Thanks Achim.

As I expected, that's still the same situation as earlier DFI BIOSes, they need to sort it out because some load temps here are actually in the 70s at least - the temps are reading lower than the CPU registers are showing, by a few degrees at minimum. The diode [CPU in EVEREST] will always be lower than Core temps, if not, BIOS has a problem with giving incorrect temps. In your case, they're adding negative offsets to read lower than actual temp values, which is misleading for all and mainly for the temp controlled HSF.

I've not had a problem with Sandra in anything yet, many different builds now. Just finished the total benchmark suite with 4450e including Java. Probably a DFI thing.

Yea, you need to find fully stable settings first. Sandra will crash if it detcts instability very quickly, it has done for me for a while now.
Then you can maybe try disabling SMBus/DMI and other unnecessary options access in Hardware>Processors>Options and check if it makes a difference [should do] whenever you want, enough to retrieve Tctl values.