"Pump-out Effect" Is it real or is it nonsense? Let's discuss...

[lexluthermiester]

The subject of "pump-out effect" has been discussed in another thread. We need to bring that discussion here.

I think it's nonsense and isn't something that we users/enthusiast need to worry about, but there is another school of thought. There needs to be testing and experimentation to prove this is a thing. Why? Because until lately it's never been heard of. I personally have never seen an instance of it and I've been working on desktops, laptops and ultra mobile devices since the 1980's.

The below response is in reference to the following;

Pretty sure this video has been posted before in this thread but it seems that you were "too experienced" to watch through it

Curing time is BS

They were wrong on that one. Please see ArcticSilver5. Curing time is even stated by the maker of it.

Arctic Silver Incorporated - Arctic Silver 5

www.arcticsilver.com

Important Reminder:
Due to the unique shape and sizes of the particles in Arctic Silver 5's conductive matrix, it will take a up to 200 hours and several thermal cycles to achieve maximum particle to particle thermal conduction and for the heatsink to CPU interface to reach maximum conductivity. (This period will be longer in a system without a fan on the heatsink or with a low speed fan on the heatsink.) On systems measuring actual internal core temperatures via the CPU's internal diode, the measured temperature will often drop 2C to 5C over this "break-in" period. This break-in will occur during the normal use of the computer as long as the computer is turned off from time to time and the interface is allowed to cool to room temperature. Once the break-in is complete, the computer can be left on if desired.

This is a known performance metric industry wide. Steve and Der8uer are smart guys, but they are not Gods, they don't know everything and they sometimes make mistakes. Additionally, when Der8uer was talking about "pump-out-effect" he described 2000 heat cycles. I'm certain that he was talking about load-heat cycles.

 

[biffzinker]

I was having this problem with a 4790K running OCed using Noctua’s NT-H1 compound. The temperatures would start out normal. A couple weeks later the CPU would be throttling. After a repaste temperatures were back to normal. The cycle kept repeating for me.

Edit: The paste would appear squeezed out and there wasn’t nothing really on the IHS. The compound looked like it all was on the outer edge of the heatsink base.

 

[lexluthermiester]

I was having this problem with a 4790K running OCed using Noctua’s NT-H1 compound. The temperatures would start out normal. A couple weeks later the CPU would be throttling. After a repaste temperatures were back to normal. The cycle kept repeating for me.

Really? I've never seen this before. Is your 4790k delidded and the heatsink plate lapped?

 

[biffzinker]

I don’t have the 4790K anymore. I was using this heatsink. https://www.scytheus.com/aircooler-kotetsu

The 4790K was delidded with liquid metal.

 

[lexluthermiester]

Interesting. I currently have some NT-H1. Ran some tests with it. Perhaps it's time for some extensive and more extreme testing with it to see if I can replicate this effect...

 

[Mussels]

I don’t have the 4790K anymore. I was using this heatsink. https://www.scytheus.com/aircooler-kotetsu

The 4790K was delidded with liquid metal.

i had a 4770k with that issue, but the problem was always the paste being shifted by gravity under the IHS - that issue went away when i went to LM (conductonaut) but was a reoccuring problem with regular paste.

 

[lexluthermiester]

i had a 4770k with that issue, but the problem was always the paste being shifted by gravity under the IHS

That's heat induced TIM bleeding, not the same principle as what "pump-out" is described as.

See the following for how Der8uer describes it;

Time index 8:52

 

[biffzinker]

Interesting. I currently have some NT-H1. Ran some tests with it. Perhaps it's time for some extensive and more extreme testing with it to see if I can replicate this effect...

I originally blamed it on the mounting hardware. I was hand tightening the two screws in the crossbar. I wasn’t over tightening the screws.

 

[Space Lynx]

i had a 4770k with that issue, but the problem was always the paste being shifted by gravity under the IHS - that issue went away when i went to LM (conductonaut) but was a reoccuring problem with regular paste.

you use conductonaut with your 5800x now or no? i'd be to afraid of a single drop slipping out and frying mobo lol

 

[nguyen]

you use conductonaut with your 5800x now or no? i'd be to afraid of a single drop slipping out and frying mobo lol

Yeah I woud be worry about using LM on vertically mounted CPU too, same reason I wouldn't apply LM on laptop since I store the laptop vertically inside a backpack.
For horizontally mounted GPU there is no issue with LM

 

[Mussels]

you use conductonaut with your 5800x now or no? i'd be to afraid of a single drop slipping out and frying mobo lol

i did use it on the 2700x and 3700x, and it DID drip out with how often i changed coolers - i found some on my AX370 mobo near the unused USB 2.0 header, and decided to not use it on any more AM4 mobos

 

[mtcn77]

He used the wrong terminology. 'Pump out' empties the chamber, but 'caking' spikes up temperatures. If anything, 'Pump out' should lower resistance.

There is no difference between 'curing' during the set in period and in the long term, so it is a misnomer to depict the big picture. You cannot use the same term for two conflicting frames of references.

 

[XT785]

Hi! Has anybody experienced pump-out effect on laptops using gelid gc extreme thermal paste?

 

[MIRTAZAPINE]

I have question about this myself. Thermal paste "degrading" in laptops is a common problem I wonder if it was because of "pump out" or just simply the thermal paste drying out from the high temperarature reducing it thermal contact.

From what I read it seems to affect the more watery and less viscous paste that can spread easily example mx4 or thermal grizzly. Thermal paste that are viscous and less easily spread like shin-etsu which was the standard that many manucfacturer like Dell and Acer use for example use are not affected much. I guess shin-etsu was used as even with poor application you still can a decent cooling due to its viscousness.

 

[ThrashZone]

I was having this problem with a 4790K running OCed using Noctua’s NT-H1 compound. The temperatures would start out normal. A couple weeks later the CPU would be throttling. After a repaste temperatures were back to normal. The cycle kept repeating for me.

Edit: The paste would appear squeezed out and there wasn’t nothing really on the IHS. The compound looked like it all was on the outer edge of the heatsink base.

Hi,
Coolers now days aren't flat that if noticed just looking at your links coolers base plate it doesn't look flat either the reflection looks funny
Neither are cpu's so if these two surfaces are off the paste will fill those inconsistencies pretty much what paste is supposed to do

If the cooler and cpu are flat it would be normal for most paste to be on outer edges
Water blocks are designed with bows in them to better mate with cpu tops because cpu's are usually convex like a bowl not sure what a 4790k is like but hedt chips are large and convex my 10900k was nearly flat just a little convex barely noticeable

Took some images a while back of an ek magnitude cold plate bow if this cold plate didn't mate well with a cpu would that be said to be pump out because no paste was in the center ?

 

[Falkentyne]

I have question about this myself. Thermal paste "degrading" in laptops is a common problem I wonder if it was because of "pump out" or just simply the thermal paste drying out from the high temperarature reducing it thermal contact.

From what I read it seems to affect the more watery and less viscous paste that can spread easily example mx4 or thermal grizzly. Thermal paste that are viscous and less easily spread like shin-etsu which was the standard that many manucfacturer like Dell and Acer use for example use are not affected much. I guess shin-etsu was used as even with poor application you still can a decent cooling due to its viscousness.

It can be both.
Kryonaut has been known to pump out on laptops and lose stability at >80C. This loss of stability was apparently improved in Kryonaut extreme as well as MX-5 (TFX and SYY-157 also are stable).
This was a major complaint about Kryonaut. In fact, one of HIDEvolution's system integrators (they are an OEM builder for several laptops) absolutely refuses to use Kryonaut because of temp degradation.
This is also one of the reasons people using delidded LGA CPU's have had bad stability using Kryonaut on the delid instead of Liquid Metal.
Laptops can be very prone to this because the CPU BGA die isn't flat. It's convex. The convexity is what causes core temp delta issues (usually, when the cores are 0-8 numbered, often cores 1,3 (on 4C/8T) and in addition, 5 (8C) end up a lot hotter than the others (those cores will always be slightly hotter even on a sanded die and great fitting heatsink). Combine that with imperfect heatsinks and you have a problem.

I encountered this issue myself with Kryonaut on my MSI GT73VR. This did NOT happen on my desktop CPUs however.

MX-4 has been known to turn into soup at >90C on laptops that are thermal throttling. I still can't find the picture but someone a few weeks ago posted a picture of this. It was on a GPU though. Nothing but liquid oils on the GPU Core in a wavy pattern and the actual paste was off to the edges.

 

[Shrek]

The subject of "pump-out effect" has been discussed in another thread. We need to bring that discussion here.

I think it's nonsense and isn't something that we users/enthusiast need to worry about, but there is another school of thought. There needs to be testing and experimentation to prove this is a thing.

How about this for testing; two blocks of metal separated by thermal grease and the top one with an out of balance wheel to achieve relative motion (pump out is supposed to be due to relative thermal expansion).

There is a certain logic, grease is carried out from the moving interface and caught on the edge when the motion returns.

I suspect that some thermal greases have lower thermal conductivity so that they remain sufficiently gooey to flow back into the crack using surface tension; this is where surfactants may play a role

Seems oleic acid may be a candidate (boiling point 360°C)
How To Make Ferrofluid (popsci.com)

I'm just playing with ideas here

 

[unclewebb]

I think it's nonsense

I think it is an issue for laptop CPUs that run for hours at 90°C or higher while gaming. Direct die or a heat spreader like desktop CPUs use seems to make a difference.

Hard to argue with what this video shows. There is virtually no thermal paste left on the heatsink or on the CPU and GPU die where there obviously used to be paste.

 

[Space Lynx]

I think it is an issue for laptop CPUs that run for hours at 90°C or higher while gaming. Direct die or a heat spreader like desktop CPUs use seems to make a difference.

Hard to argue with what this video shows. There is virtually no thermal paste left on the heatsink or on the CPU and GPU die where there obviously used to be paste.

this must be why noctua made the nt-h2

 

[unclewebb]

nt-h2

I have heard from users that NT-H2 works much better long term in the laptop environment. TechPowerUp now uses NT-H2 on their test systems.

 

[Shrek]

NT-H1 is good to 110°C
NT-H1 3.5g || Specifications (noctua.at)

NT-H2 is good to 200°C
NT-H2 3.5g || Specifications (noctua.at)


Even more interesting is they don't bother to list the thermal conductivity; I think this is a reflection that it is not that important.

 

[Mussels]

"good to x" is by whatever metric the maker decided upon

Good to 200C for... a month? a year? Good thermal conductivity?

Even with quality brands like noctua, incomplete information can lead to very wrong assumptions about what the marketed specs mean

 

[Fangio1951]

I've had my i7-7700 k delided and using Conductonaut LM for about 3+ years, with the mobo mounted vertically, in a Thermaltake Core P5 case and the temps haven't varied = normal - 26-30 C and stress = 52-55 C max.

System is fully water cooled = mobo, GPU and CPU.

 

[freeagent]

I haven't experienced it myself.. but curious if it could be from a lack of mounting pressure? Some guys are afraid to turn screws all the way, or cooler just has a weak mount? My LGMRT on Intel is super tight, but on AM4 I can deal out full titty twisters even after the screws are done turning.. No problem with my new cooler.. she's tight.. Again no pump out though, so it has to be something else.. maybe the formulation of the TIM itself, or some other QC issue somewhere down the line.

I understand thermal expansion cycles, but enough for it to pump out like the name implies leaves me with other questions, mainly on the design of the cooler itself and its mount, or the logic behind the engineering that went into it. I don't know.. I am no engineer.. but had I stayed in school and made better choices who knows lol.

 

[delshay]

They talk about most thermal paste use aluminium powered, but Arctic Silver 5 has real particles of silver, tested in a lab many years ago. Real silver has a higher thermal conductivity than aluminum.

My personal choice when it comes to thermal paste is Arctic Silver 5. I just wish all of today's modern paste can be sent to lab to verify what's in them as some were caught cheating many years ago with what it contains & they thermal conductivity claim. ...I just know OCZ at that time was one company caught lying what it contains, but there were other company's caught lying too.