Hello. I have a temp issue with my R7 3700x high temp underload and idle with Noctua NH D15s. idle is above 45C, and underload like running Aida64 default selected option stress test cause the temp is go above 85C on HWinfo64. Even playing PlanetSide 2 cause the CPU temp to 77C and slowly reaching to 80C. I did reapply thermal paste with 3 dots and check the mounting, and it seems fine to me. Ryzen Balance Power Plan selected.
ROG Strix X470F
R7 3700x
2x16GB G.Skill DDR4 3200MHz RGB
Corsair 750D with 2x 140mm fans intake front and 1x 140mm exhaust back
Ambient is 30C. Bring down ambient from 30C to 26C doesn't help the temp
Case off doesn't help either
Your temperatures for an Air cooled CPU Cooler seems to be within the Processor's Maximum Operating Temperature of 95C.
If under heavy loads or stress, the Processor's temps goes over 95C then you may need to upgrade to a liquid CPU Cooler to keep the temperature at or below 95C - Maximum Operating Temperature of the Processor.
80C is not that bad for the processor while playing a modern game.
If your CPU Cooler only has one Fan you may want to add a second fan to the CPU Cooler heat sink in a PUSH-PULL combination. It may help to keep the temperatures more stable during AIDA64 testing.
Also open one side of the computer case and see if the Processor's temperatures are lower. If they are that indicates poor air circulation in your computer case.
According to Noctua FAQ concerning your specific CPU Cooler model here is what is says about high temps on the new 3xxx processors: https://noctua.at/en/products/cpu-cooler-retail/nh-d15s/faq
Disable PBO doesn't do anything brother. Watching a Youtube video with two other tabs open in FF causing max temp raise close to 70C and over 70C too! Ryzen Master and HwInfo64 are not too far off each other, so for AMD recommend Ryzen Master over HWInfo64 is kinda nonsense to me.
Liquid cooling is largely irrelevant with Matisse. "The Stilt" has a pretty good primer on it over on overclock.net in his "Matisse, strictly technical". Here is an excerpt.
Liquid cooling is great for Threadripper as it generates a ton of heat that needs to be dissipated. Matisse on the other hand, generates a very small amount of heat, but also in a very small area. The ability of the IHS to pull the heat out of the cores becomes the rate limiting step.
That being said s_vayner has the same cooler that he is using on a 3900X with great results. Maybe he can chime in on how he installed it?
The 3700X isn't binned as high as the 3800X or 3900X so the voltage/frequency curve may not be the greatest. That's why I would shut down PBO and let the 3700X boost within the normal confines of a 65W processor. If it is still getting hot, then there is definately a problem with the cooler setup.
The issue is, transferring the heat from the cores to the IHS (integrated heat spreader). The actual part of the CPU die the generates most of the heat is much smaller in area in Zen 2 than in previous generations, so the IHS is contacting less of the hot surface. So the rate the heat diffuses out of the core into the IHS is slower. It's simply an issue of thermodynamics, and one that will predominately affect processor performance as these circuits get smaller.
The cooling solution then, is pulling the heat out of the IHS and then dissipating it into the air and exhausting it out of the case. Yes, liquid has a higher specific heat than air, so it accepts the heat more readily and is better able to cool the IHS. But the IHS heating up isn't the issue, the issue is the rate at which you are transferring heat from the cores to the IHS to begin with, which is fixed.
The voltage he is referring to, is the FIT (Silicon fitness monitoring feature). This is a safe voltage limit the processor will not boost beyond to keep heat density in check. It is why Zen2 CPUs give the exact same single threaded boost scores regardless of cooler, and whether or not PBO is enabled. In my 3900X, when I do a single core Cinebench R20 I get the same score with and without PBO enable. I'm not approaching any temperature limit (single core load is <60C) and I haven't maxed out PPT, TDC, or EDC. So why doens't it just keep boosting? Because the voltage limit kicks in, and that is the limit you hit first before any of the others. "The Stilt" discusses that at length as well.
So since you are voltage limited already (to protect the silicon from crazy heat density), even with the stock cooler, adding additional cooling capacity does nothing. Also worth noting that all that additional VRM overhead in the new X570 boards is also entirely useless. The voltage limit is hit way before that extra current capacity would make any difference.
You do get "some" added benefit it all core loads. I think s_vayner managed a 7300 multicore score on the 3900X (Noctua DH-15) while mine (custom loop) were in the 7500 range. And he did see a bigger uptick in multicore going to the Noctua from the stock cooler, but single core (and likely gaming) was exactly the same in all scenarios. With these CPUs, and likely CPUs in the future as dies shrink more a high end air cooler should really maximized performance, so it is a little strange that in this case a 65W CPU is running so warm.
Turn PBO and Ryzen Master still report 88W and not 65W. Temp doesn't help at all. I tried reseat the cooler and apply new MX4 thermal compound and no work. Here is a picture watching Youtube video with two other tabs open in FF browser with PBO turn off
I got it. And I think you're right. OP should definitely do something. Also 1.5V is normal I guess. 
Picture taken on the last thermal compound replacement 3 days ago. You see I install the mounting mechanism bar thing correct. Spacers in first then the two short bars then the screws. This is side way mounting position not upside down mounting position which require the long bars. I tighten them till them screws stop and feel the pressure that it is tighten.
Me in this video drop the cpu cooler onto the cpu with no paste and line up the screws of the cooler and mounting mechanism to check if this is the mounting mechanism flaw. Screws are not tighten in this video. If the screws are tighten then there will no wobble or wiggle around. I don't know if this is cooler fault?
This is from the ASUS Strix X470-F motherboard forum.
Sounds pretty much identical to your experience. Not sure why the temps with this board are so high. I should qualify, the lower results I posted earlier for my 3900X under custom loop are on an ASUS Crosshair VII Hero (Wi-fi) using UEFI 2801.
