2700K @ 4.5GHz but varying temps reported

TheSoupladel · 29 Mar 2018 at 22:05

So i got my new cooler today and GFX card, got it all set up eventually (the usual post build, but it wouldn't actually post drama occurred) and once i did a clean windows 10 install and upto date drivers, I turned my attention to the overclock.

As above i am running at 4.5GHz with 1.35v, with my ram voltage set at 1.35v with the speed and timings manually set as well, speed step is turned off and LLC set at 50%. I am running this on a Asus P8Z68-V/PRO and the cooler in question is the Corsair H100i V2 in a push/pull exhaust config mounted in the roof of the case

I am running a stability test in Aida 64 and my max temps are sitting between 59 & 62c, but the actual temps, as reported by both Corsair link and core temp are varying between 50 and 60c depending on the core.

So the question is, is it likely this is caused purely by problematic thermal paste application, hence the inconsistencies across cores or is it likely caused by the manner in which Aida 64 conducts its stability testing?

I ill be honest and say I had issues getting the pump mounted which absolutely will have affected the pre-applied compound, but there did seem to be a reasonable and consistent covering.

Munchies · 30 Mar 2018 at 08:55

Could you share individual core temps under load? As your post to me suggests initially that there is only 3C disparity, and then 10C disparity.
If your hottest core is 62C and your coolest core is 59C there is no issue.

It's normal to have temperature disparity between cores, assuming each core is being worked equally the reason for the disparity is (to my understanding) due to the thermal compound / application between die and heat spreader.
Your application of TIM between CPU / Cooler shouldn't affect core temps individually as such. 10C disparity seems a little high for this chip, but if max is 62C it's well within spec so nothing to worry about, however if it's 3C difference that's normal.

TheSoupladel · 30 Mar 2018 at 09:03

Ok, so from memory the max temps were, 59 for Core 0, 62 for Cores 1 & 2, then 59 for Core 3. However, the current temp on Core 0 would be 50, then 54, then 57, then it might drop back to 50 or 49, it was a similar story across the other cores.

This was all within the Coretemp app, but it was a similar story being reported within Corsair link.

Based on what i know the max temps are well within expected temps and stress/stability testing aren't a accurate representation of what the temps are under gaming as the temps are actually a good bit lower. I was more concerned with the disparity across the cores from one second to the next when all 4 cores were showing as under 100% load consistently.

Munchies · 30 Mar 2018 at 09:16

Oh, now I understand what you mean. That's completely normal in my experience with AIDA's test, if you enable only the CPU test (disable FPU / Cache / Mem etc) you'll likely see more stable cpu temps I presume it's because AIDA is switching tests, so while the cpu is showing 100% load it's being handed a different task.

TheSoupladel · 30 Mar 2018 at 10:56

Well I have come across an annoying issue and I can’t make out if it is caused by the overclock or windows 10 and my hardware.

I switched the computer on this morning and shortly after loading windows I got a BSOD then it was stuck in that cycle of rebooting and then BSOD.

One of the first error was “error by pointer” then it was “memory management”

I am using a legit download of Windows 10 from MS, albeit I am using one of those real but frowned upon “cheap” eBay licence keys

Munchies · 30 Mar 2018 at 17:16

I presume you mean reference by pointer error? Which is typically driver related, memory management could also be driver related.

What graphics card did you replace, and what with? Use DDU (Display driver uninstaller) in safe mode to ensure all drivers are removed particularly if you changed from AMD to Nvidia / reverse. Then install the latest drivers.

mistersprinkles · 30 Mar 2018 at 17:52

My hottest core on my 4790K is sometimes 7C hotter than my coldest core. It's always the same cores. This is under identical load. I wouldn't worry about it.

Your temps are fine.

TheSoupladel · 30 Mar 2018 at 18:46

Munchies said:
I presume you mean reference by pointer error? Which is typically driver related, memory management could also be driver related.

What graphics card did you replace, and what with? Use DDU (Display driver uninstaller) in safe mode to ensure all drivers are removed particularly if you changed from AMD to Nvidia / reverse. Then install the latest drivers.

I appreciate what you are saying but its didn't apply in this case because i went from Nvidia to Nvidia (680 GTX to 1060 GTX OC) but did a completely clean install of Windows 10, rather than an upgrade.

I think I may have resolved the error problem, in that the advice i took on overclocking suggested i should manually set the voltage for my RAM rather then leaving it on Auto, the i came back this afternoon the firstling i did was switch the DRMA voltage back to Auto and it has been fine ever since.