Which temp is THE temp?

rlwings · September 18, 2017

Hey guys. Just a quick question about watching temps during overclocking...

I am overclocking an i7 7700K on an ASUS PRIME Z270-A board. I am using AIDA to test...

Question: Which temps are the ones to watch. Overall average CPU package? Average per core? or each individual core's maximum? ... If the answer is to watch each core's maximum, then what about the fact that this maximum is only reached briefly and not very often? ... I would think that it is the average heat that will determine the rate of degradation over time as it happens more often than brief spikes... I'm at 4.9GHz @ 1.28V and averaging mid 70's temps, except for that damned core 0 which spikes into the early 90's... Just delidded as well, but only used AS5 under the hood.

Any help would be greatly appreciated.

Mira Yurizaki · September 18, 2017

Average over time is probably the best one to use if you want to estimate longevity. But my concern is that the sampling frequency is often low (1Hz) and either a lot can happen in between sampling periods or it samples at a point where it makes it appear to be more or less than it really is.

Although the problem with degradation is that nobody really knows what the average life expectancy of the processor is. So even if you had some reasonable amount of thermal data, it's highly unlikely you'll get an accurate prediction on when the part is expected to start showing failures.

Lord Nicoll · September 18, 2017

43 minutes ago, rlwings said:

Hey guys. Just a quick question about watching temps during overclocking...

I am overclocking an i7 7700K on an ASUS PRIME Z270-A board. I am using AIDA to test...

Question: Which temps are the ones to watch. Overall average CPU package? Average per core? or each individual core's maximum? ... If the answer is to watch each core's maximum, then what about the fact that this maximum is only reached briefly and not very often? ... I would think that it is the average heat that will determine the rate of degradation over time as it happens more often than brief spikes... I'm at 4.9GHz @ 1.28V and averaging mid 70's temps, except for that damned core 0 which spikes into the early 90's... Just delidded as well, but only used AS5 under the hood.

Any help would be greatly appreciated.

I use real temp GT, which posts actual core temps in real time as a number, I was pretty happy with it for working out the core temps, I don't use package or socket temperature as they're not as reliable as actual core temps. The temps seem kinda high for that frequency and voltage too, what cooler are you using? It might be better to invest in a better cooler than anything.

rlwings · September 19, 2017

13 hours ago, Lord Nicoll said:

I use real temp GT, which posts actual core temps in real time as a number, I was pretty happy with it for working out the core temps, I don't use package or socket temperature as they're not as reliable as actual core temps. The temps seem kinda high for that frequency and voltage too, what cooler are you using? It might be better to invest in a better cooler than anything.

Thanks, ya, my temps are actually averaging in the high 60's with two cores peaking at 91... I am using the Hyper 212 with a delidded processor and MX5 throughout... I just ordered some Coollaboratory Liquid Ultra. (Do you think I could use liquid ultra on this cooler as well. I think it has a copper base.)

Lord Nicoll · September 19, 2017

5 hours ago, rlwings said:

Thanks, ya, my temps are actually averaging in the high 60's with two cores peaking at 91... I am using the Hyper 212 with a delidded processor and MX5 throughout... I just ordered some Coollaboratory Liquid Ultra. (Do you think I could use liquid ultra on this cooler as well. I think it has a copper base.)

My i7 7700K is delidded with Thermal Grizzly Conductonaut, it's also a liquid metal TIM but it's a tiny bit better, it doesn't dry up over time (well so far that's proven true) but really they both do pretty good. I have it all round, on the die and IHS. The Hyper 212 is probably the main thing making bad temps, but also normal thermal paste pumps out and dries very fast with direct die cooling on a CPU, so that might also be an issue. I think the 212 has an aluminium part of the base, if you get liquid metal on that, the cooler will break it if is aluminium. On the heatpipes it's fine.