Thank you very much Korth for your detailed reply and research!

I totally agree with you on "going with the safe bet" on laptops, as they are quite expesnive. Believe me I tried. I spent a pretty dime on Kryonaut and the best pads I could find. But now every single more serious load turns my laptop into a jet engine. It's throttling left and right too. If one is lucky, you could always undervolt. But I also lost that bet. Mine can only do a measly ~ -0.045 undervolt which equals to almost nothing. If you can do ~ - 0.100, it's superb.

So, what is left? Repasting again with IC Diamond? Maybe that Copper thing? Meh. I am fed up with spending a day rebuilding AND risking my laptop.
Why Conductonaut? I checked Coollaboratory Liquid and people seemd to prefer Conductonaut over them. (Though I think they are pretty similar at the end of the day.)
I will post a text post once I am done. Would love to do vids but I lack a tripod or any kind of equipment.

Ps.: To be perfectly safe, I will use a clear nail polish, around 6-10 layers around the chip. I thought about using super glue, but getting off the polish might be much easier than the glue later on. There were "super glue removers" for sale in the past, but haven't seen them in years. So... yeah. If my heatpipe rots, I am in for an expensive repair. :')

https://rog.asus.com/forum/showthread.php?77546-Fourth-Repaste!-Gosh-Blue-Lady&p=662931&viewfull=1#p...

The main reason I'd avoid liquid metal or phase change/reflow TIMs is that laptops are mobile. They get moved, bumped, jostled. They get reoriented every time you carry them around in a bag. The TIMs don't leak out of socket (much) over time, but I think it's just asking for trouble.

Are you allowing sufficient "cure time" after TIM application? Have you experimented with different TIM application methods? I'm not saying "you're doing it wrong" (I have no idea how you're doing it anyhow, lol) but if you're consistently getting poor results then perhaps you should consider changing something, it doesn't hurt to go through the basics again and watch vids of other people repasting their machines, I've been pasting TIMs over silicon for years and still learn useful little details/improvements every now and then. If 10-20 minutes on youtube can postpone half an afternoon of laptop takeapart by a few more months then it's worth it.

I'd be apprehensive about using nail polish or superglue around a CPU socket. They have unknown electrical and thermal conductivities, they might partially reliquify when exposed to heat, they might contain reactive binders/additives or organic solvents which can damage socket plastics, solder masks, and PCB substrates after long-term contact/exposure. An electrical-grade conformal coating/sealant or potting epoxy would be a safer choice. Although I don't see the need to build up a sealing layer, it has dubious usefulness, might cause other problems, and would be nearly impossible to remove.

Well, I don't know how to thank you for the second barrage of super useful tips and ideas.

> Cure time
Well I heated up the machine, stress tested it. Especially since my GPU re-paste went haywire the first attempt. I purchased new Arctic pads, but to be honest, the factory thing is something else. These pads are solid you know, like small but thick post-it notes. Meanwhile ASUS uses something "sticky". If you checked your GPU components or youtube vids, you can see it's a blue sticky kind of thing. But if I would take apart my laptop right now, the Arctic pads did not deform or get sticky at all, they are just the same regular thick post-it notes.

So this was the first time I went for the GPU as well. Put a pea on it, you know, "da bess method". And of course replaced the heat pads. Put together my laptop only to see my GPU temps skyrocket - apparently the pea - even if it spreads - just does not cover the GPU die enough. So folks, don't be silly and believe memes - use the tool that Grizzly or your manufacturer gives you. It will cool just as fine.

That said I checked and re-applied my CPU paste too, it looked fine and it was cool. For a short few months.

> Flow out, burn in time
No I never used Liquid Metal in the past. In the past I sticked to generic, cheap pastes like MX-5. And to be honest on a desktop computer, I never felt the need to really even repaste because even the most basic and affordable aftermarket coolers kept the computer cool in any situation, no matter how much of a hack job I did on the pasting.

> Paste method
I used pea first, but it's just too risky. It either works, or not. Taking apart these ROG laptops are quite destructive, so take them apart as few times as possible.

> Sealant
Hmm. Other people used polish, my (stupid) idea was the glue. Well the idea for sealant is that IF it flows out because someone applied too much or because what you said (carry, shake, vibration), it should only flow out on to the surface you coated and not to the PCB itself - causing a short. The CPU has cover around by the way, but the GPU has none. This is the latest vid that comes close ( https://youtu.be/7glBCZr8ntI ) to my laptop - but yeah, my G750JZ has MXM slotted GPU and it's about a 100x times more destructive and harder to disassemble. Destructive because installing the mobo back makes it flex left and rigth AND there are the addon cards and cables and oh my god. Luckily the STRIX looks like an easier beast. Not sure if the "normal", big G752 is also easy to take apart. Mine isn't. For sure.

I'll check the shop here for some kind of heat resistant coating you just posted. It's still a few days until the Conductonaut arrives anyway.

(Out of desperation I tried to undervolt, had a few nice BSODs, I really did not win the sillicon lottery. But that much I already knew. I also tried using "FANContol" but the fan profiles were off and making my own profile seems a bit risky. In Intel Extreme Tuning Utility my package temps also just skyrocket into 90c zone way too often, so even if I had my own profile it would still just go up as well.)

ASUS is using blue goop now? I'm used to seeing their (old) pink goop, lol.

Most TIMs require many hours or a few days (or even a week) to fully "cure". They usually work fine until then, just at reduced efficiency, although running them too hot or too long before then can "thin" or "cook" them a little, enough to reduce their long-term performance.

"Pea", "dot", "rice grain", and "X" methods are all popular because they all work. But most people don't recognize when particular applications (part geometries, contact between mating surfaces, proper mounting pressure, etc) benefit more from particular application methods.

Conformal coatings are usually thermal and electrical insulators. They're typically a (transparent) polyurethane of some sort. They basically "laminate" electronics. They can cause issues with ESD-sensitive parts and can change the physical/electrical properties of inductive parts. They're basically impossible to remove without exotic chemicals and expensive washing machinery. Circuit traces and passive components generate very little heat but when they're laminated they're unable to dissipate any heat. Apply in a completely dust-free environment, as close to a sterile "clean room" as you can manage, and use ESD Although I still don't think they're necessary.

Sure I will check the alignment, I mean I am the most curious here about what went wrong - and why it worked for "so long". If you install it just plain wrong, temps will skyrocket in no time, like I had it with the GPU. Guess I will order a Kryonaut or IC Diamond (dunno which is the best, it's really hard to find a "best") and an LM paste like Conductonaut. If I fail with the LM (without breaking my machine), I will swap to IC Diamond. That's the plan at least.

I am at -0,075V undervolt - but in the past I had to tone it down to -0,045V. If someone gets a > -0,100 CPU that's real lottery.