Arctic MX-4 TIM upgrade raised my temperatures??? Temperature jumping around??

Zuhayr378

Reputable
May 15, 2014
282
0
4,860
So I went ahead to buy some new thermal paste for my GPU, which I did and I got some non-conductive MX-4 on sale. It lowered temperatures by about 6 degrees which is actually a very good decrease.

BUT, upon applying it on my Pentium G3258 which was doing fine in terms of temperature (I did it anyways) from it's H80i stock paste, I noticed that idle temperatures went from 31 degrees Celsius to 34 (wait for it), BUT the thing is now my temperatures are jumping all over the place. Minimum is 26 degrees which is a good sign over the early 29, and the maximum ON STARTUP is waaay at 52 whereas it used to be about 44 if I can remember (it never ever hit above 45 on my account). Now it's going like 27, 32, 37, 29, 41, 35, and so on just jumping up and down.


So is this up down movement bad?

P.S. It's overclocked to 4.4 Ghz forgot to mention that with a 1.34 voltage setting (keeping it higher doesn't hurt it apparently and it's my only way of stability).

EDIT: Halfway through this thread the idle became 28 and stable. I still ask why this jumping around happens.

ALSO, does the paste have to "burn in" before giving better results?
 
Solution
1-5C, I believe, is the range quoted from their website. I have yet to see an actual 5C gain from AC5, and even after curing I find that Arctic MX-4 does an equal or better job. *shrugs*

The temperature spikes are, indeed, probably related to load, or perhaps other shifts in power phases. If 60C is your maximum load, you have nothing to worry about.
Most "modern" thermal compounds - Arctic MX-4 amongst them - have very little to no burn-in, or curing, time. Arctic Silver 5 is an oldie-but-goodie that required ~200 hours of various thermal cycles to cure, but even then it amounted to 1-5C difference.

Anyway, the most important number to pay attention to is the temperature at load. As long as that isn't drastically worse, I don't think you have to worry much.

1.34 volts may be a bit high for Haswell, but, someone else is going to have to chime in on that.
 

Zuhayr378

Reputable
May 15, 2014
282
0
4,860


It's 58 and 56 on the two cores so far on full load. Haven't done gaming yet, but are those good temperatures for the basic full-load tests that I have run with Intel BurnTest?

I completely understand that 1.34 is quite high but it's the only way for sure-fire stability at 4.4Ghz and it offers these kind of temperatures that I am happy with it. Even before the TIM upgrade I never ever went beyond 70 EVER and the chances of going beyond 65 were so so little (it happened maybe 2 or 3 times in heavy gaming). My average gaming temperature is in a range of 50-57 or so usually with the Corsair H80i.

 

Zuhayr378

Reputable
May 15, 2014
282
0
4,860


Correct me if I'm wrong, 1-5 degrees is a not a small range and 5 degrees is actually a good difference in temperature.

Though I ask again why temperature may jump around. Is this related to CPU load?
 
1-5C, I believe, is the range quoted from their website. I have yet to see an actual 5C gain from AC5, and even after curing I find that Arctic MX-4 does an equal or better job. *shrugs*

The temperature spikes are, indeed, probably related to load, or perhaps other shifts in power phases. If 60C is your maximum load, you have nothing to worry about.
 
Solution

Zuhayr378

Reputable
May 15, 2014
282
0
4,860


It's all good now. The temperatures are in a very nice zone, the thermal paste DID make a difference, and I now now what the spikes are for. In both your explanation and what I've seen. Thanks for clearing up some things.