Thermal Paste: Which do you use?

Tried eBay? You can get quite a few of the top ones with free postage (Coollaboratory, Gelid Extreme, IC Diamond, MX-4 etc). To be honest as long as you're not buying cheap no branded shit you can get better temp gains elsewhere. 1-4 degrees in real world is nothing really.

 

I have CLU under the Lid of my CPU.

Coolermaster Maker Nano is my current paste. Does good, comparable to Gelid, NT-H1 and Kryonaught IMO. Will do a roundup one day.

 

So.... what conclusion did we come to... almost anything 'name brand' will do the job ? There's so little difference that for 99.9% of users, just grab what's there?

Perhaps it would be easier to say if there were any particular "must avoid" brands?

 

Pretty much just buy whatever is readily available and suits your application (as long as it's not some no name brand off ebay).

For myself I'll use pretty much anything brand name on CPU's as they are essentially all so close together performance wise. However I won't use pastes such as AS5 on GPU's as its harder to spread and control on the larger surface. Don't want to chance it spilling out onto other components close to the die with it's somewhat capacitive properties.

 

Cheers for that.

 

I've been using Coollaboratory liquid ultra for quite some time and have always had good results. i recently delidded my 4770k and replaced the tim with this stuff.

T1

 

New thermal material provides 72% better cooling than conventional paste
https://www.techspot.com/news/105537-new-thermal-material-provides-72-better-cooling-than.html

Thanks to a mechanochemically engineered combination of the liquid metal alloy Galinstan and ceramic aluminum nitride, this thermal interface material, or TIM, outperformed the best commercial liquid metal cooling products by a staggering 56-72% in lab tests. It allowed dissipation of up to 2,760 watts of heat from just a 16 square centimeter area.

The material pulls this off by bridging the gap between the theoretical heat transfer limits of these materials and what's achieved in real products. Through mechanochemistry, the liquid metal and ceramic ingredients are mixed in an extremely controlled way, creating gradient interfaces that heat can flow across much more easily.

Beyond just being better at cooling, the researchers claim that the higher performance reduces the energy needed to run cooling pumps and fans by up to 65%. It also unlocks the ability to cram more heat-generating processors into the same space without overheating issues.

 

Bit of an odd one that article, colour me skeptical.

Thought it was just going to be Honeywell PTM7950 equivalent but seems something else.
2,760W from 16 cm² would be impressive, that would effectively be 100% heat transfer between IHS and the cooling solution.

Not sure about the other claims.
72% better than conventional paste.... but then goes on to say 56-72% better than commercial liquid metal. Huh??
65% usage reduction of cooling pumps and fans is non-sensical. The heat still needs to go somewhere when it's transferred to the cooler.

More efficient transfer from the chip(s) may mean processors need slightly less voltage for a given clock as they aren't battling as much thermal resistance (see the Ryzen 9800X3D cache redesign for an example), which may lead to slightly less overall heat generation in the first place - but it has diminishing returns.

Either way, the transferred heat has to go somewhere - and that generally requires fans and pumps. TIMs don't dissipate heat (solving the problem) as this article states, they simple transfer heat (passing the problem on).

 

You have to me a bit careful with these sorts of claims, often times the temperature Delta is much higher than an enthusiast would expect.

Sure you can cool 2.7kw, as long as the 16cm2 is 2000c and cooling with 2c air with a 20k fan, or whatever.

To be more serious , if its 70% better than liquid metal, then liquid metal is about 58% as good ( 100 /170) , so 2760*0.58 = 1600W of cooling for liquid metal. That's a lot more than any of us would reasonably expect too cool even with liquid metal.

 

Gonna bump this thread up as I decided to give Phasesheet PTM (From Thermal Grizzly) a go on my 14900k (note, it's not going full bore, it has a 5.5mhz P Core and 4.1mhz E Core lock with a power limit 250 long, 230 short, using a negative 0.055 voltage offset, creating a max voltage of 1.34v). Got it from Scorptec for $17 with the OCAU member sponsor code. Came yesterday and decided to whack it in today. I did get some backup Noctua NT-H2 as well but it wasn't needed. I had mastergel marker before, and I had used a lot of it, it was all over the place . No drying out at all from what I could tell, I was surprised it looked as good as it did after a year.

It comes as a pressed sheet between a couple pieces of plastic, it's rebadged Honeywell PTM7950. It looks like regular thermal paste but in the sheet form, and is reportedly about as good as any paste on the market but not quite as good as liquid metal.

The install was a bit complicated, it didn't seem to stick down properly to the CPU (I didn't need to cut it at all because it's non-conductive and I have a contact frame) for the plastic top to lift it away. I had to eventually gently lift one corner (opposite side to where it wants you to lift) and had a little bit of tearing in the corner. Once I got the lift off in the corner it came off properly with no more rips. On further research supposedly putting it in the fridge or freezer helps but I didn't know about this and so didn't do it.

It has to do "thermal cycling" to achieve full results apparently, after a couple C23 runs that throttled a few cores, doing it again and it's already dropped from hitting ~97 on some of them to a max of 93, in ambient of 28c & even after a 10 minute run only one of them got to 97, the rest from 94 to as low as 87. E cores hit 81 max. I'm not sure how much it will improve until it's fully cured or cycled or whatever the terminology for a phase change pad is, but already the temps are low enough to call a success.

So would I recommend it? Yes. It's clearly a very good interface even with a very hot chip. It has a downside though with the potentially tricky install. Best case scenario would be when you first do a new build/install a new CPU that you aren't going to be touching for a while. You also need to have good mounting pressure, it's very flat compared to using paste.

For low & mid tier CPUs a Kryosheet or Carbon IC Pad/Panasonic PGS might be a better option as it has an easier install (as long as you're careful) and you don't need as high performance and isn't a material that could ever pump out. If you think the cooler will have to be replaced or taken off soon then cheaper thermal paste tube where you get more than one application would be the move.

The thermalsheet is as cheep as any high tier paste tube. This makes it a very attractive option for any build or situation where you aren't going to have to redo it and you are confident of your ability to install it.

 

It's damn good gear - genuine PTM7950 is great, but even the cheap knock-off stuff is proving very acceptable on a couple of other rigs in the house - laptops, especially.
I padded my old ASUS and lost 15C (CPU) over the (still reasonably gooey) Gelid paste I had on there.

If you want to go down a good rabbit hole, Igor has put together some fantastic work on his new TIM database. Worth a look a look before you buy any more. Let me know if you find DOWSIL in anything less than a kilo

 

Doing research on this I found a lot of people selling the easier to supply PCM500 or PCM8500 but labelled as PTM7950. There's also Thermalright Helios which is inspired by PTM7950 and has similar element percentages but uses coarser grains of it.

 

Yep you're right mate - I think there's a 0.05mm thickness discrepancy too, with genuine PTM7950 being 0.2mm and the cheaper ones being 0.25mm (or maybe vice versa). My calipers aren't that good, so I wasn't able to verify that

 

Doing a little more testing, I'm not entirely sure what counts as a "thermal cycle" for the supposed requirement of 10 to have it fully ready, but I'm probably halfway through that at least.

Back in August when there was a microcode/bios update for my 14900k, I made a post here about the scores I got. tl;dr is

• 1.314v voltage max
• P Core max temp 84
• P Core low temp 76c
• E Core max temp 75c
• Power usage 217w
• C23 Score 38,811

Ambient temp I didn't say, except that it was August 10, 2am and "freezing", which seems correct with a quick google of the temps that day, probably 8c or less where I was at that time.

Same settings in the bios that I had back then (5.5ghz P, 4.1ghz E core locked with a 0.055 negative voltage offset being the main ones). 28c ambient according to a desk thermometer.

• 1.324v voltage max
• P Core max temp 87c (on 3 cores)
• P Core low temp 81c (on 2 cores)
• E-Core max temp 75
• Power usage 220w
• C23 Score 38,883

Considering the difference in apparent ambient temp this is a crazy good result. I'll do more testing but if I don't post anything further consider it job done and a good result with no issues.

edit - Did another test just now. 24c ambient. HWInfo reports nothing over 84c on the P cores, low was 73c (!), some were in the high 70's and had a e-core low of 65c. I was worried about mounting pressure but damn, this stuff is magic.

I bumped my P-Cores to 5.6ghz, got a C23 of 38,950 and the highest my hottest core got was 87c. Still not using more than 220w, the voltage is still only 1.35v and the e-cores are still around 70c.

 

Thermal Grizzly Kryonaut extreme. Been using it for yrs already & pleased with its results.

 

macktheknife juzz86

I’m using:
Gelid Solutions HeatPhase Ultra Phase Change Material (PCM) Thermal Pad for AMD & Intel CPUs, with my 9800X3D, got it free with an order from OCGear.

https://www.ocgear.com.au/products/...se-change-material-pcm?variant=47223967842612

Seems 7950-ish and very good, but I’ve no prior paste with this CPU to compare to