Even though these is no load in the UEFI/BIOS, this will cause a higher temperature because of the voltage being at what is required for this high P-state.

That explains why UEFI/BIOS is a higher temp than at idle in the OS, as the OS will invoke lower P-states and Park Cores, dropping voltage and heat.  When there is work to be done, the OS will tell the CPU and so things wake up, request more volts and pull more current under various load situations.

It's also true that there is a +20°C temperature offset on the -X Ryzen CPUs.  This is something to do with TDP and some other stuff I can't remember.  It's not just AMD, this 'trick' was on the Intel Core2Duo and numerous other CPU's.  You simply need to factor in this offset when configuring fan curves or reading temperatures.

ASRock also read the T-socket temperature by default I think.  This is the temperature of a sensor in the CPU socket (?) and it takes longer to heat up and cool down than the actual CPU die or heat spreader.  It's one way to make fans ramp up/down a lot less, but it's also a way to let the CPU get a lot hotter before doing something about it.

Newer UEFI/BIOS versions typically allow for selecting a different source for temperature display in the HW Monitor UEFI/BIOS tab.

If you want to measure temperatures in Windows use HWinfo, as it displays Tctrl (value reported from on-chip sensor), Tdie (Tctrl adjusted -20°C, thus actual die temperature) and the temp from the motherboard Nuvoton sensor, or Tsocket as we've come to call it.