Which Thermal Paste to buy and apply (Traditional and Liquid Metal)

Yeah, that's why I'm not such a fan of high refresh rate monitors - it forces the FPS up to the edge of performance of the GPU / CPU, and ramps up the heat / power higher than I think needed to enjoy a game on a laptop. On a desktop you have more margin to work with as far as cooling options, so there it's much easier to manage.

There are some games that do benefit from highest refresh and lowest frame gaps for fast action, but that's only a small percentage of games, unless that's your field of interest.

I've helped guys tune their high FPS games with CPU affinity - right click on the details entry for the game manager (like Steam) and then every game started from it will inherit the Affinity setting. Here's a how to on Windows 7, but it's the same on Windows 10 except you need to go to the Details tab in task manager to see the process / affinity setting.
https://www.tech-recipes.com/rx/37272/set-a-programs-affinity-in-windows-7-for-better-performance/

If the laptop can control hyper-threading in the BIOS - you want to turn hyperthreading off - it reduces heat more than it reduces performance - you can use affinity to uncheck half the threads, effectively reducing the hot cores putting out heat. It's very effective and hardly reduces the FPS. Then you don't have to fool around with CPU affinity - that's only when the BIOS doesn't let you disable hyperthreading.

Did you use brand new MX-4 2019 formula?, look at the link to it on Amazon and Newegg I put in my previous post - Edit2 - refresh the page to see it, and see if your tube matches - it's got a red band on it.


10 Best Thermal Pastes 2019

https://wiki.ezvid.com/best-thermal-pastes

You should be able to reduce your max refresh on a 200FPS-240FPS (360FPS) monitor too, set it 25% lower and see if that helps reduce temps enough - if you are running in the 90c range then you want to drop 10c or so into the 80c range. You may need to use a Custom Resolution setting to accomplish this, or use the steps available first to see if it helps - like 120hz, 144hz, a custom 180hz, etc.

Play around with those tips and the other ones, hopefully you can tune it so that if / when the temps start rising again when the MX-4 pumps out you can avoid repasting again as long as possible.

MX-4 is just a terrible paste. Anyone else have it constantly dry up?
https://www.reddit.com/r/overclocki...is_just_a_terrible_paste_anyone_else_have_it/

GravitonNg 7 months ago
"I tried Conductonaut on my both gpu and cpu previously and got great temps but after a year renewing them was a pain and it slight tarnishes the surfaces it touch, I didn't like that so I changed to Noctua's NT-H2.

Seems to work well but it has only been a few months, temps are a few degs higher as expected but I think they should last quite a while without drying out.

I'm confident because my last NT-H1 lasted 5years on my i7-2600k with a NH-D14 without a repaste, I only got kinda worried as I read advice saying we should repast every 2-3yrs just in case, temps were higher than I liked and I thought might as well do a quick reapplication while dedusting ye olde PC"

 

100% sure it's just the old in new lipstick http://forum.notebookreview.com/threads/clevo-overclockers-lounge.788975/page-1368#post-10591228


From sales link above... The new version of the #1 bestseller: the MX-4 Edition 2019 convinces everyone with its usual quality and performance

And bottom same link.


From my harsh ironic post in 2017. Nothing have really changed.
http://forum.notebookreview.com/thr...-and-liquid-metal.806840/page-6#post-10586708


Edit...

https://www.arctic.ac/eu_en/mx-4.html

 

So it's still garbage

 

Put it this way... There is better options for notebooks

 

@Papusan I've a new fav paste Cooler Master Maker Gel with new design. No spread necessary but its thicker than ICD 7/24.
You should try it on your kids PC. Idle temps seems to be 10C lower than older version while full load temps has variations due to ambient temps. If it ambients are within 24C then full load temps on 2C Pentium LGA Chips that suck more power is 55C if ambient is above 30C then it hovers around 65C.

 

New package design won’t make it any better


Cooler Master Redesigns its TIM Packaging so Parents Don't Think Their Kids are Doing Drugs techpowerup.com by btarunr Jan 17th, 2020

“The design, however, has a downside: the "pea drop" application is impossible, and a somewhat sub-optimal method is being forced upon users. Did Cooler Master just give in to pressure from an uninformed section of the market? Could a printed disclaimer on a conventional round syringe have helped instead?”

 

That was a stupid move. Just a label that says "THIS IS THERMAL PASTE, NOT DRUGS" would have sufficed.

 

Buildzoid finally gets to the point around 16:40 - it took 4 applications of LM before he got good temperatures...

Liquid metal on top of an overclocked 3950X kinda makes sense
Feb 26, 2020
Actually Hardcore Overclocking

 

An excellent video on thermal epoxy and paste for industrial applications below. He demonstrates lapping techniques, typical TIM materials, and provides the steps for DIY thermal epoxy.

He also recommends cheap indium sheets c.150 micro thickness for non-electrically sensitive applications; they flow into defects at about 100psi. Perforated indium sheets require about half the pressure to flow so might be a better option for lower pressure applications; has anyone here tried these?

He does not have recommend thermal pads and gallium compounds.

The presenter is fascinating. He clearly has a stellar applied science and engineering background but his history is a mystery.

 

Now I understand my parent's faces when I carry my thermal paste and apply heat gently! Hmmm....

 

Hi, is it normal that Grizzly Conductonaut seems dry after some time ? or at least when you dissamble the cooling, does the liquid metal seems like a rough surface instead of the shiny one at the first application ? my seringe is pretty old too, more than 3 years.
Also is there better liquid metal solutions out there ?
Thanks you

 

Yes this can happen with any LM.
You need to make sure contact pressure is even and firm and complete, without warped heatsinks either (Buy some Innovation cooling contact paper https://www.amazon.com/Innovation-Cooling-Contact-Analysis-Heatsink/dp/B07CKHRKHQ/ )
and you also need to insulate the border with something that will also block out oxygen. Air conditioner filter foam or fan filter foam will work when cut-out in the shape of the CPU (a few mm larger shape).

 

Its been two months since I lapped and LMed my GS75 and since then there have been some interesting developments.
While the cpu temps have remained outstanding, the gpu temps have started to creep up.

I took it apart and it seemed that the LM on the gpu side had started forming little pools leaving some areas with thin layers. Everything was fine on the cpu side however. Interestingly the LM still left some residue and the mirror surface was no more even though the copper heatsink had long since turned silver.

And so i relapped the heatsink but this time i used carbonaut on the gpu side and LM on the cpu side. Temperatures were outstanding on the cpu side but most interestingly the gpu temps were as good as when it had fresh LM on it. This might be because of the mirror finish on both surfaces allowing the carbonaut to achieve its max potential.

So a few things to takeaway:
1. LM does work fine on mirror finshed surfaces but on bigger dies say GPU dies the cohesive force of the LM may start to overcome the cohesive force causing it to pool up.
2. Carbonaut works amazingly well when both surfaces are mirror finished.

The results using carbonaut really surprised me as i just got it to experiment with since most reviews of the carbonaut simply branded it as "performs similarly stock paste" or "average performer compared to aftermarket pastes"

 

Thanks you, I never insulate the whole thing, I just put some paper towel in case it would bleed out but it never happen in a years of service with my 3920XM on my M6700 and I only sand paper my heatsink (P2000) but I think I will make a new heatsink. Temperature wise it worked well for a year but since my yesterday teardown it is very bad.
Do you have any article talking about liquid metal and oxygen ? what reaction does it cause ?

 

OUR PATENTED PROCESS BRINGS EXOTIC LIQUID METAL THERMAL COMPOUND TO NEW ROG GAMING LAPTOPS rog.asus.com | April 02, 2020

Liquid metals have low melting points that render them fluid at room temperature. These alloys are highly conductive, so they’re extremely effective at transferring thermal energy between surfaces. The benefits are well-established in overclocking and DIY circles, and internal testing reinforced the appeal for gaming laptops. Our engineers observed a 10~20°C reduction in temperatures depending on the CPU.

Although liquid metal’s properties are consistent regardless of the processor, our research indicates that Intel CPUs have the most to gain. The die is small, with heat concentrated in eight distinct regions of the chip. It’s also surrounded by a safe zone on the CPU package that’s free of surface-mounted components that don’t get along with electrically conductive materials. We want to deploy liquid metal where it can have the biggest benefit and be the most reliable.



There are different kinds of liquid metal on the market. We use Conductonaut from Thermal Grizzly because it has a lower concentration of tin, which isn’t as conductive as the gallium and indium that also make up the alloy.

 

So they machined a mask and have an automated brush, that's cool I guess.
"...Zero overflow" - proceeds to pump two globs of LM onto the die.

Well now it makes sense why they have that double sided tape around die, as everything underneath it will be swimming in LM.

 

That is pretty funny that it comments on zero overflow when it puts 3x as much LM as you would need. On the plus side, having a laptop with a warranty that doesn't exclude LM is nice.

 

Those are not blobs yet but still more than needed. Welp, they have to make sure theres enough left over when it alloys with the heatsink.

Speaking of which, they dont show any lm application on the heatsink side either. Maybe they are using nickel plated coldplates.

 

I hope they will not reduce the size of the cooling system because a better heat transfer allow it while keeping same temp :/

 

I expect same cooling as last years models but Liquid metal is needed now due higher boost clocks (more heat and higher base clock frequency) on old 14nm+++++ from Intel. The cheap mans fix!

 

Dunno about Asus, but HP laptops that have LM from the factory use nickel plating.

 

They are most likely using nickel plated heatsink for the cpu.

 

Here's MX-4 after 2 months:


And on the CPU:

That discoloration is frankly disgusting. DON'T USE IT FOR LAPTOPS.

By the end of March I was noticing a significant variation in core temperatures (Up to 12C), and my 10210U would shoot to over 90C within 10 seconds of a Prime 95 (~35W Package Power Consumption). I'm assuming the high temperatures simply exacerbated the degradation.

So I repasted with Noctua NT-H1:

My laptop didn't have any TIM between the PCH (Left small rectangle) but I decided to put some on anyway. I'm sure it doesn't hurt. Anyways temperatures are a lot more stable. Unfortunately I don't have any clear benchmarks because I was busy, but core variations are within 3C of each other, so that's a good thing I guess.

Unfortunately my laptop still can't handle a full run of CB20 (Package Power 38-40W) without reaching the 97C PROCHOT limit. That's because of the feather-light (literally) single heatpipe:

(Old picture, factory stock TIM)

Lenovo also has a second heatsink version for the same model of my laptop, but with two heatpipes (for an additional dGPU). When those parts are made available, I might try to see how well two heatpipes fare.

 

Instead of that Asus laptops should use beefier heatsink and not a shared heatsink.
I doubt none will invest in LM since a slight miscalculation or a runaway LM can kill your entire expensive BGA to a paperweight having BGA vendor logos which might impress a Business client!

 

Cooling systems with shared heatsinks are fine, they just need to use more of them or use thicker ones.

I'd say going to LM is a sign of progress though. It pushes laptop manufacturers to design for it and make it safe and long lasting.

 

Do you think it would also lead to void warranty for opening laptop on your own? I mean a disassemble.

 

But everyone wants slim laptop so thin shared heatsink is bad.
I like bigger heatsinks because my ambients are always >35C with 60% humidity.

 

A very bad idea. With help from Nvidia we will now get even worse cooling than ever from the notebook manufacturers. More Junkbooks with barely enough cooling for both Cpu and Gpu. Saved by Nvidia's Dynamic Boost!

Like SmartShift, NVIDIA’s Dynamic Boost is designed to take advantage of the fact that in many laptop designs, the GPU and the CPU share a common thermal budget, typically because they are both cooled via the same set of heatpipes. In practice, this is usually done in order to allow OEMs to build relatively thin and light systems, where the cooling capacity of the system is more than the TDP either of the CPU or GPU alone, but less than the total TDP of those two processors together.

This allows OEMs to design around different average, peak, and sustained workloads, offering plenty of headroom for peak performance while sacrificing some sustained performance in the name of lighter laptops.
https://www.anandtech.com/show/15692/nvidia-details-dynamic-boost-tech-and-advanced-optimus

 

IDK why OEM's try to get peak performance rather than sustained thermal performance. It makes advances in techs useless. I still have Ivybridge Pentium CPU that hold upto 35W at sustained usage but my BGA power throttles when it reaches more than 27W.

 

Asus doesn't have a sticker on the outside to show if you have disassembled it or not but it has a sticker on one of the heatsink screws which you can easily remove and put back.

 

All of the turdbook makers are oozing bad ideas from every orifice. In the rare accident that they actually have a good idea, it doesn't mean as much because those accidents are not great enough to offset the cumulative effect of all the bad ones. Unified heat sinks are a crappy design, but it's so common now that it has become an acceptable breach of intelligence. In order for it to be even sort of effective that would require that the thermal unit cooling capacity exceed the combined TDP under an abnormally severe load of both the CPU and GPU simultaneously for an extended period of time, (such as an indefinite and never-ending load,) sufficient for the thermal unit to reach the full saturation point without either processor ever reaching a temperature high enough to induce thermal throttling. Anything less than that represents abject failure and poor engineering. Nobody cares enough about their products to build them that way, and even if they did it wouldn't address the secondary issue of sloppy fitting garbage that they are equally famous for. Plus, they'd have to be thicker and heavier to not overheat, so the cry-baby whiners would come unglued over that. They value form over function.

 

From what I've seen high midrange and above gaming laptops rarely skimp on cooling.

Take my GS75 for example, I'm running a shunt mod and removed the cpu power limit so the cpu can run at 90w and the 2080 Max-q is runs at 130-140w. I can't run both at 100% usage due to the 230w adapter limiting it but the heat output is at least 1.5x higher than stock and the cpu is still running in the 85-90 while the gpu is sitting in the high 70s at max fans while looping metro exodus' rtx benchmark.

Midrange and lower is where I see manufacturers skimp on cooling like on the helios 300 and lenovo y540 with only one shared heatpipe, one on the cpu and one on the gpu

 

Well, I think you're giving way too much credit to cooling on high midrange and above turdbooks based on your own experience. The GS75 Stealth has a much better thermal design than other systems in the same class and better than most in a higher price range. You've proven that with the results you are seeing in temperatures. It's unfortunate that you needed to do those mods to help un-gimp it. Most similarly-equipped systems are running 15-20°C hotter at lower watts and clocks.

More heat pipes and more radiators and a third fan make a big difference. Just imagine how cool it would run if they made it an extra 1/4 inch thicker added another heat pipe, along with larger radiators and fans.

But, this is not the industry norm. Most of the products in the same class run much hotter, and it's because they were poorly engineered by people that don't care. Even my inexpensive Tongfang turdbook runs cooler than most costing much more because people are not getting what they pay for when they open their wallet to "level up" on an overpriced pile of trash.

 

I have a friend who also raised the gpu power limit by flashing a max-p vbios on the Asus GX701 and at max fans it hits mid 70s on the gpu and high 70s on the cpu. Granted the cpu power limit is still there so it's not producing as much heat as mine.

The Asus GX502 also seems to have significant thermal headroom as its gpu runs in the high 60s and cpu in the low 70s at max fan.

MSI's higher end tend to have better heatsink designs imo. Take a look at the new GS66's design.

The GE series isn't as good because of its cooling system hasn't been updated for a long time and I trust you've seen the GT76's cooling.

Acer's Triton 500 (same as Gs65) and helios 700 have good cooling as well.

Razer simply doesn't have enough heatsink surface area and dell just has garbage quality heatsinks where at least half are uneven.

The one I hate most would be gigabyte's aero series that thermal throttles at stock even when you apply liquid metal.

Oh and BTW, carbonaut is SINGLE USE ONLY.

It can't be reused unlike what thermal grizzly wants you to believe. If you have good mounting pressure, carbonaut will be squeezed extremely thin and will stick to the heatsink upon removal. Once you try to remove it from heatsink it will tear and be rendered unusable.

 

Yup, that's the sad truth. Unified heatsinks might look decent when you run either cpu or gpu intensive task, since they can utilize more of the available overhead thermal capacity.

 

After looking at several ryzen 4000 laptop reviews, I'm coming to the conclusion that the ryzen cpus could use liquid metal as much as intel's cpus.

Why? While the amd cpus are more efficient and use less power, they are 7nm and that makes their die size very small. That leads to a high thermal density, possibly as high as Intel's which could be why we are seeing such high temperatures with ryzen mobile cpus.

On desktops its not as noticeable as the heat is spread out by soldering a large IHS on top of the small dies, significantly reducing the thermal density but on laptops we are using regular thermal paste in between the die and the heatspreader which may be insufficient for dealing with such high thermal densities.

What are your thoughts on this?

 

Naked die cooling works well on desktops. Conclusion... Not good enough cooling on the Ryzen laptops.

 

I have not seen anyone do direct die cooling with paste before, only liquid metal.

Do you have any links?

 

Sorry if my post was unclear.

I meant of course with Liquid metal. Going with naked die and paste is stupid. All I meant is the heat spreader don't do much if the cooling is up to date. Put it this way... Newer chips have sTIM but the cooling will still be better if you go with LM and naked die.

Your reply #On desktops its not as noticeable as the heat is spread out by soldering a large IHS on top of the small dies, significantly reducing the thermal density#.

Processors regardless if its from Intel or AM will benefits from Liquid metal. But the end results won't be better than the total thermal transfer capacity.

 

You're missing the point, the issue here is the use of paste.

Laptops cooling systems are all direct die. And most laptops use paste by default and like you said, direct die with paste is stupid which is why I started this discussion in the first place.

 

Maybe I do. I replied more on this...

You said the problem wasn't as noticeable as the heat is spread out by soldering a large IHS on top of the small dies, significantly reducing the thermal density. This doesn't matter if the cooling is adequate and both chips use Liquid metal or sTIM. See... Better cooling with naked die vs the soldered large IHS on top of the small dies. The large IHS on top doesn't help much. Neither on large or small dies.

 

Many of us have been using liquid metal on our laptops for a long time, including @Papusan, and nothing is better under the right circumstances.

The problem is "right circumstances" with laptops. Too often using liquid metal is not enough. It does not work when parts fit is sloppy. And, it is not enough to make up for poor engineering and form factors that are to thin and light to function well, and lack the physical capacity for adequate thermal control. The only remedy for these trashbooks is that people will stop paying money for garbage and idiots will stop building garbage that nobody wants.

Unfortunately, that is not going to happen because we are outnumbered by silly people that value form over function, and they are going to continue wasting money on trash because it looks cute and their friends on Facebook are precious to them, and their acceptance in the clique is what they value most.

 

Let me rephrase that.

People don't feel ryzen desktop is hot because of the STIM + IHS, spreading out the heat from the small dies.

But now on laptops there is no STIM +IHS to spread out the heat. And using regular paste is like what Intel used to do, use regular paste under the IHS.

I'm not talking about the cooling system here.

 

Put a IHS on AMD's mobile chips in same way as for desktop chips (this to try reducing the thermal density on new small die), you'll get significantly worse cooling with same notebook cooling setup.

The Ryzen chips won't feel as hot is because of the lower max power consumption.

 

The heatspreader on laptops performs the same function as the IHS on desktops.

All I'm saying is that the current situation of cpu die -> paste ->heatsink in laptops would be like doing a cpu die -> paste -> IHS on desktops.

The paste is the thermal bottleneck here.

Desktops consuming less power than laptops? Really?

This is why I say you're missing the point.

I am just talking about the TIM in between die and heatspreader. Nothing more.

And to simplify it even more

Desktop: die -> solder -> heatspreader
Laptop: die -> paste -> heatspreader

 

It won't as long the cooling isn't good enough. Remove the IHS and add in naked die cooling on Clevo LGA laptop and you'll get better cooling.

Liquid metal won't save you from a bad working notebook cooling. Regardless... Of course paste is a worse choice than liquid metal. Have always been like that. But won't do magic if the cooling is a mess. Or that the cooling can't remove more heat than it already does with thermal paste.

Your post... I didn't talk about laptop in this context. It's about the chip in itself.

 

No it wont, but how good/bad the cooling system is is not what we're discussing here.

I never said anything about power consumption. I said people didnt notice the increase in thermal density of 7nm dies when ryzen desktop arrived because the STIM + IHS helped spread out the heat. The 7nm dies on desktop are just as thermally dense as on laptops, people just didnt notice.

Since we'll never see laptop heatsinks soldered to the cpu dies liquid metal might be the only way forward as cpus will be getting more and more thermally dense as we move on to 5nm, 3nm etc. I'm just asking for thoughts on this.

 

Put it this way. Pascal graphics was a hot mess vs. Turing due the smaller die. But liquid metal vs thermal paste on the smaller Pascal die didn't help as much as people thought.

 

Shrinking is never a positive thing. Bigger is better. It provides more room for awesomeness. Give me more of everything. More cores, more power (watts), more voltage, more cooling. Can't have more when you shrink those things. You can only find the bottom and then there is nowhere left to go.

What does that get us? Less of everything. Soldered turdbooks, fewer RAM slots, no RAM slots, fewer drive bays, no drive bays, fewer heat pipes, smaller radiators. Fewer options. The only thing you get more of is heat and problems, and more compromises. More lame excuses for mediocrity.

It is still true... "Go big, or go home."

Shrinking can be good when you cram more into the same space, or even better when take up more space than before, cram in more than ever, give it more power and more voltage and more cooling.

But, the dumb-dumbs don't do that. They shrink everything and the end result is you end up getting less. Doing more with less is always a bad plan. Doing more with more is a much better strategy.

 

Nvidia increased the Turing die size but added in the wrong cores We will see the same pattern with Ampere.

And more cores in the Cpu chips means lower all core boost. This so it can fits in the same TDP.