But Intel Core i7-8700 only as nominal has TDP 65W. Real TDP about 105W. That is why I have 100C of CPU temperature after only 10 minutes on 100% CPU loading.
Those numbers are crazy! You're not overclocking the CPU are you? If not, then either Intel lied about the TDP or there's something seriously wrong with your cooler setup.
I replaced stock termal tape on Thermal Grizzly Kryonaut, (TG-K-030-R), but still it is not enough. 2 of case sides warms up to 70C when CPU has 100C.
I just used regular thermal paste everywhere — a combination of the stuff that came with the DB4 and LH6, some Arctic Silver 5, and Noctua NT-H1. Probably 30g all-up (for CPU and GPU combined). I coated the inside of all of the channels that heat pipes would slot into, and applied liberal amounts to the exposed pipes and the spreader sections between them. I probably over-did the TIM — to be honest — but it seems to have worked.
If your walls are getting to 70°C then it would seem that your cooler setup is actually working (if it wasn't, then the CPU would be hot, but the walls would be cool). That would suggest even more strongly that Intel just lied about the TDP figure.
I think heat sink should be fitted to case sides more tightly than design can alow . Universal brackets are too fexible and also, case sides still have small gaps between each other :-(
I just popped the cover off my DB4 and inspected the heat spreaders. They are all nicely and evenly bonded to the walls. Nothing seems to have moved since I put it all together. The brackets are snug and seem to be doing their job (of holding spreaders in place) just fine.
I'm not sure what you mean by gaps between the sides of the case. Mine interlock very, very tightly — you couldn't slip a sheet of paper between them even if you tried.
What is your temperature of the case surface during 100% CPU loading?
I wrote a couple of articles about the thermals, starting with: Passively-cooled CPU Thermals
Short answer: In a 20°C room, sustained 100% CPU loads result in the CPU stabilising at 58°C and the hottest part of the hottest exterior wall reaches 42°C. The graphs only show a one-hour test period, but since then I've had overnight jobs (~6 hours @ 100% load) and the results are the same.
The hottest wall is the one that cools the CPU using the three short (stock) heatpipes (i.e. the wall directly opposite the CPU). Unsurprisingly, the hottest part of that wall is immediately on the other side of the heat spreader. Measuring even a few centimetres away from this point results in temperatures that are much lower.
Based on what you've written, it would seem that heat
is getting to the walls, so the cooling system
is working as it is supposed to. Rather than try to blame the case for the high temperatures, maybe it's time to consider the increasingly-likely possibility that Intel simply lied to you about the TDP of the 8700K?
Last month I attached a power meter to my computer to measure power consumption in various states and under various loads. At 100% CPU load the
entire system draws ~75W. Given that our systems are very similar, you should be able to do the same test and get a very similar result. If your system draws substantially more than 75W then the problem's not the cooling or the case, it's the CPU. A power meter reading » 75W would prove that Intel lied.
