Not sure how to read the temperatures when stresstesting the OC

steffeeh · December 2, 2016

You see my build in my sig if it's helpful

Basically I have a 100% stable overclock on my 5820k at 4.4 GHz with 1.264V with my Noctua NH-U14S aircooler.
Now previously I have pretty much stated to myself that I won't be able to reach 4.5 GHz unless I watercool this thing with a good watercooler with 2x 140mm fans or so.
Though lately I've started to be unsure if that really is the case... Because now I'm starting to have the feeling that I might have been focusing on the wrong temperatures.

Basically I've kept an eye on the maximum temperatures on the cores and how they behave as I stresstest it in Aida64 but also when I game or do other CPU intensive tasks. I've pretty much ignored the temperature of the overall CPU as it's always lower, and I felt I should keep an eye on the hottest part.
But is this the proper way to do it? Because I'm starting to think that maybe the correct way to view the temperatures is to run the stresstest and just keep a basic eye on temps so it doesn't get wayyy too hot, but then decide wether it's too hot or not when the stresstest presents the average temperature and not focus too much on the maximum core temps?

For example, running a stresstest now and looking the temperatures as they go, the hotter cores hover around (i.e not calculated average!!) 70-75C while the cooler ones hover around 66-70C, and the whole CPU package showing roughly 66-67C.
But there are regurarly occurring temperature spikes in the cores, especially in core #3 (hitting 84-87C), and core #1 in 2nd place temp-wise (hitting 82-84C). Other cores spike at 74-76C or so.
On the other hand the calculated average temperature by Aida64 are 64-66C on all cores + the CPU package except core #3, the core #3 being at 70C (i.e well within safe limits by reading them alone).
And I'm trying to wrap my head around all this - sure those spikes hit bad temperatures... but they're spikes... they're no sustained temperatures, but still they're reaching high temps.

What temperatures should I let determine wether or not I can go further or if I should stop... maximum temps vs average temps... core temps vs package temps? Could I in theory reach higher safe temperatures (considering the average temperatures), or have I hit the ceiling for now (considering the temperature spikes)?

steffeeh · December 3, 2016

Anyone?

Shadow_Storm56 · December 3, 2016

23 hours ago, steffeeh said:

You see my build in my sig if it's helpful

Basically I have a 100% stable overclock on my 5820k at 4.4 GHz with 1.264V with my Noctua NH-U14S aircooler.
Now previously I have pretty much stated to myself that I won't be able to reach 4.5 GHz unless I watercool this thing with a good watercooler with 2x 140mm fans or so.
Though lately I've started to be unsure if that really is the case... Because now I'm starting to have the feeling that I might have been focusing on the wrong temperatures.

Basically I've kept an eye on the maximum temperatures on the cores and how they behave as I stresstest it in Aida64 but also when I game or do other CPU intensive tasks. I've pretty much ignored the temperature of the overall CPU as it's always lower, and I felt I should keep an eye on the hottest part.
But is this the proper way to do it? Because I'm starting to think that maybe the correct way to view the temperatures is to run the stresstest and just keep a basic eye on temps so it doesn't get wayyy too hot, but then decide wether it's too hot or not when the stresstest presents the average temperature and not focus too much on the maximum core temps?

For example, running a stresstest now and looking the temperatures as they go, the hotter cores hover around (i.e not calculated average!!) 70-75C while the cooler ones hover around 66-70C, and the whole CPU package showing roughly 66-67C.
But there are regurarly occurring temperature spikes in the cores, especially in core #3 (hitting 84-87C), and core #1 in 2nd place temp-wise (hitting 82-84C). Other cores spike at 74-76C or so.
On the other hand the calculated average temperature by Aida64 are 64-66C on all cores + the CPU package except core #3, the core #3 being at 70C (i.e well within safe limits by reading them alone).
And I'm trying to wrap my head around all this - sure those spikes hit bad temperatures... but they're spikes... they're no sustained temperatures, but still they're reaching high temps.

What temperatures should I let determine wether or not I can go further or if I should stop... maximum temps vs average temps... core temps vs package temps? Could I in theory reach higher safe temperatures (considering the average temperatures), or have I hit the ceiling for now (considering the temperature spikes)?

Thoes aRe some pretty high Spikes..... the average is lower but it seems like you have some really hot cores. this is normal.. I don't think ur current Temps are really bad but I wouldent go higher... that 0.1ghz increase can put out alot more heat depending on the cpu. Idk what ur supposed to look at but I look at the hottest core as it is still part of the cpu that is getting really hot and can be damaged by too much heat.