Thanks for those epic screenshots
For just a bit of peace of mind, despite that crazy high cpu diode read, how much is the socket diode reading? If you let it stabilize, cpu diode reads shouldn't be higher for more than ~5-10ºC in old K8s. Even tough some funny stuff to that interval might happen if those temps are accurate, you never know.
These cpus changed a lot with time since the first clawhammers/newcastles, it might be that your bios is simply misreading the CPU diode. I'm giving you a much older example, in the 754 days the older C0 CPUs reported different temps when compared to later CGs, and bios were either set for one or the other (this is easy to check on a dfi 250gb, from stock to later bios). Given the experience with 3 cpus, I could confirm differences up to 10ºC (cpu diode) for the same socket temp reading (stable); this was all before 90nm stuff, with just a slight architecture change.
939s had way more variants, and a greater chance for things to go wrong or updated on what wasn't a top board. An example is the DFI NF4 Infinity, it would report crazy low diode temps with 90nm stuff - a friend of mine had his 3800+ x2 reporting about 15ºC lower than socket diode.