[EOL] Arctic MX-5 is here!!Tests incoming! Completed. Now its MX-6 testing time!

[Shrek]

Hard to remove might actually be a very important quality to avoid pump out.

 

[Space Lynx]

Hard to remove might actually be a very important quality to avoid pump out.

i mean mx-4 never had any issues, nor did mx-2 to my knowledge. both perform better than mx-5, and are easier to clean for re-pastes... so it just seems like mx-5 loses all the checkboxes to me... i don't fully know what you mean by pump out though...

 

[Shrek]

Pump out, perhaps one of the most important aspects of thermal grease

As the chip thermally expands and contracts relative to the heatsink, it slowly pumps out the thermal grease; it may take thousands of cycles.

(1) Der8auer Deep-Dives on Thermal Paste: Misconceptions, Curing, & More | LTX 2019 - Bing video

 

[8BitZ80]

Pump out, perhaps one of the most important aspects of thermal grease

As the chip thermally expands and contracts relative to the heatsink, it slowly pumps out the thermal grease; it may take thousands of cycles.

(1) Der8auer Deep-Dives on Thermal Paste: Misconceptions, Curing, & More | LTX 2019 - Bing video

Pump-out might be relevant for thinner pastes but normally thermal paste needs to be replaced because the paste itself has expired. For a stock Intel heatsink I've found that the paste goes bad after about five years. Typically that's because the paste has dried out and isn't making good contact, which has very little to do with the original viscosity (stock heatsink pastes are very thick). There is no singular trait that determines whether a paste will be good or bad.

 

[lexluthermiester]

@lexluthermiester what's taking so long on those temps?!?!?! womp... I still can't decide what paste to buy.. lol

I'm not done yet and I've had to restart. Though I can tell you, keep your MX-5...

In case you're wondering, I had a fan fail during testing which invalidated all of the tests because there was no way for me to know if the fan was causing variances. So, I ordered a new fan and am waiting for it to arrive. Murphy's law in effect...

 

[Shrek]

Pump-out might be relevant for thinner pastes but normally thermal paste needs to be replaced because the paste itself has expired. For a stock Intel heatsink I've found that the paste goes bad after about five years. Typically that's because the paste has dried out and isn't making good contact, which has very little to do with the original viscosity (stock heatsink pastes are very thick). There is no singular trait that determines whether a paste will be good or bad.

I am not convinced that dried out paste is not effective, after all an indium pad works just fine, and it might be a way around pump out.

Take a grease like SuperLube thermal compound; it is good to 260°C so should have no problem not evaporating at 100°C; now GD900-1 is only good to 120°C and that does not seem a good enough safety margin to me (the original GD900 is good to 200°C)

Let's use Arrhenius' law (probably not applicable here) which say reactions double in rate for ever 10°C, then a safety factor of 20°C gives one 4 times the life (GD900-1), while a factor of 100°C gives a factor of 1000 (GD900) and 160°C gives 65,000 (SuperLube)

Noctua NT-H1 has a 110°C top temperature NT-H1 3.5g || Specifications (noctua.at)
Noctua NT-H2 has a 200°C top temperature NT-H2 3.5g || Specifications (noctua.at)

MX-5 specs states: Limited 8 years Durability (as does MX-4)
MX-5_EN_V02.indd (arctic.de)

If all we are interested in is day 1 performance we may end up with the wrong compound.

 

[8BitZ80]

I am not convinced that dried out paste is not effective, after all an indium pad works just fine, and it might be a way around pump out.

Take a grease like SuperLube thermal compound; it is good to 260°C so should have no problem not evaporating at 100°C; now GD900-1 is only good to 120°C and that does not seem a good enough safety margin to me (the original GD900 is good to 200°C)

Let's use Arrhenius' law (probably not applicable here) which say reactions double in rate for ever 10°C, then a safety factor of 20°C gives one 4 times the life (GD900-1), while a factor of 100°C gives a factor of 1000 (GD900) and 160°C gives 65,000 (SuperLube)

Noctua NT-H1 has a 110°C top temperature NT-H1 3.5g || Specifications (noctua.at)
Noctua NT-H2 has a 200°C top temperature NT-H2 3.5g || Specifications (noctua.at)

MX-5 specs states: Limited 8 years Durability (as does MX-4)
MX-5_EN_V02.indd (arctic.de)

If all we are interested in is day 1 performance we may end up with the wrong compound.

I don't know where you get all these theories from but you must have a head like Megamind from overthinking everything.

I am not convinced that dried out paste is not effective, after all an indium pad works just fine, and it might be a way around pump out.

I've seen Core i3 processors with the original heatsink thermal throttling at 90C, and after replacing the thermal paste barely touching 50C. Dried out paste definitely isn't effective. Indium pads, yeah they sound great, they're a completely different product though and we gotta be careful to not compare olive oil to motor oil.

Take a grease like SuperLube thermal compound; it is good to 260°C so should have no problem not evaporating at 100°C; now GD900-1 is only good to 120°C and that does not seem a good enough safety margin to me (the original GD900 is good to 200°C)

Let's use Arrhenius' law (probably not applicable here) which say reactions double in rate for ever 10°C, then a safety factor of 20°C gives one 4 times the life (GD900-1), while a factor of 100°C gives a factor of 1000 (GD900) and 160°C gives 65,000 (SuperLube)

Lets say I've got a cup of water at 25C. It freezes at 0C and boils at 100C. Does that mean that it's got a 25% chance of boiling and a 75% chance of freezing? Instead it's actually doing nothing at that temperature. What I'm saying is that a paste that breaks down at 260C provides no indication of what it might be doing at 100C, or 80C, or any other temperature. For all you know it might have a melting point of 150C and below that temperature it's actually a solid. I see what you're trying to do with extrapolating results but it doesn't really mean anything. A paste rated for 120C will perform just fine at 120C, however beyond that temperature there's no guarantee that it will be chemically stable.

 

[Mussels]

^ what... sigh, "poopipeboy" said
you're overthinking things, and using info in the wrong context

Fire can handle thousands of degrees, it must be the best thermal compound - flamethrower heatsinks anyone?

 

[ThrashZone]

I am not convinced that dried out paste is not effective, after all an indium pad works just fine, and it might be a way around pump out.

Take a grease like SuperLube thermal compound; it is good to 260°C so should have no problem not evaporating at 100°C; now GD900-1 is only good to 120°C and that does not seem a good enough safety margin to me (the original GD900 is good to 200°C)

Let's use Arrhenius' law (probably not applicable here) which say reactions double in rate for ever 10°C, then a safety factor of 20°C gives one 4 times the life (GD900-1), while a factor of 100°C gives a factor of 1000 (GD900) and 160°C gives 65,000 (SuperLube)

Noctua NT-H1 has a 110°C top temperature NT-H1 3.5g || Specifications (noctua.at)
Noctua NT-H2 has a 200°C top temperature NT-H2 3.5g || Specifications (noctua.at)

MX-5 specs states: Limited 8 years Durability (as does MX-4)
MX-5_EN_V02.indd (arctic.de)

If all we are interested in is day 1 performance we may end up with the wrong compound.

Hi,
If I make it through 6 months before I alter my systems that I'd change the paste I'd be surprised so 1-2-3-4/.... years durability is great but a wasted spec since it would never see that long installed
Shelf life well that would be a different story

 

[Shrek]

Lets say I've got a cup of water at 25C. It freezes at 0C and boils at 100C. Does that mean that it's got a 25% chance of boiling and a 75% chance of freezing?

At 75°C it evaporates much faster than at 25°C; the problem is the Boltzmann distribution of velocities, some have what is takes to 'boil' off; so actually it both boils and freezes, some evaporates off, the rest cools down.

Taking into account evaporation and pump out is overthinking? These are real things.

A paste rated for 120C will perform just fine at 120C, however beyond that temperature there's no guarantee that it will be chemically stable.

A capacitor rated at 105°C will perform just fine at 105°C? Have you seen their rated life at that temperature? months, not years.

Fire can handle thousands of degrees, it must be the best thermal compound - flamethrower heatsinks anyone?

Straw man argument.

 

[Falkentyne]

At 75°C it evaporates much faster than at 25°C; the problem is the Boltzmann distribution of velocities, some have what is takes to 'boil' off; so actually it both boils and freezes, some evaporates off, the rest cools down.

Taking into account evaporation and pump out is overthinking? These are real things.


A capacitor rated at 105°C will perform just fine at 105°C? Have you seen their rated life at that temperature? months, not years.


Straw man argument.

Its the same reason why dried out oxidized liquid metal gives worse temps on low pressure heatsinks and needs to be replaced.
Gallium gets absorbed into the copper, causing the remaining low gallium+indium+tin mixture to harden because there's too little gallium for it to remain liquid anymore. Combine that with the regular thermal expansion and contraction, you now have imperfect, bad contact and major hot spots with what's left. (There are ways to limit / prevent this).

With thermal paste, this is even worse because the components themselves end up separating and drying out and once again you are left with imperfect contact and very poor heat transfer.

Some pastes just completely break down (MX-4 has turned into soup for some people, on very hot, high wattage chips).

Some pastes will become dryer and thicker (like a putty) but will not separate and will maintain their thermal properties while becoming even more resistant to further shifting around (Thermalright TFX and SYY-157 paste is like this).

 

[lexluthermiester]

Some pastes just completely break down (MX-4 has turned into soup for some people, on very hot, high wattage chips).

In these cases, it was applied too thick and the excess dripped away under load. TIMs do not need more than just a tiny amount, spread very thinly.

I am not convinced that dried out paste is not effective

And you would be correct. TIM's do NOT loose effectiveness after "drying" out. In most cases, the fluid components are only a delivery mechanism and do not actually assist in the function of heat conduction.

 

[ThrashZone]

In these cases, it was applied too thick and the excess dripped away under load. TIMs do not need more than just a tiny amount, spread very thinly.


And you would be correct. TIM's do NOT loose effectiveness after "drying" out. In most cases, the fluid components are only a delivery mechanism and do not actually assist in the function of heat conduction.

Hi,
Indeed too many watching GN on youtube all his chips look like he uses an entire tube on one mount lol

 

[Space Lynx]

Hi,
Indeed too many watching GN on youtube all his chips look like he uses an entire tube on one mount lol

to my knowledge he only did that once or twice and was just part of the testing to see if it mattered or not temp wise. i doubt he actually uses that much in a real world build.

 

[ThrashZone]

to my knowledge he only did that once or twice and was just part of the testing to see if it mattered or not temp wise. i doubt he actually uses that much in a real world build.

Hi,
He shows applying excessive thermal paste, grizzly to be exact on the beginning of every youtube video I believe lol

 

[lexluthermiester]

Hi,
Indeed too many watching GN on youtube all his chips look like he uses an entire tube on one mount lol

I roll my eye's everytime I see Steve or even Jay apply a TIM... just wasteful..

 

[ThrashZone]

I roll my eye's everytime I see Steve or even Jay apply a TIM... just wasteful..

Hi,
Have you seen GN plastic case with chips in it ?
I saw it all of them were soaked in paste it was pretty funny should of saved it lol

 

[Shrek]

I had a fan fail during testing which invalidated all of the tests because there was no way for me to know if the fan was causing variances. So, I ordered a new fan and am waiting for it to arrive. Murphy's law in effect...

I have a tendency of finding a way to get some oil into the bearing (assuming it was that which failed)

 

[Space Lynx]

I have never had a fan fail, like ever. That being said, I switch out all my gear every 5-7 years on average, and I turn my PC off when not using it. Also, I probably just don't use it as much as other people.

 

[lexluthermiester]

I have a tendency of finding a way to get some oil into the bearing (assuming it was that which failed)

It wasn't the bearing or I would have reoiled and started over. It was a motor coil burn out. The smell was nasty.

I have never had a fan fail, like ever.

It happens from time to time. I see one fail like this every few years.

 

[Falkentyne]

In these cases, it was applied too thick and the excess dripped away under load. TIMs do not need more than just a tiny amount, spread very thinly.


And you would be correct. TIM's do NOT loose effectiveness after "drying" out. In most cases, the fluid components are only a delivery mechanism and do not actually assist in the function of heat conduction.

We're not talking about that. We're talking about pastes degrading in a few months, and needing reapplication (or a completely different paste).
Applying too much isn't going to affect this. It's going to get squeezed out anyway.
I already tested this on my GTX 1070, R9 290X and RTX 3090.
I yeeted the TFX and Kryonaut Extreme and MX-5 (MX-5 and Kryo Extreme tested on 290X and 1070 only) by intentionally applying "too much". There was no variance in longevity. This was easy to see with Kryonaut Extreme and MX-5, after some high temp torture testing (Intentional 95C on 290X) then back to normal temps to check for degradation.

In fact, on my last repaste on my 3090 FE, I intentionally yeeted the TFX by painting over the entire core (and a second thinner layer on the heatsink) instead of doing the X + dots method, and got the best results this way. Got this information from notebookreview (due to them fighting constant convex cores and imbalanced heatsinks, some very experienced people there).

(Some new updated paste results also. Kryonaut and Kryonaut Extreme are on a previous page with a different application).

TechnologyGuide

Thank you for visiting the TechnologyGuide network. Unfortunately, these forums are no longer active. We extend a heartfelt thank you to the entire community for their steadfast support—it is really you, our readers, that drove

forum.notebookreview.com

As far as 'Applying too much' and spreading: I was using the X pattern with dots with TFX until I read this:

TechnologyGuide

Thank you for visiting the TechnologyGuide network. Unfortunately, these forums are no longer active. We extend a heartfelt thank you to the entire community for their steadfast support—it is really you, our readers, that drove

forum.notebookreview.com

Now I just follow Seanwee's method.

TechnologyGuide

Thank you for visiting the TechnologyGuide network. Unfortunately, these forums are no longer active. We extend a heartfelt thank you to the entire community for their steadfast support—it is really you, our readers, that drove

forum.notebookreview.com

 

[lexluthermiester]

Now I just follow Seanwee's method.

TechnologyGuide

Thank you for visiting the TechnologyGuide network. Unfortunately, these forums are no longer active. We extend a heartfelt thank you to the entire community for their steadfast support—it is really you, our readers, that drove

forum.notebookreview.com

Oh I'll pass on that.

 

[Shrek]

We're not talking about that. We're talking about pastes degrading in a few months, and needing reapplication (or a completely different paste).

My Mini-mind reasoning was that if the paste can take 260°C then it probably is not going to degrade at 100°C

 

[Falkentyne]

My Minimind reasoning was that if the paste can take 260°C then it probably is not going to degrade at 100°C

Yeah I get you. Everyone here is making some good points and no one is what I would consider "wrong."
But with thermal pastes theres just so many things that can cause things to go south.
Counterfeiting, imbalanced heatsinks, convex cores, terrible mounting pressure, you name it. That's why notebookreview forums has been a great learning place for discussion about thermal pastes (and pads) because the laptop users have always had the worst of it.

Also, always take any spec with the smallest grain of salt possible. Just because a paste is rated to take a certain temp doesn't mean it can take it long term. We've already seen plenty of posts about pastes breaking up or as you mentioned yourself, thermal stresses and pumping out, and the weaker the mounting pressure, the worse the pump out effect.

I'm still trying to find a picture of the "Soupy" MX-4. I don't even remember what forum it was on, because I disregarded it (I don't use MX-4 anymore, even though I still have a 6 year old tube--I even remember where I bought it from!). Someone was complaining about high temps (degraded with time). But it was not on a 'small' BGA die. It was either on a GPU or IHS. The application itself looked "fine". Not too much was applied. There was the normal expected amount outside the chip and the chip was completely covered up. The problem was the appearance of the paste. It looked like the consistency was completely ruined, I can't describe it, except as "soupy and wavy", and the only part that looked like normal paste was the part outside the chip. It was clear that the paste had completely separated into its liquid vs solid components despite the liquid components covering the entire die.

 

[80-watt Hamster]

We're not talking about that. We're talking about pastes degrading in a few months, and needing reapplication (or a completely different paste).
Applying too much isn't going to affect this. It's going to get squeezed out anyway.
I already tested this on my GTX 1070, R9 290X and RTX 3090.
I yeeted the TFX and Kryonaut Extreme and MX-5 (MX-5 and Kryo Extreme tested on 290X and 1070 only) by intentionally applying "too much". There was no variance in longevity. This was easy to see with Kryonaut Extreme and MX-5, after some high temp torture testing (Intentional 95C on 290X) then back to normal temps to check for degradation.

In fact, on my last repaste on my 3090 FE, I intentionally yeeted the TFX by painting over the entire core (and a second thinner layer on the heatsink) instead of doing the X + dots method, and got the best results this way. Got this information from notebookreview (due to them fighting constant convex cores and imbalanced heatsinks, some very experienced people there).

(Some new updated paste results also. Kryonaut and Kryonaut Extreme are on a previous page with a different application).

TechnologyGuide

Thank you for visiting the TechnologyGuide network. Unfortunately, these forums are no longer active. We extend a heartfelt thank you to the entire community for their steadfast support—it is really you, our readers, that drove

forum.notebookreview.com

As far as 'Applying too much' and spreading: I was using the X pattern with dots with TFX until I read this:

TechnologyGuide

Thank you for visiting the TechnologyGuide network. Unfortunately, these forums are no longer active. We extend a heartfelt thank you to the entire community for their steadfast support—it is really you, our readers, that drove

forum.notebookreview.com

Now I just follow Seanwee's method.

TechnologyGuide

Thank you for visiting the TechnologyGuide network. Unfortunately, these forums are no longer active. We extend a heartfelt thank you to the entire community for their steadfast support—it is really you, our readers, that drove

forum.notebookreview.com

Opening caveat: I don't re-paste many open dies.

It wastes a few tenths of a gram, but I've landed (for the moment) on spreading the contact area with the edge of an old credit card (or equivalent).