This is 2021 R1. Note there is a bug # associated with this if you are interested out to development. I am forced at the moment to run basically with less than 4-6 cores, after this number the problem immediately shows up again. Here's an example of the joint area iterations which receive the irradiation beam and should correspondingly be heating up:
Working case (~<4-6 cores):
Max temps start right around ~100 C

Parallel case (24 cores):
Max temps ~initialization temps (26.85 C)