Finally took the disappointing plunge into liquid metal

FireDemon

Limp Gawd
Joined
Feb 10, 2006
Messages
256
Went from MX-5 to Thermal Grizzly Conductonaut. While I had the block off, I figured I'd clean it up and go old-school a bit. Lapped it all the way up to 5000 grit. Applied a thin layer of liquid metal with the included gothic q-tips to both the IHS and coldplate. Put it all back together and the result?

Nothing. Literally not 1 single degree difference over MX-5 with the exact same ambient conditions between tests. Gaming it likes to sit in the mid 50s (Now sitting at 49C during gameplay and 50s while loading). But it's hard benchmarking I was interested in the results of.

Both TIMs with Prime95 small FFT test running for 30 minutes top out at and sustain a temperature of 91C with 5ghz PBO. 5800x in a custom loop.

I did notice that with the liquid metal, it does seem to buffer small/transient changes a little better but that's about it. Where are these guys you hear about claiming insane differences of 15-20C over top tier pastes getting their numbers from?
 

Attachments

  • IMG_20210524_020916.jpg
    IMG_20210524_020916.jpg
    382 KB · Views: 0
  • IMG_20210524_020913.jpg
    IMG_20210524_020913.jpg
    283.5 KB · Views: 0
  • IMG_20210523_231756.jpg
    IMG_20210523_231756.jpg
    274.4 KB · Views: 0
  • IMG_20210524_001329.jpg
    IMG_20210524_001329.jpg
    274.4 KB · Views: 0
Last edited:
The thermal paste has long been in the realm of 'law of diminishing returns' territory. When you are talking about 1-2 degrees variance the differing factors involved make it all moot.

As long as your CPU it running at acceptable temps while doing what you need it to do...you are done.

I've just stuck with MX-4 and done.
 
Went from MX-5 to Thermal Grizzly Conductonaut. While I had the block off, I figured I'd clean it up and go old-school a bit. Lapped it all the way up to 5000 grit. Applied a thin layer of liquid metal with the included gothic q-tips to both the IHS and coldplate. Put it all back together and the result?

Nothing. Literally not 1 single degree difference over MX-5 with the exact same ambient conditions between tests. Gaming it likes to sit in the mid 50s (Now sitting at 49C during gameplay and 50s while loading). But it's hard benchmarking I was interested in the results of.

Both TIMs with Prime95 small FFT test running for 30 minutes top out at and sustain a temperature of 91C with 5ghz PBO. 5800x in a custom loop.

I did notice that with the liquid metal, it does seem to buffer small/transient changes a little better but that's about it. Where are these guys you hear about claiming insane differences of 15-20C over top tier pastes getting their numbers from?
Ryzens are soldered to their IHS.

Most often, the large temp improvements are with CPUs which aren't soldered. My 7600k drop 12 - 18c. However, some soldered CPUs end up with a poor solder job and can also benefit.

You could potentially knock off another 1 -4C. with a fancier, modern paste. Thermal Grizzly Kryonaut Extreme, Thermalright TF8 or TF10, Coolermaster Mastergel Maker, etc.
 
You already lapped the CPU. You won't see a ton of difference as long as you are already using a high-end paste. Thermal Grizzly Kryonaut and MX-5 are two of the best right now. I've used them both.

Arctic Silver 5 and Arctic Ceramique II give about 5c+ higher temps.
 
You already lapped the CPU. You won't see a ton of difference as long as you are already using a high-end paste. Thermal Grizzly Kryonaut and MX-5 are two of the best right now. I've used them both.

Arctic Silver 5 and Arctic Ceramique II give about 5c+ higher temps.

Yeah Arctic pastes are in toothpaste realm they are so old. Hello 2003!
 
Liquid metal is only beneficial for use between a die and an ihs/block. Usage between the ihs and block/cooler isn't recommended.

Yup all about surface area. Once there's enough surface area (like the top of the heat spreader) there's not a lot better paste can do. As mentioned, when we were delidding Skylake, etc it was a different story. The processor die is so small in comparison that it makes a huge difference. I saw a drop of over 10C myself. Used regular paste between IHS and water block.
 
Yeah Arctic pastes are in toothpaste realm they are so old. Hello 2003!

I keep trying to tell this to people who are still stuck on using AS5, it's like they literally have not bothered to look up any new pastes in the last two decades. And then they act like I am recommending some exotic hyper-expensive paste when many modern good pastes cost the same as a tube of AS5.

That being said, it's not exactly a terrible paste still today, but there is really no good reason to use it IMO unless you have a non-expired tube left over and really don't want to spend a few extra bucks. I would say the bigger difference is not the better cooling, which while there is a difference with modern thermal pastes is not some significantly huge difference, but the fact that many modern thermal pastes are non-conductive while AS5 is.

Of course, it's not like the AS company has just been sitting on their behinds after they made AS5 either, there are modern AS pastes too. Normally I just stick with Noctua or Thermal Grizzly pastes.
 
I keep trying to tell this to people who are still stuck on using AS5, it's like they literally have not bothered to look up any new pastes in the last two decades. And then they act like I am recommending some exotic hyper-expensive paste when many modern good pastes cost the same as a tube of AS5.

That being said, it's not exactly a terrible paste still today, but there is really no good reason to use it IMO unless you have a non-expired tube left over and really don't want to spend a few extra bucks. I would say the bigger difference is not the better cooling, which while there is a difference with modern thermal pastes is not some significantly huge difference, but the fact that many modern thermal pastes are non-conductive while AS5 is.

Of course, it's not like the AS company has just been sitting on their behinds after they made AS5 either, there are modern AS pastes too. Normally I just stick with Noctua or Thermal Grizzly pastes.
I agree. I was at MC one time and this old dude was saying AS5 is the best paste to buy. I was like no. Gt the thermal grizzly or MX-4.
 
I've used it on several CPUs and GPUs. It's good stuff, but it's only going to do so much. I think the people experiencing huge drops didn't have a good mount in the first place. These are the results I've gotten, the best that I can remember:

Delidded 4690K: 15C drop on hottest core (delidded at the same time - did not try with regular thermal paste)
Delidded G3258: 12C drop on both cores (delidded at the same time - but did not try with regular thermal paste)
Zotac GTX 980TI reference card #1 with 5+ year old OEM paste: 4C drop from full load temp
Zotac GTX 980TI reference card #2 with 5+ year old OEM paste: 2C drop from full load temp (did this one twice to make sure - only got 1C the first time)
Between 5960X IHS and Phanteks TC14PE: 2C drop compared to a really good application of TG Kryonaut or Phanteks PH-NDC.
 
While your work is impressive, you are applying the LM wrong. Due to Ryzen‘s IHS being soldered to the die, the use of liquid metal was not really as much of an improvement over Intels “toothpaste“ they used between the die and the IHS. If you want to go down that route…



You can find out how he does it around the 9 minute mark, the results seem to be 2-4c.

You can try direct to die cooling, http://der8auer.com/ryzen-3000-oc-bracket/ , its not for the 5000 series though.
 
I've been considering doing LM on the CPU in my laptop, that sounds like a situation where it could actually help since it's direct-die. But yeah I'll echo everyone else who's said that LM is wasted on chips with an IHS.
 
Just make sure the laptop's heatsink isn't made out of aluminum. If it is, things will end badly... Eventually.
 
Just make sure the laptop's heatsink isn't made out of aluminum. If it is, things will end badly... Eventually.
An excellent point. Fortunately the only AL parts are the fins and part of the VRAM headspreaders- CPU contact plate and pipes are all copper.

I feel like I remember hearing of folks accidentally dissolving their waterblocks back when full aluminum blocks were still a common thing, yikes!
 
Just make sure the laptop's heatsink isn't made out of aluminum. If it is, things will end badly... Eventually.

I made the mistake of using it on a Coolermaster 212 once. It fell apart in under an hour and was just dangling from the motherboard. Apparently the only thing made of copper on that unit are the heat pipes themselves. WHOOPS.
 
I keep trying to tell this to people who are still stuck on using AS5, it's like they literally have not bothered to look up any new pastes in the last two decades. And then they act like I am recommending some exotic hyper-expensive paste when many modern good pastes cost the same as a tube of AS5.

That being said, it's not exactly a terrible paste still today, but there is really no good reason to use it IMO unless you have a non-expired tube left over and really don't want to spend a few extra bucks. I would say the bigger difference is not the better cooling, which while there is a difference with modern thermal pastes is not some significantly huge difference, but the fact that many modern thermal pastes are non-conductive while AS5 is.

Of course, it's not like the AS company has just been sitting on their behinds after they made AS5 either, there are modern AS pastes too. Normally I just stick with Noctua or Thermal Grizzly pastes.

I still have a tube of AS5 sitting around collecting dust. Odd thing is I stopped using it for the most part many years ago when I got some MX-4 (or was it 3? Whatever).

I found a good use for that stuff. Attaching little heatsinks to the memory on my video cards. They used to make this epoxy stuff called arctic alumina. Ceramic based, did a great job. Except that once it hardened nothing short of the jaws of life would get that little heatsink off again.

But ....if you were to mix up the epoxy and then mix in an equal part of AS5, it wouldn't harden the same. It would still adhere more than enough but it wouldn't be permanent. It became kind of rubbery and would peel off.

Unfortunately all the supply of alumina epoxy has dried up (ha ha! Pun!) So I have a useless tube of AS5.

And that's my pointless story.
 
Back
Top