Difference between thermal pastes

[ChrisZH]

Hello all, just got another question I want to float around to people. So when I installed my waterblock I did ensure that my thermal paste covered the entire chip with a thin layer. Is that all you need, just a very thin layer or should I have a little excess?

 

I am also using the stock thermal paste that came with the block (about 2 years old). The paste still had a very good consistency to it and was not chunky at all but I am also left to wonder if they have a shelf life.

 

I have also ordered some thermal grizzly hydronaut paste because I want to repaste my 5800x3d with some quality paste but also think it might be a good idea to repaste my 2080ti because it is also using the stock paste. Is it possible that there will be a difference between the stock thermal paste that came with my Bykski blocks and thermal grizzly paste? If so do you think there could be a measurable difference between the two?

[ToboRobot]

You want to use as little as possible.  

Difference between pastes is very hard to measure, as variables like mounting pressure etc can impact it.  Personally I don't bother repasting unless it's failing/causing a problem, but it's not worth it to try to get a few extra degrees unless you are doing serious overclocking.

[ChrisZH]

1 minute ago, ToboRobot said:

You want to use as little as possible.  

Difference between pastes is very hard to measure, as variables like mounting pressure etc can impact it.  Personally I don't bother repasting unless it's failing/causing a problem, but it's not worth it to try to get a few extra degrees unless you are doing serious overclocking.

Well one thing I am worried about that I remember is one of my washers broke so I replaced it but the thickness was different. I was thinking of getting new washers for every screw for the block so the pressure is the same across the board? This might cause an issue right? 
 

with using PBO2 I never see above 81 c in stress tests anymore and my average is about 68-70 now. So it's not anywhere close to throttling anymore but I just want the most out as possible I guess?

 

So for the thermal paste you really just want as thin of a layer as possible? Is it because the mounting should push the paste to fill in the micro holes of the block and IHS?

[ToboRobot]

2 minutes ago, ChrisZH said:

 

So for the thermal paste you really just want as thin of a layer as possible? Is it because the mounting should push the paste to fill in the micro holes of the block and IHS?

Yeah.

[ChrisZH]

2 minutes ago, ToboRobot said:

Yeah.

Hmm, is there a way to see if mounting pressure is off without taking the block off? Like should I look at specific core temps and see if there is an outlier?

[TeraSeraph]

11 minutes ago, ToboRobot said:

You want to use as little as possible.  

This is how people end up with bad paste jobs. There's no such thing as too much thermal paste, given it's non-conductive.
 

 

[Kaetemi]

17 minutes ago, ChrisZH said:

I am also left to wonder if they have a shelf life.

 

Initial performance should last at least a year, then it'll slowly degrade. Normally you'll be fine for 4 years on average, and you can stretch it as long as there's no performance troubles.

[ToboRobot]

4 minutes ago, TeraSeraph said:

This is how people end up with bad paste jobs. There's no such thing as too much thermal paste, given it's non-conductive.
 

 

Love GN and the team, but I don't think at the time they were able to record the data needed to make these claims.  This is a fun video, it is not empirical science.  

 

[ToboRobot]

11 minutes ago, ChrisZH said:

Hmm, is there a way to see if mounting pressure is off without taking the block off? Like should I look at specific core temps and see if there is an outlier?

Nope.

You can control the environment (temp, etc) and remove and mount the cooler & preform a number of controlled tests to get a baseline average and numbers for a good mounting and then try to replicate it.  

Or just be like, "it's good enough until it's a problem then I'll fix it"...

[freeagent]

If its too thick it acts as an insulator. 

 

That video is why so many people have shitty temps lol.

[TeraSeraph]

17 minutes ago, freeagent said:

If its too thick it acts as an insulator. 

 

That video is why so many people have shitty temps lol.

Unless they're using dried up paste or not mounting their cooler properly it should just squeeze out the extra. 

[ChrisZH]

3 minutes ago, TeraSeraph said:

Unless they're using dried up paste or not mounting their cooler properly it should just squeeze out the extra. 

I guess this was a problem for me because I don't want excess to get on the board because the cleanup I feel is a nightmare

[freeagent]

Just now, TeraSeraph said:

Unless they're using dried up paste or not mounting their cooler properly it should just squeeze out the extra. 

You would think.

 

There is such a thing as too much. Not all TIMs are created equal.

[ChrisZH]

23 minutes ago, freeagent said:

If its too thick it acts as an insulator. 

 

That video is why so many people have shitty temps lol.

If it acts like an insulator when it's too thick then how do you really know when it's a good amount on the chip? When I put mine on I got a nice thin layer but at the same time I was worried that it wasn't enough but I didn't want to just add more because I figured it might inhibit heat transfer

[ChrisZH]

1 minute ago, freeagent said:

You would think.

 

There is such a thing as too much. Not all TIMs are created equal.

This is what I thought but then unfortunately there is just so much conflicting points of views out there. So it's hard to figure out which way is the best way.

[Somerandomtechyboi]

48 minutes ago, ToboRobot said:

You want to use as little as possible.  

Difference between pastes is very hard to measure, as variables like mounting pressure etc can impact it.  Personally I don't bother repasting unless it's failing/causing a problem, but it's not worth it to try to get a few extra degrees unless you are doing serious overclocking.

Decent advice for the cpu, you really only need a decent size dot or an x, cooler will spread it out for you

 

But for the gpu you really wanna go ham, you do not want even one area of the die not covered as that will lead to horrific hotspot temps problably even premature failiure depending on how hot it gets (>115c)

 

 

I went ham on both and so far temps are pretty good for a dinky 92mm tower cooling an x5650 at 4.2ghz, gpu temps are shit but the cooler design itself is horrible so makes sense

[freeagent]

Just now, ChrisZH said:

If it acts like an insulator when it's too thick then how do you really know when it's a good amount on the chip?

Trial and error. I have spent time and money trying for myself. 

[TeraSeraph]

4 minutes ago, ChrisZH said:

If it acts like an insulator when it's too thick then how do you really know when it's a good amount on the chip?

You should see a bit of tiny bit squeeze out around the edges but not a lot, if there isn't squeeze out then it's likely there's not enough to ensure full coverage.  

[ToboRobot]

3 minutes ago, Somerandomtechyboi said:

Decent advice for the cpu, you really only need a decent size dot or an x, cooler will spread it out for you

 

But for the gpu you really wanna go ham, you do not want even one area of the die not covered as that will lead to horrific hotspot temps probably even premature failure depending on how hot it gets (>115c)

 

 

I went ham on both and so far temps are pretty good for a dinky 92mm tower cooling an x5650 at 4.2ghz, gpu temps are shit but the cooler design itself is horrible so makes sense

Yeah, even on a GPU you want to use as little as possible (as still cover the whole GPU).  

I wrote some articles on paste and cleaning it in the early 2000s for overclockers.com but they appear gone so I can't link to it and I'm lazy... 

[SorryBella]

9 minutes ago, ChrisZH said:

If it acts like an insulator when it's too thick then how do you really know when it's a good amount on the chip? When I put mine on I got a nice thin layer but at the same time I was worried that it wasn't enough but I didn't want to just add more because I figured it might inhibit heat transfer

Rule of thumb, if it squeeze unreasonable amount to the outside of the cooler and IHS contact, you probably applied too much. But really your method work.

[freeagent]

Just now, TeraSeraph said:

You should see a bit of tiny bit squeeze out around the edges but not a lot, if there isn't squeeze out then it's likely there's not enough to ensure full coverage.  

Unless you are a pro

[Somerandomtechyboi]

1 minute ago, ToboRobot said:

Yeah, even on a GPU you want to use as little as possible (as still cover the whole GPU).  

I wrote some articles on paste and cleaning it in the early 2000s for overclockers.com but they appear gone so I can't link to it and I'm lazy... 

Think ill have to do some of my own testing but im selling my 580 so i guess whenever i buy a gpu that i wanna keep for myself ill do some testing for myself

 

Theres also a thing called pump out that can happen at high temps iirc, basically makes the corners of the gpu have no thermal paste after enough high temp cycles so im also inclined to test that

 

Just need a gpu that has a proper temp range, minimum 100c tjmax but id prefer something that can go up to 115c for optimum testing

[freeagent]

Just now, Somerandomtechyboi said:

Theres also a thing called pump out that can happen at high temps iirc

Some very popular TIMs pump out, and it is known.. yet people still buy. Crazy.

 

I wont name names, but they are easy to find.

[Somerandomtechyboi]

1 minute ago, freeagent said:

Some very popular TIMs pump out, and it is known.. yet people still buy. Crazy.

 

I wont name names, but they are easy to find.

Oh so its not really an issue on all pastes?

 

maybe i overdid my gd007 application on that 580 but oh well less risk of uncovered areas since i went absolutely nuts

 

I guess this is something interesting i can test out, but what cards aside from the rx 6000 cards can run at 115c? i dont mind bios modding a random junk card to run at 115c but id prefer something that isnt gonna die when run up to 115c

[freeagent]

13 minutes ago, Somerandomtechyboi said:

Oh so its not really an issue on all pastes?

Nope not at all. I am using SYY-157 on my CPUs right now. I think I even have some on my GTX 580 Matrix Platinum. It is on my 980 Classified for sure.

 

Other TIMs that I have tried recently that are good are TF7, TF8, TFX. Even AS5 and Ceramique 2 are ok as long as you aren’t pushing limits. Those two would be better on lower wattage parts though.. but they are still ok.