difference in reported core temps using EVGA E-LEET utility

[Citizen Snips]

I have been running prime95 for several hours on my new X58 SLI/core i7 920 build and have reached the following steady-state temperatures as reported by EVGA's E-LEET utility (it looks exactly like CPU-Z but also has a special tab for overclocking settings):

core 1: 58.0 C
core 2: 56.0 C
core 3: 58.0 C
core 4: 54.0 C

This is at stock speeds. Is the difference of 2-4 degrees C normal, given that all cores should be loaded equally by prime95? What differences in temperature between cores do the rest of you see under load? I see the same difference in temperatures between the cores at idle.

Is the difference in temperature due to:

1. inaccuracy of the temperature sensors
2. imperfectly applied TIM
3. imperfectly secured HS
4. imperfections in the HS construction
5. all of the above
6. something else

Also, are these sensors on the die of the CPU itself? If so, can someone tell me which core number in E-LEET corresponds to which physical core on the die?

E-LEET also has another temperature sensor labeled "CPU". Is this one located on the motherboard?

 

[silent-circuit]

That's not a big differential at all. Nothing to worry about. Ignore the "CPU" temp, it means very little. Core temp is all that matters.

 

[Zero82z]

I have no idea how E-LEET gets its temperature readings, but you should probably be using the latest version of either Real Temp or Core Temp to monitor your temperatures instead. And temperature variances between cores are perfectly normal and occur for a number of reasons, so you've got nothing to be concerned about.

 

[Citizen Snips]

I have no idea how E-LEET gets its temperature readings, but you should probably be using the latest version of either Real Temp or Core Temp to monitor your temperatures instead. And temperature variances between cores are perfectly normal and occur for a number of reasons, so you've got nothing to be concerned about.

Even when all cores are theoretically being loaded equally by prime95?

 

[Zero82z]

Even when all cores are theoretically being loaded equally by prime95?

Yes. Temperature variations are not caused only by asymmetrical loading. Like I said, there are several reasons, some of which are manufacturing variances between different transistors on the chip, the heatsink mounting not being perfectly even, background processes using more of some cores and less of others, thermal sensors not being calibrated properly, measurement inaccuracies, etc.