CPU not properly throttling in Ubuntu 23.10

jferments · 04-21-2024, 10:39 PM

My CPU (AMD 7965WX) is not properly thermally throttling, and is exceeding the max temperature. I am running Ubuntu 23.10, and using an ASUS WRX90E-SAGE motherboard.

If I run CPU intensive processes that max out all cores (especially if my dual GPUs are running heavy loads at the same time), I can quickly exceed the 95C maximum temp for my CPU, and Ubuntu does nothing to throttle the CPU. I'm worried that my CPU is going to get damaged.

When I look at

Code:

/sys/devices/system/cpu/cpu0/cpufreq/scaling_available_governors

there are only "powersave" and "performance" available ("powersave" is what it is currently set to). It is my understanding that neither of these will throttle the CPU.

Also, when I try to set to "performance" governor in /etc/init.d/cpufrequtils and reboot, it has no effect. It just stays in "powersave" mode

When I check status of thermald, this is what it says:

Code:

username@ML-tower:~$ systemctl status --lines=50 thermald

○ thermald.service - Thermal Daemon Service
     Loaded: loaded (/lib/systemd/system/thermald.service; enabled; preset: enabled)
     Active: inactive (dead) since Mon 2024-04-22 14:34:09 PDT; 26min ago
    Process: 1878 ExecStart=/usr/sbin/thermald --systemd --dbus-enable --adaptive (code=exited, status=0/SUCCESS)
   Main PID: 1878 (code=exited, status=0/SUCCESS)
        CPU: 11ms

Apr 22 14:34:09 ML-tower systemd[1]: Starting thermald.service - Thermal Daemon Service...
Apr 22 14:34:09 ML-tower thermald[1878]: Unsupported cpu model or platform
Apr 22 14:34:09 ML-tower systemd[1]: thermald.service: Deactivated successfully.
Apr 22 14:34:09 ML-tower systemd[1]: Started thermald.service - Thermal Daemon Service.

Just to reiterate: this CPU (AMD 7965WX) is definitely compatible with my motherboard (Asus WRX90E-SAGE), so I am not using an incompatible CPU.

How can I ensure that my CPU will not exceed a specified temperature threshold (ideally ~92C or less)? Is there any way to also throttle the GPUs when CPU temps are too high?

business_kid · 04-24-2024, 04:28 AM

It sounds like your heating is out of control. On a Ryzen, Temperature should be managed by using a semiconductor junction on the actual wafer as a temperature controller. It should reduce or increase core voltage, & frequency to keep the temperatures within safe boundaries.

I gather you're into gaming. Now if the cpu goes up to 95° and stays there, your cpu is working as designed, but your heating is inadequate. So your cpu will be throttling back, and all talk of governors is meaningless. I have a gaming type PC, but I'm no gamer - it's something future proof that I don't intend to replace, just to repair. I don't have puny IGPUs, but a separate card. I have proper cooling and rarely see 40°.

Investigate a bit more, and see what you come back with.