thermal paste vs. thermal adhesive: performance difference

[Stefan1024]

I know you are not supposed to use thermal adhesive between the CPU and the heat sink. However I like to do something similar because there is no other way.

 

Therefor I'm interested in the performance difference between thermal paste and thermal adhesive (Arctic Silver Premium). The adhesive has about 7,5 W /(m*K), paste can range from 5 to about 11 W(m*K) so they should perform more or less the same.

Has anybody by accident glued the CPU to the heat sink and has some experience with the performance?

[WoodenMarker]

I haven't had personal experience comparing the two but there shouldn't be much of a difference. The important part is that the paste and/or adhesive fills the gaps that would otherwise be air which is much worse at transferring heat. 

[Stefan1024]

8 hours ago, WoodenMarker said:

I haven't had personal experience comparing the two but there shouldn't be much of a difference. The important part is that the paste and/or adhesive fills the gaps that would otherwise be air which is much worse at transferring heat. 

That's waht I thougth, but I have no experience / proof.

Also the glue will displace mineral oil and not air.

[Cyracus]

yeah, you can use toothpaste for thermal compound and it'll work pretty close to thermal compound for a short while, so as long as you're using a product designed for this kind of thing it'll work just fine

[VJK]

The idea is to minimised heat conduction resistance. So the paste and pad is to fill in the gab between cooler and cpu plate. So less gab is better.

And comparing pad and paste.

Is small gab paste better. For bug gap, pad can fill in more firmly. So depends on size of gap.

[Stefan1024]

14 hours ago, Cyracus said:

yeah, you can use toothpaste for thermal compound and it'll work pretty close to thermal compound for a short while, so as long as you're using a product designed for this kind of thing it'll work just fine

With an overclocked Titan X and passive cooling, I can not afford to waste some performance to a bad thermal compound. I'm already on the edge

10 hours ago, VJK said:

The idea is to minimised heat conduction resistance. So the paste and pad is to fill in the gab between cooler and cpu plate. So less gab is better.

And comparing pad and paste.

Is small gab paste better. For bug gap, pad can fill in more firmly. So depends on size of gap.

I have only a smal gap, so I don't need a pad. But I'm wondering between the performance of thermal past and glue.

[Cyracus]

tested differences in paste/pads is a few degrees, that's the low end to top end so if you get good pads you're looking at a degree or two from good thermal paste

[Stefan1024]

2 minutes ago, Cyracus said:

tested differences in paste/pads is a few degrees, that's the low end to top end so if you get good pads you're looking at a degree or two from good thermal paste

The temperature difference is dependet on the thermal conductivity of the paste / pad, but also on the thickness, area of the heat transfer and the power that is going througth the connection. So My results can vary considerably from an other test with other parameters.

 

But I got your intention, with the high quality termal adhesive I should be fine.

[Lays]

If you're worried about temps on the GPU whilst being passive, just use CLU.  It will have the best delta's, I just put it on a 780 Lightning to test last night, and saw a 3c drop on watercooling under load. My room was 18c and I was seeing load temps hover at 23-24c in Valley at 1.287v 1332 mhz.

Just be careful when you put it on, it only takes like a minute or so, very easy if you just take your time.

 

I think the issue with a pad may be it's too thick, and some of the heat transfer could be lost? I'm not sure if the pad would melt or anything under conditions on top of a gpu, I've never seen anybody do it before so IDK.

[WoodenMarker]

10 minutes ago, Lays said:

If you're worried about temps on the GPU whilst being passive, just use CLU.  

Liquid metal in a mineral oil pc is a bad idea. It might leak out and get to other components. 

[Lays]

39 minutes ago, WoodenMarker said:

Liquid metal in a mineral oil pc is a bad idea. It might leak out and get to other components. 

It's fairly solid when  you put it on, I think it'd be OK.  Not 100% sure though, what did Linus use in his build log? Normal paste? 

[thekeemo]

Had a project that I used it for.

Very similar

[Stefan1024]

7 hours ago, Lays said:

If you're worried about temps on the GPU whilst being passive, just use CLU.  It will have the best delta's, I just put it on a 780 Lightning to test last night, and saw a 3c drop on watercooling under load. My room was 18c and I was seeing load temps hover at 23-24c in Valley at 1.287v 1332 mhz.

Just be careful when you put it on, it only takes like a minute or so, very easy if you just take your time.

 

I think the issue with a pad may be it's too thick, and some of the heat transfer could be lost? I'm not sure if the pad would melt or anything under conditions on top of a gpu, I've never seen anybody do it before so IDK.

Liquid metal is very agressive towards aluminium. As my heat sink is made out of aluminium, it's going to be funny.

[Lays]

3 hours ago, Stefan1024 said:

Liquid metal is very agressive towards aluminium. As my heat sink is made out of aluminium, it's going to be funny.

O I didn't see that part