trulyspinach / SMCAMDProcessor

Power management, monitoring and VirtualSMC plugin for AMD processors
BSD 3-Clause "New" or "Revised" License
1.04k stars 90 forks source link

CPU frequency stays at max level #125

Closed wtfloris closed 3 years ago

wtfloris commented 3 years ago

Hi all,

I'm not sure if there's actually something wrong or if I just don't understand it, but it seems to be the CPU freq should automatically be dialed down when there's little workload, right? Mine doesn't seem to do that according to AMD Power Tool. Screenshots included, it stays at 4.2 GHz (with auto speed adjustment enabled of course). Even when I manually click a lower speed step, it stays at 4.2.

It's a Ryzen 5 3600 on a ASUS TUF X570 PLUS mobo.

Am I missing something here? I couldn't find any other mention of this issue online. Thanks

Screenshot 2021-02-13 at 21 17 49 Screenshot 2021-02-13 at 21 17 57

wtfloris commented 3 years ago

Asked for support in the AMD OS X Discord server, just wanted to share what trulyspinach told me in case someone runs into the same problem:

[3:36 PM] trulyspinach: i remembered some have problem setting speed when they have undervolt enabled in bios [3:37 PM] trulyspinach: did you enabled any such options as well? It seems that bios will constantly use smu to override cpu settings thus the changes in OS will not be effective [3:38 PM] Shaneee: Yeah. I've only got PBO enabled and with Chrome, Discord and background apps running it'll settle down to 2.9GHz when I leave it [3:40 PM] WTFloris: Ah that makes sense, it's overclocked and undervolted through bios yea [3:45 PM] WTFloris: Hmmm I've set cpu voltage to "auto" but that doesn't seem to change anything [3:50 PM] trulyspinach: there must be other settings overriding it

Will update if I find out what the issue is.

wtfloris commented 3 years ago

The issue was a manual overclock through the BIOS. Undervolting works fine, as long as you don't manually set a clock speed/multiplier.

Now using PBO with some manual undervolting and it works fine.