My guess: you shouldn't have used the 1 mm pads over the GPU VRMs. They are too thick to allow the right pressure and contact between CPU and copper CPU heatsink
Maybe even the 0.5 mm could be too thick.
With my latest repaste on G751JY I went with thermal paste everywhere, because even with 0.5 mm pads I noticed higher temperatures on some CPU cores (more than 10°C differences from hottest and coldest core), so I thought that heatsink was not completely "flat" on the CPU die.
I initially went with 0.5mm for everything, but after I put the heatsink back on, I inspected the thermal pads from the sides with a lamp, and I could clearly see the light coming through between the VRMs and thermal pad, meaning there was no contact between them at all. That's why I went with 1mm for them. Other parts seemed to make contact with 0.5m just fine so I left them like that.
EDIT: Forgot to ask, what temperatures should I normally expect for cpu/gpu after cleaning everything and reapplying the paste?
I know that with 0.5 mm pads or with thermal paste the VRMs are not well in contact with heatsink, but I gave priority to CPU and GPU die contact-pressure-cooling.
Anyway, in my case, using thermal paste over the VRMs resulted in a well spreaded thermal compound: I considered it quite enough.
The fact that other parts make a good contact with 0.5 mm can be seen in a negative way, they are relieving pressure from the chip dies.
About the temperature, I assume you have the 4710HQ or 4720HQ. I don't know exactly because my 4710HQ is undervolted in daily use and overvolted/overclocked when needed, but with stock clocks and voltages you should not touch thermal throttling like you experienced, unless your room temperature is over 30 °C.
Maybe in the middle 80s seems to be plausible and acceptable when playing demanding games.
Right now, with an ambient temperature of 28 °C (it's hot here), my CPU idles at 38-42 °C with a little undervolting (that has little/zero impact on idle temps anyway).
Yes, I have a 4720HQ. it's not overclocked or undervolted right now. During stress test in Intel XTU, the cpu temperature doesn't go above 82 degrees but it keeps showing "Power limit throttling: yes". Also it doesn't let me undervolt any lower than 0.090. I tried -0.045 yesterday and there was no difference in temperature.
- 0.090 V is not a little undervolting, it's already a good result if your system is stable. Mine is -0.80 V, just for.
For Power limit throttling, you can raise the Turbo Boost Power from actual 47/58 W to 50/60 W if I remember correctly, and the Power Boost Windows from 8 seconds to maximum (I don't remember the value, maybe 64 sec).
This allows your CPU using 2-3 W more, with a little increase in clocks and temperatures as well. So in your situation this is not an improvement, because you are temperatures limited. So you should lower temperatures first.
In XTU you notice Power throttlink because it's a short test and temperatures didn't stabilize at higher values, and because you're stressing only CPU, while during games you are stressing both CPU and GPU, and the whole cooling system will show its limits.
I don't know if it's stable at -0.090 yet, I haven't tried lower than -0.045. Will try that later.
I was hoping I wouldn't have to open it up again, but it seems like I'll have to. It's a nightmare getting to the motherboard 😄
I know, I think I have opened it up dozens of times, now I can remove the motherboard in less than 20 minutes 😄
I'd give it a try, removing thermal pads. But it's your choice.
You can try with NBFC and a custom fan profile, to let them kick in with lower threshold. With NBFC, using 165 and 166 registers, you can also "unlock" the full speed of these fans, that should be about 4100-4150 rpm for CPU and 3800-3850 rpm for GPU fan. At these speeds, the noise is really high 😛
Hmm I'll give it a try. There's no risk like damaging the fans or something?
Btw, 20 minutes, nice man, that's really fast 😄