Safety limit not respected during auto tune.

[dust0vich]

Bug Report

System Info

• CPU: Ryzen 7700x Raphael + Thermalright AXP90 X47 Full Copper

• Zen generation: Zen 4

• Motherboard: Gigabyte Aorus B650I Ultra

• BIOS version: F42

• Kernel: 7.0.11-arch1-1

• ryzen_smu version: 4.84.4.0

• CoreCycler version: commit f0a6bc8 (HEAD -> main, origin/main, origin/HEAD)

• Install method: Arch method, install all modules and then 'sudo python src/main.py'

Description

What happened?
Auto tuner does not respect max core temp in config tab during test.

Expected Behavior

What should have happened?
Should have respected temp value so cores don't give false negative for brief temp spike. Or at the minimum allow for a brief temp spike before calming down, or allowing the coarse to repeat if it fails at same setting.

Steps to Reproduce

1. Adjust safety temp higher than 95c
2. Press play in auto tuner tab
3. Watch cores randomly fail due to temp spike after 2 seconds of test.

Relevant Output

Terminal output / error messages

CPU temperature 100.5 C exceeds safety limit 95.0 C — stopping test
CPU temperature 99.1 C exceeds safety limit 95.0 C — stopping test
CPU temperature 100.6 C exceeds safety limit 95.0 C — stopping test
CPU temperature 97.4 C exceeds safety limit 95.0 C — stopping test

For now what I have done is turned down the max temp to 92-93 in the PBO BIOS settings to mitigate the issue even though it still happens intermittently. My cooler doesn't have much headroom and am trying to drop the temps as much as possible for my minipc build. Having to do this makes me have to redo the tuning over and over as it stops testing on certain cores when the temp spikes during the coarse phase.

Your program is a lifesaver and literally the only one that I have seen that supports 7xxx cpus. Thanks for the awesome work.

Additional Context

• Runs as root
• ryzen_smu module loaded
• Issue occurs consistently (not intermittent)