Windows 10 Power Plan Bug for 5950x?


I’m not sure if this is by design or a bug, but when I’m running my 5950x in Windows 10 with the “Balanced” power plan, despite having a 420mm Arctic Liquid Freezer II with fan speeds around 1000-1200RPM+ in a ~25C ambient room, I’m averaging 60C+ when idling and 80C-90C under synthetic full load. It appears the voltages stay around 1.45v - 1.5v and the clock rate stays around 4GHz-4.5GHz, which seems rather high for idling (seems more like “High Performance” than a “Balanced” plan).

Meanwhile, if I drop the power plan to the “Power Saver” option, my temps become more normalized with 30-35C idles and 60C+ under synthetic full load; unfortunately, the clock rate drops to a min ~2GHz and a max of ~3GHz (0.9v - 1.2v).

Things got a little stranger… I ran prime 95 on small FFTs and the clock rate and voltage went DOWN when the CPU utilization went UP.

I also ran Cinebench R23 and sure enough… the frequency and voltage went down as utilization went up. Check out the CPU Core Voltages…

Before running Cinebench and idling at the desktop (1.487v):

and while running Cinebench (1.038v):

It’s almost as if the Balanced plan is somehow inverted?

I dropped the maximum processor speed to 99% in the Balanced power plan and now it won’t boost above 3.30GHz (or just below the 3.4GHz base), but at least, while idling, my temps hover around ~35C and voltages around ~1.1v.

It also appears when waking from the sleep, the power plan reverts back to Balanced despite being set to Power Saver. I have to set it back to Balanced, wait 30 seconds, then reset it back to Power Saver for the CPU to downclock from 4GHz-4.5GHz/1.45v-1.5v.

Thinking that this is a Windows only issue, I switched over to Linux Mint 20 and I got what I expected to see with the CPU idling around 2.2GHz and boosting to ~4.5GHz during a short burst synthetic full load.

Is anyone getting the same results with their 5950x with an ASUS x570 motherboard?


Motherboard: Asus x570 Hero VII Wifi (BIOS Rev. 3003)
CPU: 5950x (stock)
CPU Cooler: Arctic Liquid Freezer II 420mm (Rev. 3)
GPU: EVGA 3090 FTW3 Hybrid (stock)
RAM: G.Skill Trident Z Neo 32 GB (4 x 8 GB) DDR4-3800 CL14 Memory (XMP enabled)
Boot Drive: Sabrent 1TB Rocket 4 Plus NVMe 4.0
Case: Fractal Meshify 2 XL

To be thorough, I’ve updated the BIOS on my ASUS x570 Hero VIII (tried running everything stock) to the latest version (3003), updated Windows (20H2), installed the Asus chipset drivers, installed the AMD x570 chipset drivers and I’ve tried reapplying thermal paste and reseating my AIO on the CPU socket multiple times… but alas, same high idle frequency/high vCore/high idle temps.

This is common on the x570 hero I have the same board and my 3950x has way higher idle but when utilization goes up the readings go down. It is not a bug. It happens to be that a program pings it and as it pings it raises the voltage because the cores are calling for more voltage. Atleast that is my understanding. idle with a monitoring program, the cpu has high voltages. When I run cinebench or ffmpeg that idle voltage goes down. @Adubs knows of this all to well. He and I discussed this issue once upon a time.

Welcome to AMDs messed up agesa and ASUS not doing their due diligence. This board is straight garbage thanks to these two being so incompetent. My suggestion is to disable PBO entirely. The chip will still boost without it.

@Adubs Hmmm, disabling PBO didn’t seem to do a whole lot (other than drop the idle frequency)…

Are you running a negative offset vcore by chance?

Thats kinda the point, reduce idle so you can get some boosting due to cooler not being completely saturated. I never ran any offset because this almost always resulted in performance loss. My setup suffered from the exact same issues you’re facing now but I was on a full custom loop and using a 3900x. This board has an extremely immature bios coupled with AMD constantly changing things in agesa.

I’m currently using a 5800x and its pretty manageable. I could see how doubling of cores would really cause me some heart ache.

The other thing you could look at is your SOC voltage to make sure thats also in line.

The problem is this board is agressive off idle and by off idle I mean just a few %CPU usage is off idle. This is basically impossible to eliminate in windows so its constantly scaling core 0 up and down. Thus the insane idle voltages and temps.

Your cooler does seem to be a bit inadequate though if I’m being honest. I was on a full custom loop obviously but I’d expect the arctic coolers to be doing better here since they are chart toppers for AIOs. I’m currently on the EK240 and manage to keep my 5800x sub 75C while mining monero.

I can maintain 4.4ghz all core with this setup in a meshify 2.

Granted I only have the one CCX

For good measure I exported my current power plan settings. These were some kind of special sauce AMD recommended with their chipset drivers a while back. This is what I’m currently using. Don’t think it will do anything for you but I suppose it doesn’t hurt to try.

Ryzen™ (1.2 KB)

I’m curious if this is just an ASUS problem or an x570 platform problem?

If this is just an ASUS problem, then it’d make more sense for me to return it and swap it out for another x570 board.

At least I know I’m not the only one experiencing the same issue. That said, I’m a bit shocked that a 420mm AIO isn’t adequate enough to cool the CPU at stock. I’d prefer not to have to spend even more money for a custom water cooling unit (already spent over 4k for the current setup)! That said, the biggest issue of why this cooler is failing is because the CPU is being pinned between 1.475v and 1.5v during idle (essentially boiling the water before a load is even applied). If the CPU downclocked/upclocked properly, the temperature/vCore really shouldn’t be an issue.


The truth is its both. The grass may be slightly greener with an aorus master, but the real concern to me isnt your idle temps, its your peak temps. The ASUS may be a little high but if you cant manage to keep this chip from bouncing off 90C with this board, I have doubts the gigabyte will change anything.

Frankly this happened with the release of ryzen, and then again with zen 2, and here we are again dealing with it on zen 3. It will take time to iron out some of the quirks, as AMD changes and breaks things. As much shit as people give intel for skylake++++++++++++ and the 14nm meme, theres something to be said about the stability of a tried and tested platform.

Something you may not quite understand about ryzen is when its loaded lightly it scales voltage up and this is normal. The current is the real killer, not the voltage. It doesnt behave like any of the chips that came before it and is counter intuitive. your 1.5 is high but not insane. I’m sitting at 1.44 right now cooking away at alt coin mining.

I wonder if that cooler is just too much rad for the pump its got. Too much pressure drop and therefore low flow. I’ve seen tests on some corsair stuff that showed the 360 performing worse than the 240. Not sure what to make of that. I’m on half the rad here myself at 240 and things are great temp-wise.

1 Like

Good to know that it’s an x570 platform issue, so I’ll just play the waiting game for now.

Tomorrow I think I’ll play with the offset voltages and see if I can at least lower the idle temps a bit. The only time I’ve been able to hit 90C is running prime 95 (blended or smallest FFTs, can’t remember). With gaming, it only seems to briefly spike to 80C, then dip to 65C-70C. That said, I think I need to do more testing to determine if a cooler swap is necessary for long-term (the temps are going to be much worse when its in the summertime).

Thanks for the help!

1 Like

you could try a .05v negative offset and that would put you somewhere around where I’m at. In my limited experience though this almost always resulted in no noticeable change in temps or a loss in performance.

1 Like

Cooling doesn’t matter if you are running small fft it will bost till it hits 90 or the power limit.
What do you have open when “idling”. Doing petty much anything will cause it to jump up temporarily when a few cores boost Becasue it reports the hot spot temp.

Combine that with some monitoring programs refreshing rather slowly what it shows as 50-60 idle is probably it jumping between 30s and 60s depending on what its doing.

I can confirm that Gigabyte part: I have a 5950 on an Aorus Master and it schow exactly the same behaviour as in your Asus board.

Especially smallest FFTs in prime 95 will run it at 90 pretty much from the start.

I’d have to reboot into Windows to provide exact numbers, but I think it was way better when running other stuff, e.g. Furmark and the like.

From my understanding high idle voltage is a non issue so long as you have low idle current. Voltage with cores pinned @ 100% should be your main concern.

Huge update… Epic Games Launcher was causing my high idle temperatures. Killing the task, my idle temps went from around high 40s-low 50s all the way down to mid 30s!

With EGL running:


Welp, seems like everyone blaming AMD in this thread was perhaps wrong?
We seriously need to find a new consensus to troubleshooting AMD stuff. It seems to me that it simply isn’t the case that AMD has messed up on the hardware or software every single time. In the past few months I’ve found cases where people were claiming gpu driver issues find out that they were using shitty hdmi cables, or people not having good enough PSUs etc. I guess I’m saying that nothing should be “overlooked” in troubleshooting these days.


I don’t have epic games and it still has this problem. Crosshair viii hero wi fi, 5950x with 65c idle temps, kraken x73 aio. I still think there’s a problem with the chipsets and bios.

I’d recommend killing all non-essential Windows 10 tasks. As mentioned above, any sort of small load (like RGB software, monitoring software, etc) on the CPU causes it idle higher than normal.

Optionally, you can run Linux and run Conky and see if you’re getting the same temperature readings. If you are, then it’s your cooling solution/air flow.

What I did to improve my thermals even more was to switch to Thermal Grizzly compound (from Artic Silver 5) and instead of the radiator fans exhausting case air, they were moved to the top of the radiator to push air into the case through the radiator.

Except they have and it is their fault mostly.

Ironically even when windows says it’s ‘idle’ it’s really not very idle. I monitor windows vms from my unraid server and it’s constantly hitting cores for a small amount while in the task manager it shows effective 0%. Adding any little bit of overhead just compounds this.

Completely agree. Nevertheless, the small loads shouldn’t be causing idle temps in the 55-65C range.


Power Saver: