this forum, because they care.

Those who care usually also push the limits of temperature by bringing down the airflow in order to limit noise. This can results in very high temp loads at near 90C.

In this situation, a 10C temperature difference can actually be clearly audible in corresponding fan output reduction.

And there is the added benefit of psychological comfort of owning 'one of the best' - which funnily enough seems to work, even if you don't believe in it :)

...we make the TIM less important. But yes, the difference can be audible.
One thing I have never read about. I think, not all TIMs are good for lapped surfaces. OCZ Freeze seems too thick to me.

The FrozenCPU 88 thermal compound roundup found AS5 to be on of the top five A+ TIMs, still after all this time.

I tried other TIMs like IC Diamond and Thermalright Chill Factor, but ended up going back to my trusty AS5.

Addition 1/2010:

Indigo Extreme vs MX-3, TX-3, PK-1, GC-Extreme & Liquid Pro
http://lab501.ro/racire/cele-mai-bune-p ... s-oxizi/17

More info about application @ Xtremesystems
http://www.xtremesystems.org/forums/sho ... p?t=232141

See the full review for diff. on temps with multiple mountings (fairly mount insensitive, so should get consistent thermal transfer)
http://vapor.skinneelabs.com/TIM/Indigo ... -main.html

following test (in portuguese) has tooth paste included.
looking at the graphic the last 4 are:
"batom rosa" = pink lipstick
"creme dental" = tooth paste
"creme dental (12h depois)" = toothpaste after 12 hours
"sem pasta térmica" = without any product

http://www.clubedohardware.com.br/artig ... 011/2319/5