[SOLVED] Core 1 Temperature higher than others?

[spazbandicoot]

Hi there,

I have a few concerns and questions:

I'm currently doing a stress test in AIDA64 on my i5-4690k to find a stable 4.3GHz Overclock on all cores.

Nothing seems to be too out of the ordinary apart from the fact that I've noticed the package temperatures are spiking up in relation to Core #1's temperature.

Cores #2,#3 and #4 are running at around 60-70, while Core #1 is running at around 70-80.c peaking to almost 90.c every few seconds.

I should add that I re-mounted my CPU Cooler (be quiet! Dark Rock Pro 3) yesterday and re-applied some MX-2 thermal paste as I had done a bad job at both before. I don't remember if this was happening before I remounted the cooler or not, I never really realised it until now.

I'm just a bit curious as to why this is happening, if it's normal, if this is where my thermal headroom ends and/or if it is something that can be 'fixed'.
Also wondering as to why temperatures in AIDA64 spike up every 4-5 seconds.

Also, I'm a bit of a noob when it comes to overclocking - how long should I run this stress test to know if I've reached a stable Overclock? 15 Minutes? An Hour? 4 Hours??

Specs if needed:

CPU - i5-4690k

RAM - 24GB DDR3 HyperX Fury 1600MHz

GPU - GIGABYTE G1 GTX 1060 6GB rev.2

Mobo - ASRock Z97 Anniversary

PSU - Rosewill Capstone 1000w 'Gaming' (80 Plus Gold)

Any help is appreciated.
Cheers.

Temp Graph

Edit: Added Graph Photo

 

[Darkbreeze]

Variance of UP TO 10°C, CAN be normal, and is within the acceptable variance between cores according to the Intel specifications. That does not necessarily mean that it IS a normal condition however. It could certainly be due to a bad paste job or mount, where the coverage or mounting pressure are not equally distributed across the relevant areas of the heat spreader.

It can also be caused by an unequal internal thermal interface material application as well.

It would have been helpful to know if this existed prior to the repaste or not.

What sort of application technique are you using and how much paste are you putting down? How are you tightening the heatsink as a process? Are you tightening in increments, moving from one corner to the next, until full tighness is achieved or are you tightening down one side or corner all the way and then moving to the next side or corner?

 

[spazbandicoot]

Variance of UP TO 10°C, CAN be normal, and is within the acceptable variance between cores according to the Intel specifications. That does not necessarily mean that it IS a normal condition however. It could certainly be due to a bad paste job or mount, where the coverage or mounting pressure are not equally distributed across the relevant areas of the heat spreader.

It can also be caused by an unequal internal thermal interface material application as well.

It would have been helpful to know if this existed prior to the repaste or not.

What sort of application technique are you using and how much paste are you putting down? How are you tightening the heatsink as a process? Are you tightening in increments, moving from one corner to the next, until full tighness is achieved or are you tightening down one side or corner all the way and then moving to the next side or corner?

I mounted the Cooler accordingly to this YouTube video.

It would be very, very difficult for me to mount it according to the Manual as that was what I did last time when mounting and resulted in a bad job due to it's complexity and.. 'finickiness'.

I don't exactly remember if this is true or not but I think this high temperature Core thing may have been happening before the re-mount.

My thermal paste application was an estimated pea-sized dot in the middle of the CPU cooler's contact point (since I was pretty much mounting the Motherboard to the cooler, instead of vice versa). I tried not to move the motherboard and cooler around too much while sitting them together.

When tightening the screws, I did increments on each corner, moving from one to the other until what I felt was a decent amount of equal force when trying to tighten again on all the corners.
I see that in the YouTube video linked above, the person mounts the cooler with what seems to be tighter screwing.

Should I try tightening the screws more? When would I know when to stop?

Edit: Fixed Typos

 

[Darkbreeze]

If you can't easily continue to tighten using "wrist" strength, then it is tight enough. If you have to start using your arm and shoulder to get it tighter, then it's already as tight as it needs to be. In other words, if it's tight, it's tight. It does not need to be "torqued" down. So long as the hardware is equally tight all the way around and you can "feel" that it is tight, then it is.

It is impossible to explain through words what the feeling of "tight enough" is. You'll have to judge that for yourself but suffice to say that OVERTIGHTENING is also bad. It can cause all kinds of problems such as shorting the CPU in the socket or damaging the motherboard. So get it tight, and call it good.

Sometimes it takes a few attempts to get the mount just right, especially on a very big, very heavy cooler. As with anything, practice makes perfect.

 

[spazbandicoot]

If you can't easily continue to tighten using "wrist" strength, then it is tight enough. If you have to start using your arm and shoulder to get it tighter, then it's already as tight as it needs to be. In other words, if it's tight, it's tight. It does not need to be "torqued" down. So long as the hardware is equally tight all the way around and you can "feel" that it is tight, then it is.

It is impossible to explain through words what the feeling of "tight enough" is. You'll have to judge that for yourself but suffice to say that OVERTIGHTENING is also bad. It can cause all kinds of problems such as shorting the CPU in the socket or damaging the motherboard. So get it tight, and call it good.

Sometimes it takes a few attempts to get the mount just right, especially on a very big, very heavy cooler. As with anything, practice makes perfect.

I'll try tightening it down a bit more with my wrist only.

 

[spazbandicoot]

So, here's an Update.

I tightened the cooler on a bit more, the temperatures didn't change at all.
Then I loosened them a bit, same result.
I painstakingly re-mounted the CPU cooler and now the temperatures are higher - peaking to 81.c while playing Battlefield 1 and reaching 96.c in a stress test.
I can't put my finger on what I'm doing wrong - Thermal paste application was the same, made sure there would be no air bubbles when mounting, tightened each corner in increments, moving from one to the other with equal force, made sure none were over-tightened, completely cleaned out the cooler, dusted off the motherboard, completely cleaned the top of the CPU (checked for anything on the bottom too) and socket and set both fans on the cooler to spin full speed, case fans too.
The fan configuration is push/pull for the whole PC. Room temperature is pretty cool.

Turning down the overclock to 4.2GHz (1.2v) doesn't show any improvement - temperatures still hot at around 80.c while gaming and about 92.c in a stress test.
This 4.2GHz OC, I've used it before and the temperatures have never EVER been this high. Something's wrong but I don't know what it is.

This is really, really frustrating me. I have no idea what is going on.

Update Update:

I've tried less tightening some of the bottom screws on the cooler and the temperatures on Core #1 (and package) have decreased substantially, peaking to 75.c with an average of 61.5.c in a stress test (yet to test gaming, I imagine it's fine), but it's still slightly hotter than the other cores regardless of the tightening of the cooler. Maybe this is normal, after all?

 

[Darkbreeze]

There will ALWAYS be one core that is hotter than the others on almost every CPU out there, especially if it is not a model that uses internal solder rather than a liquid TIM (Paste) inside the unit between the die and heat spreader.

I've just never found one ever where every core was the same. There is always about a 2-5 degree variance and on some it can be as much as ten degrees. Anything beyond ten degrees, is a problem, and either requires attemps at repasting or reseating the cooler, which you have done, or replacement by the manufacturer as a higher than ten degree variance is a warrantable issue.

What speed are you running your memory at? Because that can have an affect on core temperatures.

What case model do you have, how many case fans are installed, what direction (Intake or exhaust) are each fan in each location blowing/oriented.

Is your CPU cooler mounted to blow front to back, like it should be, or do you have it turned to blow from bottom to top?

Part of your problem might simply be your motherboard. It's not a terribly impressive model when it comes to overclocking. Motherboard quality and design can play a huge rule in overclocking and CPU temperatures.

What is your core voltage set at?

What is your LLC set at?

 

[spazbandicoot]

There will ALWAYS be one core that is hotter than the others on almost every CPU out there, especially if it is not a model that uses internal solder rather than a liquid TIM (Paste) inside the unit between the die and heat spreader.

I've just never found one ever where every core was the same. There is always about a 2-5 degree variance and on some it can be as much as ten degrees. Anything beyond ten degrees, is a problem, and either requires attemps at repasting or reseating the cooler, which you have done, or replacement by the manufacturer as a higher than ten degree variance is a warrantable issue.

What speed are you running your memory at? Because that can have an affect on core temperatures.

What case model do you have, how many case fans are installed, what direction (Intake or exhaust) are each fan in each location blowing/oriented.

Is your CPU cooler mounted to blow front to back, like it should be, or do you have it turned to blow from bottom to top?

Part of your problem might simply be your motherboard. It's not a terribly impressive model when it comes to overclocking. Motherboard quality and design can play a huge rule in overclocking and CPU temperatures.

What is your core voltage set at?

What is your LLC set at?

Memory speed is set to 1600MHz.

I have two front fans (pulling air into the case), two top fans (blowing upwards, out of the case. Might be slightly overkill) and 1 rear exhaust fan (blowing backwards, out of the case). All are set to full speed (all of this is in a Corsair 270R).

The cooler is set as front to back.

Core voltage is currently set to 1.229v, and seems to be working out well so far. Anything reasonably lower causes a BSOD reading "WHEA_UNCORRECTABLE_ERROR" after 10-15 minutes of stress testing.

LLC is enabled.

 

[Karadjgne]

You have a Haswell cpu. With that is other things to consider when overclocking. VID is important, you'll want to keep that around 0.05v ± from the vcore, you'll use ± offset to get there. Ring voltage should be no more than 1.1v-1.2v, Ring cache should be somewhere around the same speeds as the cpu. You really shouldn't have much if any need for LLC until reaching @ 4.5GHz, so either leave that on auto or set it for medium. Too much LLC is just as bad as not enough. You'll also need to disable much of the eco settings, like c-states (c1-e is ok to leave), any phase controls etc.

You shouldn't be over 1.2v until @ 4.4 to 4.6GHz, so there's other factors than vcore affecting your stability, but having a higher vcore usually can muscle in stability, at the expense of higher temps.

My i5-3570k at 4.3GHz, 1.108v under Prime95 small fft torture test on a little 120mm H55 aio gets 70°C. Never exceeds 55°C gaming.

I'd not worry about exactly what your temps are until after you fix what's causing them to be so high. At 4.2GHz you should realistically not be higher than @ 1.14v area.

 

[TJ Hooker]

If it's less than 10C difference I wouldn't worry about it. If it's greater than that and it's consistently the same core that's hotter across multiple re-pastings/re-mountings then pretty much your only option if you want to fix it would be to delid.

Edit: I'll mention that this is exactly what happened to me. Delidding made a huge difference. That being said, I cheaped out and just used a utility knife to delid rather than buying a delid tool, and it was a pretty nerve wracking 20-30 minutes of going at the CPU with a razor

 

[spazbandicoot]

Alright, that's good to know.

Cheers for the help everyone. I'll be keeping everything said here in mind (or a notepad!).

 

[Darkbreeze]

Good luck man.