Also, the internal thermal path for each core will differ if the layout on the
die is not perfectly symmetrical or if activity (and thus heating) is not
absolutely evenly distributed over the rest of the die.
I did wonder about system threads and have been watching task manager, but
the only processes running are taskmanager itself and speedfan. perhaps the
speedfan monitoring software causes the core0 to tick over. This it the
nature of chaos - by simply observing the CPU, I am affecting its
temperature and operation!
System processes have many threads.
Each can run at different times and different loadings - and usually on just
one of the cores.
At idle, those threads make up a lot of the activity.
The room was pretty warm yesterday but today the room is cold and the cores
read 40 and 36, so still a 4 degree difference. Even in a warm room, 50
degrees is not bad for silent air cooled system, so I won't worry about it.
Next time I'm in there, I'll perhaps reseat the cooler. I'm sure I have some
thermal paste kicking around from my last job.
Although your core sensors were intended to be identical, they can still have
minor differences that cause them to read differently.
Despite every effort, the die fabrication process is not perfect; in fact,
according to NewEgg customer comments, at least one very modern quad core is
arriving with some totally broken core temperature sensors.
So, I certainly wouldn't worry about any *small* differences between cores...