Help debugging source of non-thermal BD PROCHOT CPU throttling
Recently, performance under sustained CPU intensive loads have resulted in CPU throttling to below base clock: 800Mhz, while all thermal sensors reported in HWinfo are in the mid 60s. The throttle begins after 30-40 seconds of sustained workloads, or during the multicore portion of synthetic benchmarks (eg geekbench5), while temps remain in the mid-high 60s, with fans forced at max. The throttle signal remains for 1-2 minutes following idle cooldown, then stops.
BD PROCHOT (Bi-directional Processor Hot) signal triggers the throttle, shown by Throttlestop 8.70 (no undervolt/altered settings).
BD PROCHOT is described having multiple input sources, including non thermal ones. While there are reported issues with VRM (Voltage Regulator Module) high temperature throttling in more recent (2019) MBP models, the 2017 MBP CPU VRMs do not have a thermal interface (that I have found). In addition, the low overall thermal data suggest another culprit.
Disabling BD_PROCHOT via Throttlestop solves the issue, however I am concerned about damage resulting from ignoring the signal source.
Does anyone have a suggestion about how to trace or debug the BD PROCHOT signal source? Happy to provide logfile/data.
What could the culprit be?