Please do you know where I can find it on msi b450 tomahawk with newest bios? I checked bios, manual too but i cant find it 
Also, AIDA stress measures the CPU temperature and not the CPU diode temp which is at least 10°C higher!
I am talking always about the diode temp.
See the difference?
AFAIK this difference depends on motherboard/BIOS. I and my colleague have almost the same data about power consumption(Gigabyte X570 Aorus Elite F10a) - limit is 90 W according to different software. It is strange to have almost 170W without changing power limit. In my case AIDA heats CPU more than prime95 as I wrote before.
I got you now, you have disabled PBO and power consumption is on 90watts limit.
And was wondering about what was this 90watts limit you were talking about...
I did the same and results are identical to yours and temps even better, and indeed now AIDA stress CPU harder than prime95, but this is because with prime95 when PBO is disabled, frequencies stay low around 3600.
So, you are working your CPU in shallow waters, because like this performance is worse, an example: CINEBENCH20 gives me 5000 with PBO enabled and around 4650 with PBO disabled.
Your point about heat big density due to very small die makes sense, this explains why cooler remains cool all the time, of course, its physics but this isn't something good, in fact, is something very very bad.
Honestly, if it really reaches 90C, Where all this heat goes? it goes to cooler but very very slowly, it is a problem!
Tomorrow I will try to measure the temperature at the side of the IHS with a laser thermometer when the sensor shows 90, I bet it won't be more than 30-40
I can see underneath of cooler plate because my box's top is removable.
putting my finger down there I can feel no heat at all, it is like dead!
Still, bet that sensors give false readings.
See photo.
No point for a screenshot, all software including Ryzen Master shows the exact same temps.
If the only way to run this chips cool is by limiting consumption at the expense of worse performance, then this zen2 is not meeting the expectations influenced by the exaggerated reviews before zen 2 platform release, remember they were talking about overclocks near 5GHz and other unreal things.
Liquid metal doesn't corrode any processor, after many years of use I never saw the slightest corrosion in copper or nickel, also it is very easy to remove it without leaving any trace on IHS, everything they say about removed letters and scratches is a myth driven by competition or scared novices who never attempted to give it a try, I never sanded the surfaces prior to applying because there is no need for it.
Also, I didn't have a 20C different, settings like boost and consumption were slightly different and I was exaggerated a bit as most people in forums do, my apologies for that, there was an obvious difference though!
I found a screenshot from 2 days ago when liquid metal was used, and I took another shot today using the paste provided by Dark Rock Pro 4 which I applied last night, see both and compare, I must admit that there is no difference between liquid metal and Dark Rock Pro 4 paste, which is proven to be much better than MX-4.
AIDA Stability test, with PBO Enabled and liquid metal. (Don't mint about the max, there were higher because I had stressed with prime95 before AIDA)
AIDA Stability test, with PBO Enabled and Dark Rock Pro 4 thermal paste.
Conclusions are easy, liquid metal doesn't worth the hassle, although I enjoyed the time spent.
Another thing is, this unnamed thermal paste provided by BeQuiet is really decent, I will ask them where can be found.
UPDATE
Today I tested Arctic MX-2, no difference at all, absolutely the same, finally it's not worth the time with these thermal pastes, there all about the same.
With MX-2 AIDA Stress(even better than LM!)
you´ve got to remember that a stress test is nothing like everyday use.If you´re not close to 90 c you´re golden. Everyday use doesn´t strain your cpu like a stress test,it´s kinda in the word.... 
