I had done this, I did a lot of things, the only thing that has worked so far is to decrease the memory clock, mine is like 2667mhz. I've been using it for 15 days and so far it's normal, it still hasn't given me the problem.
I was using version 1.something from November 2019. Updated to latest 2.04.04.111. Will monitor if this issue presents again.
How this can be connected with CPU multiplier being locked on the lowest value?
I have Gskill 2x16Gb with 3200Mhz XMP profile. Are there any known issues with my configuration? (3700X and B450 Tomahawk Max and Gskill)
Never got any issue with Windows power plans on any other system configuration. Even this problem happens like once every two months. Definitely not primarily Windows problem.
I had this same issue on my first install of W10, the OS kept switching the power plan on every single boot.
I did some reading on this and I found out that there must be some conflict, or Windows thinks there is one (regardless if there really is). When I removed 1usmus CPP this all seemed to stop for the most part and I haven't reinstalled it on my new OS install.
I am not sure I can recommend that CPP anymore, for the first 4 months I had the 3950X, I was 100% behind recommending it. However, I wonder if it was part of the reason that I had to re-install W10 after only 4 months. For now, I am just using Ryzen Balanced and Power Saver modes, not worrying about any CPPs.
If you are having a similar issue, where your power plan is randomly switching on you, finding it in PS mode or a non-Ryzen plan, then you have a conflict with the CPP too. I can't guarantee removing the 1usmus plan will stop it, as I did a full re-install shortly after that. It sounds like a good place to start though.
There is also a chance we missed some of the needed settings to use the 1usmus plan, although I triple checked all of mine. My ASUS X570-E STRIX is missing some of the requested options, though, so maybe that is why?
For some reason, swapping between power plans seems to refresh the CPU and the cores essentially, wake up.
I'm not sure what is going on that so many people are reporting this, but I think some of you have the behavior that AMD intended.
Only certain scenarios will you see every core running at max speed. When you do get all the cores running up to max speed, the speed they hit will be lower than if you had 4 cores hitting max speed. Usually a few hundred megahertz and I think this is what is happening. Perhaps for some of you, the software you are running, just isn't utilizing the full core/thread count.
By the way, you can force all cores to run at max speed, but if you do, they won't downclock at all. To do this, either set the clock to a manual speed (OC/UC) or disable Precision Boost 2.0 or PBO (whichever you had selected).
I did a decently deep dive into AMD and W10 power plans in this AMD Community Thread if anyone is curious about my results.
(AMD 3950X + ASUS X570-E + G.Skill TridentZ 3733)
I figured out what is causing all of this behavior. The power plans all make the CPU core voltage behave differently.
In Ryzen Balanced mode, my CPU + Mobo holds the core voltage steady at 1.38V. The Average voltage and the Peak voltage are equal.
In Ryzen HP mode the average CPU voltage is lower than Balanced about 1.18V, but the peak is higher, 1.435V. This means that the power is being delivered to the cores that need it most and less to the others. This could in theory lead to higher boost clocks, so it makes sense I guess.
Then in both of the 1usmus CPPs, the CPU voltage is about 1.41-1.42V steady, pretty much like the Ryzen Balanced mode, except it does vary a tiny amount
Note: Power Saver Plan puts it at 0.9V and holds it there.
You are right indeed that the "refreshing" of the CPU seems to wake things up but you're wrong about the software utilization portion of your comment.
While experiencing this bug, an Arnold vray or renderman render or even a simple stress test on the Ryzen master software produces a "sawtooth" graph not a flat/full utilization one, however high the Mhz count. The TDC and EDC flactuate erratically from ~20% and ~85% and this directly affects clocks.
By switching any power plan to any power plan, the cores seem to just... wake up. Sometimes even a couple of consecutive stress tests will do the trick. It happens quite often after a cold boot and it is an issue of stability and reliability.
I said it's possible that is why "some" people reporting it, might be experiencing that.
I wasn't saying that about you (or anyone) specifically. Just wanted to point it out as a possibility, since this can be the correct behavior, depending on what was happening during the screenshot.