Rate my liquid metal job...

[blkout]

@cvaldes
I mean paying for LM is pointless. Better be invested on cpu block
Or perhaps good quality thermalpads, of course after thickness is firmly decided

I don’t know why people are bashing you, you’re right.

I’m still trying to figure out why anyone would put LM on top of thermal pads.. the whole point of liquid metal is metal on metal contact. That’s the primary reason it’s used in CPU deliding. The fact that he’s getting worse temps and performance pretty much sums this up. Every cpu I delided and put LM on dropped 10-20c.
Also, with this application, again he’s correct in saying you’d be better off with a water block. Obviously what you’re doing here isn’t working and the amount of LM there is just going to heat up and splash all over like hot grease on water.

 

[Mussels]

i wonder if im doing it correctly... so you decide...
tempretures.. wel they are worse then i started with... thermal trotels immediately after firing up furmark...
tough idle is 31c...

I've used LM plenty of times on GPU's and CPUs


1. That's too thick. It's going to run. You need a super super thin layer, you shouldn't see lumps or enough liquid to move around after it's on both sides.
Think putting two sheets of glass together, a single drop would cover a huge space.


Your green thermal pads look insanely thick, i'm with others thinking you've got too thick pads preventing contact at all.

I don’t know why people are bashing you, you’re right.

I’m still trying to figure out why anyone would put LM on top of thermal pads.. the whole point of liquid metal is metal on metal contact. That’s the primary reason it’s used in CPU deliding. The fact that he’s getting worse temps and performance pretty much sums this up. Every cpu I delided and put LM on dropped 10-20c.
Also, with this application, again he’s correct in saying you’d be better off with a water block. Obviously what you’re doing here isn’t working and the amount of LM there is just going to heat up and splash all over like hot grease on water.

LM... on thermal pads?
I missed something horrible, didn't I?

LM is only good for naked dies, CPU or GPU - and somewhat useful on lapped IHS.

Whatever, they're too thick.

Because it's air cooled now
Because a 210w gpu doesn't need liquid metal
Because watercooling and liquid metal...

@cvaldes
I mean paying for LM is pointless. Better be invested on cpu block
Or perhaps good quality thermalpads, of course after thickness is firmly decided

LM isn't focused on wattage, your views are off here.

LM is important when you've got high heat density and you need to transfer it fast, out of a very small area.
Naked dies to high quality heatsinks (Be they water or air) is where it shines.

The only time it's no good on air, is if the plates aren't flat like on basic coolers like the famous budget hyper 212
All these are 212 variants over the years and explain it well.

212 classic: LM happy (just protect the alu, or be super careful)

212 plus: the alu parts get thanos'd, and you get LM leaking out along the pipes
Even without the Aluminum, the varying height makes it a bad choice. You need a lot more LM to fill the space, where LM is best in very thin, very flat contact situations.

212 evo:
if you covered the side alu parts with the nail polish, you'd be okay - but you gotta also worry about leaks along the pipes, although its a lot smaller risk than the prior version.

212

 

[Morgoth]

Obviously what you’re doing here isn’t working and the amount of LM there is just going to heat up and splash all over like hot grease on water.

not sure what ur on but i think its my dilithium crystals

any way il be replacing thermal pads with thermal paste

those green pads i have used thinners before had issues with those.. thats why i tried 3mm ones to see if it makes a difference

thats not a vega 56 so it doest COUNT
and stop telling me to put a liquid cooler on my cpu....

 

[Lei]

any way il be replacing thermal pads with thermal paste

doe dat niet. luister alstublieft

 

[Morgoth]

doe dat niet. luister alstublieft

wat wil je dan? een zegt dit andere zegt dat? wat is het nu? wat wil je ?

 

[xtreemchaos]

i dont know what people want, me id pick though and take what i need, i wish you the best of luck chap.

 

[1100R]

wat wil je dan? een zegt dit andere zegt dat? wat is het nu? wat wil je ?

I found this:

Vega hotspot problem

So I mounted a water block on my reference vega 64 because I was tired of hearing a blower card all the time. I know this is a problem for a lot of vega owners, especially those who water cool. When I look I my "GPU temperature" they have improved greatly, but the hotspot temperature has gotten m...

linustechtips.com

It may be useful to you

 

[Morgoth]

YES its FIXED!!!
hotspot new max temp 44c after finishing time spy!

final result

 

[Zach_01]

So what you did differently and worked? Other than remove some pads and replaced them with paste… or was that it?

 

[Morgoth]

So what you did differently and worked? Other than remove some pads and replaced them with paste… or was that it?

yes that was it

 

[Mussels]

YES its FIXED!!!
hotspot new max temp 44c after finishing time spy!
View attachment 258888
View attachment 258889

final result

Good work with the detailed photos of the LM and pads
Either that thing needed very thin thermal pads, or it had stock paste initially?
Almost always pads will be better if that's what stock had. Thermal compounds inevitably sag due to gravity, so if it's not a tight fit they'll get worse over time, and sooner or later you'll have temps spiking on whatever components it was used on.

That still looks like too much LM to me and i'd be worried about leakage.
Some can get absorbed into the heatsink, but after that initial absorbtion and heatsink discoloration, it stops doing so and the excess could possibly leak.
Yes, I am aware you protected around the die - but if you ever take it off, reduce the excess.

This is my waterblock *after sanding it back* - notice the amount absorbed in?

I'm having trouble finding photos of when i applied mine, this is one near completion when it also had some lumps and uneven spreading

I used a lamp to put some radiant heat to soften it up, which helped.

This is what that application looked like months later when the waterblock was removed (~2 weeks ago)
Note the blob of excess - that was working its way down and suction from removing the waterblock pulled it back up like that (was similar on the block, as well)
. Given another year, or maybe just transporting the PC somewhere that might have dripped down.

 

[Morgoth]

i hope with the gravity pulling towards the coldplate would't be mutch of a problem. btw what did you use to sand it? looks so course?

not sure if it came with thermal past or not i had it on it for a year

i did stir it around with a fresh cotton swab and hoped it would absord into it. cus i could't absord it back into the srynge

btw i wonder how well does gold coating protects against corosion from liquid metal??

 

[Zach_01]

btw i wonder how well does gold coating protects against corosion from liquid metal??

This is what I know about LM (thermal grizzly)

You dont need gold plated surface for LM.

LM harms only Aluminium. They form an alloy together and after it goes downwards quickly. Alu then goes brittle and its completely destroyed. Crumbles like an old bread.
Nickel plated has nothing to worry about, a little if its sanded and depends what will (if) surface under it.

Copper has 1 thing with LM.
LM is mostly composed from Gallium and Indium (and some Tin).

In contact with copper, Gallium is infiltrating it (they have opposite charged ions +/-) and copper gets stained by it. Copper doesn't affected by it in the first place. Its integrity or its lapped surface...
but...
What LM is left behind over time is poor in Gallium and it can go dry and turn into a solid.
That can potentially hurt surfaces of IHS and Copper coldplate and in most cases (if dried out) surfaces (both) need a lot more than just sanding.
Especially if you used the minimum quantity required, like a very thin layer. The migration of Gallium inside Copper is continuing until later is saturated by the former, so this process is also waning overtime.
Thats why on applications of LM on Copper is suggested to replace it frequently the first 6-8 months. Replace it 3-4 times until you can leave it long term.

It works as I did it on my H110i AIO Copper coldplate and after 2.5 years its just fine. Maybe even better than what was before as its contact surface is Gallium stained now. The worst case is that its left the same.

 

[Shrek]

I’m still trying to figure out why anyone would put LM on top of thermal pads..

Who does that?

 

[cvaldes]

Who does that?

Probably someone who says "I read it on the Internet" or "I saw it on YouTube"...

Back to the original topic, I'm pleased to see that indeed by changing his approach, OP was able to resolve his GPU's thermal issue.

 

[xtreemchaos]

me ! only joking i find it lifts the moral . at least he got it sorted and thats what counts.

 

[Morgoth]

You're absolutely right.
I was trying to keep his hands off LM for the time being because he used an entire tube in one go.

^
i did not

you also told me not to use any kind of thermal compound... who does that??

than another guys tells me not to use the orginale back brackets....

btw the manual said to place thermal pads in the grooved parts to you know thatright?

and you showed me a video of a grapics card that is not a vega 56.....

you even told me and ur quite absolute into that i never did water cooling??

you also claimed i had no water in..

 

[Mussels]

i hope with the gravity pulling towards the coldplate would't be mutch of a problem. btw what did you use to sand it? looks so course?

not sure if it came with thermal past or not i had it on it for a year

i did stir it around with a fresh cotton swab and hoped it would absord into it. cus i could't absord it back into the srynge

btw i wonder how well does gold coating protects against corosion from liquid metal??

It was coarse - As i said, those images were taken halfway and i forgot to take any when finished. At the time, i was demonstrating to someone how LM can be removed, but it leaves stains.


You're mixing and matching bits of info and confusing them.

GPU and CPU dies need thermal paste, because the heatsink has close contact with high pressure - they directly touch.
Anything with an air gap requires thermal pads - the RIGHT thickness thermal pads.

 

[delshay]

YES its FIXED!!!
hotspot new max temp 44c after finishing time spy!
View attachment 258888
View attachment 258889

final result

Wow,wow, wow.. Thank god you found the issue. Do a superposition benchmark 1080p Extreme download here Superposition benchmark | UNIGINE Benchmarks.

Then run Unigine Heaven for about 20 -30 mins. Download from the same website.

Good work with the detailed photos of the LM and pads
Either that thing needed very thin thermal pads, or it had stock paste initially?
Almost always pads will be better if that's what stock had. Thermal compounds inevitably sag due to gravity, so if it's not a tight fit they'll get worse over time, and sooner or later you'll have temps spiking on whatever components it was used on.

That still looks like too much LM to me and i'd be worried about leakage.
Some can get absorbed into the heatsink, but after that initial absorbtion and heatsink discoloration, it stops doing so and the excess could possibly leak.
Yes, I am aware you protected around the die - but if you ever take it off, reduce the excess.

This is my waterblock *after sanding it back* - notice the amount absorbed in?
View attachment 258896


I'm having trouble finding photos of when i applied mine, this is one near completion when it also had some lumps and uneven spreading
View attachment 258897
I used a lamp to put some radiant heat to soften it up, which helped.

This is what that application looked like months later when the waterblock was removed (~2 weeks ago)
Note the blob of excess - that was working its way down and suction from removing the waterblock pulled it back up like that (was similar on the block, as well)
. Given another year, or maybe just transporting the PC somewhere that might have dripped down.
View attachment 258898

Yes, yes, yes. I use my very high power bike light when applying LM. It helps with the bonding effect, ie more surface tension. This will bring down the temperature even further & it's more stable.

 

[Morgoth]

Wow,wow, wow.. Thank god you found the issue. Do a superposition benchmark 1080p Extreme download here Superposition benchmark | UNIGINE Benchmarks.

Then run Unigine Heaven for about 20 -30 mins. Download from the same website.



Yes, yes, yes. I use my very high power bike light when applying LM. It helps with the bonding effect, ie more surface tension. This will bring down the temperature even further & it's more stable.

going to run it

gpu tens to stay at 1.400 mhz might try to do bolt 100 mhz overclock

funny amd auto overclock doest do a thing.. setted to 1670 mhz stays at 1400mhz...

turbo gives me a score of 4200

 

[delshay]

going to run it

View attachment 259035
gpu tens to stay at 1.400 mhz might try to do bolt 100 mhz overclock

You are beating my card because your temperature is so low. You can go much faster than this if you are into overclocking. If you are not into overclocking leave it as it is. However I recommend undervolting the card. Your superposition benchmark score should not change.

 

[Morgoth]

and all that with a single 120mm 68mm thick radiator and i also use liquid grapphite as coolant

 

[Colddecked]

It looks like you are ready for a monster overclock.

 

[Morgoth]

might also be a good sign that i wil be enabelded to run more power hungry grapics cards

 

[ThrashZone]

Hi,
How are you oc'ing thew gpu now ?
Msi afterburner ?