[SOLVED] Individual Core Temps

[Archaic59]

Just curious why core #2 runs hotter than the rest. It's obviously not a problem and my temps are great at idle and under load, but I get the same results with Core Temp, Hwinfo and OCCT. It's been this way since I built the system more than 7 years ago and I've used 3 different coolers during its lifetime with the same results. Currently using a Cryorig H7. As I said I don't have any problem with my system, just curious why core #2 is consistently 10c hotter than the rest at idle. The only thing I can think of is imperfections on the IHS but I defer to the experts here.
My Build

 

[haseeb98ahm]

Usually one core runs hotter than others because it turbos higher but not 10c hotter. Ivy bridge benefits from delidding because intel used crappy thermal interface material on those CPUs. So it's most likely poor contact between CPU die and IHS.

 

[geofelt]

A variation is normal.
Some of the cores are more efficient users of voltage.
Also, where the cores are located under the cooling lid can make a difference.
Not anything to worry about.

 

[Archaic59]

Usually one core runs hotter than others because it turbos higher but not 10c hotter. Ivy bridge benefits from delidding because intel used crappy thermal interface material on those CPUs. So it's most likely poor contact between CPU die and IHS.

Thank you for replying. I probably should've known about the TIM for as long as I've been doing this, but I didn't. It would be nice to assume Intel manufacturing practices and methodologies were above these kinds of issues but I guess that wouldn't be very realistic.

 

[Archaic59]

A variation is normal.
Some of the cores are more efficient users of voltage.
Also, where the cores are located under the cooling lid can make a difference.
Not anything to worry about.

Thank you for replying. Can you elaborate on core efficiency and voltage?

 

[geofelt]

All chips from a wafer do not perform exactly the same.
There are small variations.
Chips are tested(or binned) and the better chips are used in K suffix versions that are overclocking friendly.
When overclocking or raising the stock multiplier, a voltage increase is required to maintain stability.
This increase in voltage is what heats up the processor.
If a given core requires less voltage at a given multiplier, it will run cooler.
The converse is true.
So within a processor itself, the cores can behave differently.

The indivicdual cores within the processor chip are located in different physical parts .
Those closest to the cooling heat spreader will get better cooling and may run cooler.
With K suffix processors, Intel only guarantees that it will run at advertised stock speeds.
But, the luck of the bin lottery will allow the chip to run at higher clocks if thermal requirements are met and if the voltage does not exceed limits.
Since you have a Z77 motherboard, and a good cooler.you will be able to overclock some.

Try raising the all core multiplier a bit at a time and stress test.
Monitor the vcore and keep it under 1.3v for safety and keep the stress test temperature under 85c.

 

[CompuTronix]

... core #2 runs hotter ... temps are great at idle and under load ... core #2 is consistently 10c hotter than the rest at idle ...

Archaic59,

Each Core has its own individual Digital Thermal Sensor (DTS) which functions independently from the other Cores. Intel's specification for DTS accuracy is + / - 5°C across the entire operating range. This means that at steady-state 100% TDP workload, the deviation between the highest and lowest Core temperatures should not exceed 10°C, so you're at the upper end of the specification.

Intel had a known problem with DTS sensors in 45 nanometer Core 2 processors where some Cores wouldn't idle as low as others. Although typically accurate, DTS calibration issues such as linearity, slope and range are always a possibility.

As geofelt suggested, thermal dissipation is affected as different processor microarchitectures place the Cores in different locations within the Die relative to other elements, such as the Integrated Graphics Processor Unit (IGPU). Also, inner Cores tend to run slightly warmer under load as they're insulated by the outer Cores. Moreover, deviations in Core temperatures tend to converge at idle, but diverge as load increases.

Concerning your question regarding excessive deviations between individual Core temperatures, there are 3 possibilities:

(1) If your deviation diverges at idle but converges as load increases, then Core #2 has a nonlinear DTS consistent with 45 nanometer Core 2 processors described above.

(2) If your deviation is consistent across the entire operating range, then Core #2 has an "offset" that was miscalibrated into its DTS sensor at the factory.

(3) If your deviation converges at idle but diverges excessively as load increases, then Intel's factory application of Thermal Interface Material (TIM) between the Die and the Integrated Heat Spreader (IHS) has a flaw or "void" above Core #2. However, this doesn't seem to be the case in your instance.

See Section 9 - The TIM Problem

Intel Temperature Guide

CT

 

[Archaic59]

Outstanding explanations guys. After stress testing the CPU at 100% on all cores @ 3.8Ghz. for 5 minutes the max temps on all cores were as follows-
core 0 - 59c
core 1 - 66c
core 2 - 66c
core 3 - 58c

 

[CompuTronix]

No problems here; 8°C deviation between the highest and lowest Cores is within the + / - 10°C spec.