i7 4790K Overclocking - One core much hotter than the others

Monitor your voltage with something like CPUz on your second monitor to see where it lands when under heavy load. You can try to reapply the paste, couldn't hurt. Temps are far too high though, you may need to back off on the voltage.

Do you mean the side panel fan? That is mainly for providing additional cooling for the graphics card.

Is the thermal paste evenly applied across the CPU, it could be that you have bubbles in the paste or uneven distribution, that could result in a really odd temperature reading.

This is what I did with my 4690K? I put a small grain rice size dab on the CPU lid. I then took the thin plastic clam-shell my CPU came in and used its thin edge to spread it evenly over the entire surface resulting in an ultrathin coating. I then repeated that step on the base of my 212 Evo. I then installed it.

Is it possible you are simply overclocking the one core? I know my Asus board allows this.

If temps are changing after mounting/remounting the cooler it sounds like there might be something going on with the mounting process. 20c is pretty significant between cores under load.

What are temps at stock? Have you tried clearing the bios completely and running at default settings and comparing the temps you get? That's where I would start, clear out any of those offsets and overrides. 4.3ghz isn't really much of an overclock, the 4790k will reach 4.4ghz on a single core under full load, 4.2ghz when all 4 cores are under full load. See if the temperature variation is still present at stock settings between the cores and work on getting those closer to one another.

Make sure when mounting the cooler to tighten the screws in an "X" pattern, for example top left, bottom right, top right, bottom left and turn them just a couple turns at a time working your way around back and forth until they're all tight. That will help ensure the most flat mounting of the cooler to the cpu. Make sure nothing else is getting in the way like ram or vrm heatsinks, anything that might be 'nudging' the cooler as it's tightened down.

What are the ambient room temps? If it's 20c in the room or 30c in the room that will impact overall cooling. Here's an article with various thermal paste application techniques and some results. I usually use the 'X' pattern (without overlapping the lines to avoid a heavy glob in the center). Haven't found it necessary to spread the paste out and it shouldn't need applied to both the cpu and the base of the cooler. Greater chances of using too much. Both surfaces cleaned (ihs on the cpu and the base of the cooler) with some isopropyl rubbing alcohol and a lint free cloth, the alcohol evaporates pretty nicely leaving the surfaces clean/dry.

https://www.pugetsystems.com/labs/articles/Thermal-Paste-Application-Techniques-170/

On a side note, you may want to consider a larger power supply for that system. 450w to run a 1080ti card and overclock the cpu is asking a bit much from it. Nvidia recommends 600w for a system running the 1080ti, even if you take it with a grain of salt and decided to go with a 550w it's still a noticeable amount more than 450w.

Just put a tiny blob in the middle of the cpu and let the weight of the cooler spread the paste, don't spread it yourself and don't lift it off to check as you will make air bubbles

Have you tried another program to test like prime95? It may be a bit exhausting trying different things but it will help eliminate variables to determine if it's the cpu itself. If you try p95 look for version 26.6 rather than the newest and run small fft's for a steady state load for temp monitoring. Running blend will be a pain to keep an eye on temps since it's a cyclic load which will fluctuate up and down. Realbench is a decent test but more suited to stability rather than thermal testing.

You said you set everything to stock, did you reset the bios using the jumper on the mobo or by removing the cmos battery? Just double checking everything is reset to defaults.

What program are you using to read temps with? Hwinfo64 is a solid program, realtemp also does a good job.

These are only some possibilities, another could be the cpu itself. Unlike sandy bridge cpu's the ihs (metal cover of the cpu where you apply the thermal compound) isn't soldered to the top of the actual cores. There's a bit of thermal compound placed on top of the cores that gets sandwiched between the cores and the ihs on the inside. It's possible that during manufacture the thermal compound used was applied a bit heavily and just doesn't allow for optimal heat transfer on that core like with the others.

Did you buy the cpu new or used? If you bought it new and you're still within the warranty period (3yrs) you could try contacting intel and see what they have to say about it. If they think it's enough variation to warrant a replacement. They may or may not. If it's a used cpu, no dice and could be why someone was getting rid of it.

It could be a flaw in the cooler, maybe the base isn't machined flat or something is causing it to mount with uneven pressure. The only way to really confirm that is to try using other parts, a different cooler, swap out for a different cpu with the same cooler and settings in the bios, see if the results are similar or more normal.

I suspect a faulty sensor.

You can try contacting the motherboard manufacturer, however, they often do not reply or blame your problem on something else [improperly mounted cooler, for example]. I would link this page to them so they can see what you have done and advice given you. - - Good luck with motherboard support from the manufacturer [you might get lucky].

The ability to externally checking readings depends upon the motherboard, some have external locations where you can make external readings, however, most do not.

- - - - - - - -

Further comments:

I started working in the computer industry over 30 years ago and have been overclocking for about 25 years. In general, with overclocked Intel CPUs for the past 7 years it is not uncommon to see a 13 degree temperature spread, you noted a 20 to 25 degree C spread, and up to 100 degrees C.

If your temperature sensor is correct, danger comes into play when temperatures are too high, and I consider 80 degrees C and above too high.

NOTES: Some benchmarks and games stress a single core and that will cause it to increase temperatures, although an extreme spread such as yours should not be happening unless something is seriously wrong with your cooling, such as improper TIM application [which you implied in your initial post] or improperly mounted cooler. Your cooler, by itself, is a fairly good one and should not be the source of your problem.

To make this post shorter, here is a link to one of my TIM suggestions to another person http://www.tomshardware.com/answers/id-3332372/miss-spot-thermal-paste-matter.html?siteFrom=EPR-8807#xtor=EPR-8807

Of great concern to me is you are getting high temperatures with no overclock. Something is either seriously wrong with cooling or, as I mentioned you have a bad sensor. Run your CPU fan at 100% [some motherboards let you do this in BIOS, some have software that can be installed to your operating system, and sometimes you simply have to connect it to a 12 volt power connector. If needed places like Micro Center or Newegg sell adapters to change plug connection types, if necessary.]

I use a program that looks at many sensors. https://www.hwinfo.com/download.php After downloading that product, use something like OCCT to stress your CPU. http://www.ocbase.com/index.php/download

Check for updated motherboard BIOS and look a what each newer BIOS version fixes or addresses [it should be at your motherboard's web site], just make sure your power input is stable and you do not turn off your computer while the BIOS is being updated.

Again, please run your fan at 100% and let us know what your temperatures are - also, what your room temperature is.

Hi xLiqxo, you're already getting good help here so I'd like to just add a little bit: It's possible the IHS "lid" of your chip is not optimally seated over the die. At idle, your temps look fine because IHS is able handle the low thermals. But at full load, any insufficient contact becomes amplified and manifests as an uneven thermal gradient. While a lower temp on the 4th core is expected, the 37 degree gradient difference between inner and outer core is substantially out of norm.

I suggest requesting RMA if possible, otherwise a de-lidding procedure may be necessary to properly re-seat the IHS.