CPU throttling 2 weeks after re-applying thermal paste.

[Raz3R8023]

Hello,

I've got a 7700k I like to run at 4.9 all core under load, I got a liquid cooler a few months ago after running a hyper 212 since 2017, after re pasting, the cpu will run at ~65 degrees full load, no worries but a re-occurring thing is after ~2 weeks it suddenly decides it wants to run at ~95 degrees, every two weeks. I thought i was overtightening the block thus forcing the TIM to flow out when hot because that's what i can see when i take the block off, but nope, even when i tighten by hand it only wants to behave for 2 weeks. I am not a wealthy man and decent thermal paste is expensive so i can't afford to re paste every other week. 

So I ask, fellow tech enthusiasts, what do i do because re-pasting every 2 weeks is expensive and mentally exhausting.

Thanks. 

[Somerandomtechyboi]

What is the thermal paste that you are using?

 

Cause some thermal pastes degrade faster than others, i personally reccomend arctic mx4 or kryonaut in terms of thermal paste

[LIGISTX]

29 minutes ago, Raz3R8023 said:

Hello,

I've got a 7700k I like to run at 4.9 all core under load, I got a liquid cooler a few months ago after running a hyper 212 since 2017, after re pasting, the cpu will run at ~65 degrees full load, no worries but a re-occurring thing is after ~2 weeks it suddenly decides it wants to run at ~95 degrees, every two weeks. I thought i was overtightening the block thus forcing the TIM to flow out when hot because that's what i can see when i take the block off, but nope, even when i tighten by hand it only wants to behave for 2 weeks. I am not a wealthy man and decent thermal paste is expensive so i can't afford to re paste every other week. 

So I ask, fellow tech enthusiasts, what do i do because re-pasting every 2 weeks is expensive and mentally exhausting.

Thanks. 

What paste are you using? Thermal past lasts years…. Not weeks. 
 

Either something else is happening, or whatever TIM your using is not meant for CPU IHS to heatsink interfaces.

 

Also, how are you testing full load, what stress test? And at the 2 week mark are you doing the same stress test? I guess my question here is why continually stress test? If it’s stable it’s stable. What temps does it hit while playing games and doing your normal workload?

[Raz3R8023]

I use Mastergel pro from cm, 8W/mk decent for the application, didn't cause problems before, when i re-paste I let the system idle for ~2h and then stress with aida 64 to get that voltage- stability balance then i leave it there, i play a lot of cities:skylines (heavily modded) and do the occasional CAD render, that's when i notice ~2 weeks has passed when the qled reads 80+ the instant the load hits, It's winter here and i don't use heating in the pc room so ambient is sub 20. usually get ~65c before the 2 week event ingame (cities), ~35c - ~55c (Horizon 4) and i don't really feel comfortable cities with the thermal issue but what i did notice is the instant a load hits the temp rockets, no buffer till even the cold plate gets hot. I brushed it off till now because i thought it was due to me being a dunce when mounting but now i see there's a pattern, even lapping the cold plate and ihs didn't help, same 2 week thing, oh and i've delided too.

[LIGISTX]

4 minutes ago, Raz3R8023 said:

I use Mastergel pro from cm, 8W/mk decent for the application, didn't cause problems before, when i re-paste I let the system idle for ~2h and then stress with aida 64 to get that voltage- stability balance then i leave it there, i play a lot of cities:skylines (heavily modded) and do the occasional CAD render, that's when i notice ~2 weeks has passed when the qled reads 80+ the instant the load hits, It's winter here and i don't use heating in the pc room so ambient is sub 20. usually get ~65c before the 2 week event ingame (cities), ~35c - ~55c (Horizon 4) and i don't really feel comfortable cities with the thermal issue but what i did notice is the instant a load hits the temp rockets, no buffer till even the cold plate gets hot. I brushed it off till now because i thought it was due to me being a dunce when mounting but now i see there's a pattern, even lapping the cold plate and ihs didn't help, same 2 week thing, oh and i've delided too.

What are you using between the IHS and CPU die? Hopefully liquid metal?

 

I would get some thermal grizzly liquid metal for the die to IHS, and use those kryonaut for IHS to cold plate. With it being delid, the issue may not be in the IHS to cold plate surface, could be under the IHS… you removed the black sealant that originally held the IHS to the substrate? 

[Raz3R8023]

Liquid metal? nah lad, i wanted to go for the thermal grizzy cryonaut but no one sells it around where i live + lockdown, what you're saying makes absolute sense as the silicone to IHS gap is far larger then the IHS to cold plate gap and the lack of surface tension in thermal paste compared to LM may cause it to migrate when it heat cycles. i removed the sealant on the ihs and lightly lapped that surface to bridge the gap ( didn't want to overdo it and risk putting mounting pressure on the silicone), didn't re seal it, just held by the socket. My problem is why ~2 weeks?, very strange. 

[LIGISTX]

19 minutes ago, Raz3R8023 said:

Liquid metal? nah lad, i wanted to go for the thermal grizzy cryonaut but no one sells it around where i live + lockdown, what you're saying makes absolute sense as the silicone to IHS gap is far larger then the IHS to cold plate gap and the lack of surface tension in thermal paste compared to LM may cause it to migrate when it heat cycles. i removed the sealant on the ihs and lightly lapped that surface to bridge the gap ( didn't want to overdo it and risk putting mounting pressure on the silicone), didn't re seal it, just held by the socket. My problem is why ~2 weeks?, very strange. 

If you delid it... you need liquid metal. What are you using between the die and the bottom of the IHS? Normal thermal grease will not work... It doesn't migrate, if anything liquid metal will "migrate" more then normal paste will... normal paste will not suffice from die to IHS, you need liquid metal under the IHS.

[narrdarr]

when you stress test how long are you stressing for?

what liquid cooler are you using ?

how long have you had the liquid cooler?

what is the rad position ?

[Raz3R8023]

4 hours ago, LIGISTX said:

If you delid it... you need liquid metal. What are you using between the die and the bottom of the IHS? Normal thermal grease will not work... It doesn't migrate, if anything liquid metal will "migrate" more then normal paste will... normal paste will not suffice from die to IHS, you need liquid metal under the IHS.

I ment to say conductonaut, yes i use the mastergel pro inside the IHS and yes now it is obvious that this in not designed to bridge that large of a gap and definitely not for long term use. 

Thank you sir for confirming what I didn't want to admit, i'm gonna order me some conductonaut, even with $15 shipping for peace of mind.

[LIGISTX]

38 minutes ago, Raz3R8023 said:

I ment to say conductonaut, yes i use the mastergel pro inside the IHS and yes now it is obvious that this in not designed to bridge that large of a gap and definitely not for long term use. 

Thank you sir for confirming what I didn't want to admit, i'm gonna order me some conductonaut, even with $15 shipping for peace of mind.

Its actually not a very large gap under the IHS, but it is a very small amount oof surface area which is why you need something much more thermally conductive then paste. Do not use liquid metal outside the IHS though..... liquid metal eats copper, but itll be fine under the IHS.

[Raz3R8023]

6 hours ago, LIGISTX said:

Its actually not a very large gap under the IHS, but it is a very small amount oof surface area which is why you need something much more thermally conductive then paste. Do not use liquid metal outside the IHS though..... liquid metal eats copper, but itll be fine under the IHS.

By large gap I meant compared to the IHS- cold plate gap, An advantage of the 7700k is there's nothing else on the Substrate except for the silicone but precautions will be taken, and sir I think you meant aluminum?, gallium which LM is the majority of corrodes aluminum but just stains copper, almost no reaction with nickle, of which the IHS is plated.

[LIGISTX]

8 hours ago, Raz3R8023 said:

By large gap I meant compared to the IHS- cold plate gap, An advantage of the 7700k is there's nothing else on the Substrate except for the silicone but precautions will be taken, and sir I think you meant aluminum?, gallium which LM is the majority of corrodes aluminum but just stains copper, almost no reaction with nickle, of which the IHS is plated.

It definitely eats aluminum, I think it can pit copper, but maybe it is just a stain. Either way, definitely don’t use it externally to the IHS.

 

And the gap between die and IHS should be extremely minimal, I am not aware of it being any larger than IHS to cold plate. When you spread liquid metal on the die and IHS, it’s a very thin layer. I did this with my 4770k, it worked phenomenally well, temps dropped over 20c under load. 

[Jeppes]

Pump-out is normal if you don't use liquid metal between die and ihs. Thick pastes move a bit slower.

[Raz3R8023]

8 hours ago, LIGISTX said:

It definitely eats aluminum, I think it can pit copper, but maybe it is just a stain. Either way, definitely don’t use it externally to the IHS.

 

And the gap between die and IHS should be extremely minimal, I am not aware of it being any larger than IHS to cold plate. When you spread liquid metal on the die and IHS, it’s a very thin layer. I did this with my 4770k, it worked phenomenally well, temps dropped over 20c under load. 

I'm definitely not an expert but from what I know, the die - IHS gap is intentional to avoid mounting pressure on the die potentially cracking the silicone, which is far from ideal, IHS - cold plate gap is mostly due to the microscopic unevenness of the two surfaces. I'm not calling you a liar good sir, just some insight. 4700k?, that's a hot chip, we wouldn't be having this discussion if intel soldered the chip like a normal chip manufacturer, atleast for the k series. 

 

7 hours ago, Jeppes said:

Pump-out is normal if you don't use liquid metal between die and ihs. Thick pastes move a bit slower.

Explains the two weeks of usability, thanks lad.