Sign in with
Sign up | Sign in
Your question

cpu weird temperatures

Tags:
  • Cooler Master
  • Cooling
  • Games
  • CPUs
  • Intel i5
Last response: in CPUs
Share
September 21, 2014 9:24:55 AM

So im having a i5 4670k cpu and cooler master hyper 212 evo cooler.

I have been wondering my weird temps.
So i have been monitoring my temps in demanding games and 3dmark benchmarks and my cpu temp haven't went like over 61c which i think is good.

But then i was thinking about overclocking and i used prime 95 and intelburntest and right away my cpu temp went up to 76c. I dont understand how is this possible because like in cpu demanding games it never goes even near that.
im so nervous and i need help.

More about : cpu weird temperatures

a b à CPUs
September 21, 2014 11:22:00 PM

Prime95 and IBT load your CPU far, far more than any normal application or game ever could. It's not intended for comparison to standard usage. Those are for stress testing to check stability. The fact that temperature level checking can be done is just a byproduct, not that actual use that was intended. So it's normal to have higher temps while stress testing than when gaming or running any normal processes.
m
0
l
a b à CPUs
September 21, 2014 11:24:55 PM

Even an overclocked CPU usually won't ever get as hot as a standard clocked cpu does when under full load in a stress test, if everything else is right and the voltage does not exceed what the cpu is capable of handling and sufficient cooling is present. If you don't have a good aftermarket cpu cooler, don't even think about overclocking. Not even a little. I don't even like enabling the turbo boost and turbo core features on rigs without a decent cooler. There's just too much heat from the system ramping up voltage.
m
0
l
Related resources
a b à CPUs
September 21, 2014 11:29:51 PM

while what darkbreeze has said is correct that does still seem rather high for stock settings, Many people will stop overclocking when their 4670k gets into the mid 70s on those tests and you havent even added any voltage yet.
m
0
l
a b à CPUs
September 21, 2014 11:42:08 PM

It would be high for stock settings if those temps were recorded during normal usage. Those temps at 100% load on all cores in a stress test, 76c might not be acceptable, but it might not be ABNORMAL either. A lot depends on the board and also there might be an issue of the cooler either not being mounted exactly right or a bad paste job. How much thermal paste did you use? If you used a pea sized amount, I'd say it's too much.

A rice grain sized amount is plenty. Also, the 212 EVO cooler has an adjustable thumbscrew in the middle that most novices neglect to tighten. It should be tightened down when the four outer hold down screws are a couple of turns out from fully tight and then those should be tightened. If you can press down on the heatsink against the top of the cpu and watch the temps come down at all, the cooler isn't mounted correctly.
m
1
l
September 22, 2014 9:53:51 AM

well sure it will stress cpu far more but really that much that my cpu temp will increase like almost 20c???
besides as far as i have watched others results with same cpu and cooler they still get temp like 60c temp by running IBT

And i just reinstalled my cpu cooler and placed new thermal paste but same happens again.
Do i even dare to overclock even to 4.0Ghz because of these temps?And if i do should i test my stability with IBT?

I just hope that hwmonitor and msi afterburner are showing wrong temps which i doubt.


And i used the rice grain method. Oh and my cooler was hyper 212 evo.... not plus
m
0
l
a b à CPUs
September 22, 2014 11:32:10 AM

Well you could OC to 4 without a voltage change so you wouldn't make more heat. It would be worth trying but you might not be able to OC this chip very well
m
0
l
a b à CPUs
September 22, 2014 3:12:22 PM

In reality, games rarely demand much of the CPU. Almost all games rely heavily on the GPU. So when gaming you're not going to cause CPU temps to come anywhere near its load max. As I said before, there is not much you can normally do on a system that compares with the 100% CPU loads for all cores that are applied by the 3 common stress tests. But you still want them to be able to handle the thermal and computational load at whatever clock speed you're currently sitting at before going any higher with it. If your rig can't handle running Prime while it's at the stock clock, don't overclock. You need to determine why your temps are excessive before you mess with mulitpliers or voltage.

I've personally experienced the same EXACT problem on at least three other systems using the EVO cooler and two of the three were solved by unscrewing the four outer hold down screws about halfway, tightening the center thumbscrew all the way tight that is right where the X of the hold down bracket comes together underneath the heatsink body and then tightening the outer four screws back to where they belong. The third unit had a pin hole in the copper tubing which had allowed the inner coolant that is in the heat pipes to escape and had to be RMA'd. I've never had similar issues with other coolers like Noctua and Phanteks. What you do is up to you but I'd seriously look at that first and go from there.
m
0
l
September 23, 2014 11:45:16 AM

Hey I have found new results. I tested another stress test, aida64 and results was like 65c max what seems like to make more sense.
I ran this stress test with same settings like in prime95 and with it i hit 76c.

Is this trustworthy? Besides now i monitored my temps with aidas own temp monitor, does that make any difference to that i run prime95 with hwmonitor?
m
0
l
a b à CPUs
September 23, 2014 5:56:18 PM

aida64 is the "weakest" stress test of the big 3

Prime95 is the middle one

Intel Burn Test is the hottest

I personally dont stress with aida because I have had overclocks that look perfectly stable under aida and then after half a second under prime crash. The temp differences you are getting between aida and prime arent suprising
m
0
l
a b à CPUs
September 23, 2014 6:57:59 PM

Jameson Clark said:
aida64 is the "weakest" stress test of the big 3

Prime95 is the middle one

Intel Burn Test is the hottest

I personally dont stress with aida because I have had overclocks that look perfectly stable under aida and then after half a second under prime crash. The temp differences you are getting between aida and prime arent suprising


Actually I was talking about IBT, Prime95 and OCCT. And to my knowledge, although I suppose things could change, Prime95 on Large FFT's has always been the "hottest" test when testing max temps. I've seen a lot of OC's that were fine on Intel burn test but threw errors within minutes on Prime small FFT's. Computers are made by humans. There is probably no way to ever be certain they are "stable" under all possible conditions. Anyhow, I never use Aida anyhow. I guess that's just personal preference.
m
0
l
a b à CPUs
September 23, 2014 7:03:07 PM

Dastan said:
Hey I have found new results. I tested another stress test, aida64 and results was like 65c max what seems like to make more sense.
I ran this stress test with same settings like in prime95 and with it i hit 76c.

Is this trustworthy? Besides now i monitored my temps with aidas own temp monitor, does that make any difference to that i run prime95 with hwmonitor?


Aida tests a lot of other stuff along with the CPU, like memory and hard drive function. If you can run a full 8-10 hours on Aida it's probably ok. If not, it will never make it through a Prime run. I run Prime with HWinfo or Core Temp open, so that should be fine.
m
0
l
September 24, 2014 11:16:18 AM

Okay im running out of ideas why my temps are so high. Maybe the cooler isnt just doing its job or my chip is just bad.

Btw one thing i havent done is to update my bios. I just build my pc like month ago and i didnt update bios because don't fix something if it isnt broken and i havent have any problems until now this one.
So is it possible that bios update can lower temps? What i have read that it can but it can also make it even worse :/ 
Can outdated bios tell wrong temperatures to monitoring?
m
0
l
a b à CPUs
September 24, 2014 8:04:49 PM

Actually, it IS possible, if the BIOS update addresses voltage issues in the automatic overclocking routines. It isn't common, but it is possible. Modern BIOS are not like in the old days where every update was a crapshoot. These days it doesn't make sense to not keep the BIOS updated though their flash programs or in many cases they have live or internet updates available right in the BIOS or by way of a Windows utility. It just depends on the board model and version in most cases. Don't expect it to cure the problem, but don't be surprised if it helps either. And also, incorrect calculation tables for temperature is another possibility although I haven't personally seen this one happen before. In most cases, if it seems to be overheating, it most likely is.
m
0
l
!