temps dangerous for 5960x at 4.4ghz?

Idontcare

Elite Member
Oct 10, 1999
21,110
64
91
I was running realbench h264 stress test x10. I'm getting up to 77C from 2 of the 8 cores at max. Can i push a little furthur? lets say 4.6ghz under 85C?

http://postimg.org/image/yn3tpy7h9/

Don't worry about clockspeed and temperatures, those two things are not the concern.

Be worried about voltage and temperature, those two are what kills your CPU.

Underclock your CPU but overvolt the bejeesus out of it and it will die in a matter of minutes as the temperature rises.

So the question you have to answer is "what voltage at that temperature?" The clockspeed itself is absolutely irrelevant to your concerns (killing the chip).
 

hunkeelin

Senior member
Feb 14, 2012
275
1
0
Don't worry about clockspeed and temperatures, those two things are not the concern.

Be worried about voltage and temperature, those two are what kills your CPU.

Underclock your CPU but overvolt the bejeesus out of it and it will die in a matter of minutes as the temperature rises.

So the question you have to answer is "what voltage at that temperature?" The clockspeed itself is absolutely irrelevant to your concerns (killing the chip).

In the image I am only putting in 1.275v I'm pretty sure the voltage is safe. I'm not even passing the 1.3v barrier.
 

BonzaiDuck

Lifer
Jun 30, 2004
16,889
2,202
126
Don't worry about clockspeed and temperatures, those two things are not the concern.

Be worried about voltage and temperature, those two are what kills your CPU.

Underclock your CPU but overvolt the bejeesus out of it and it will die in a matter of minutes as the temperature rises.

So the question you have to answer is "what voltage at that temperature?" The clockspeed itself is absolutely irrelevant to your concerns (killing the chip).

Didn't notice the poster until I'd read the post! "Of course that makes sense!" I said.

It could be a matter of personal judgment. TCASE on those E's is 66C. There's a likely error in sensor calibration; some cores may seem to run warmer than others. 2x 77C balances against 6x @ < 77C. You're only heating them up that much during the stress tests.

As for the voltage, we once had a "maximum safe" limit for the range @ 1.375V-something -- 32nm processors. I'd say 1.3V is the upper limit on these, but again- that's a personal judgment. With the temperatures, it could be less.
 

guskline

Diamond Member
Apr 17, 2006
5,338
476
126
hunkeelin: If you are TRULY stable at 4.4Ghz on a 5960X I would NOT try to go higher, it's just not worth it.

I have a custom water cooled rig for my 3930k and though some OCers go higher, I'm fine at 4.6 Ghz. It passes all the stress tests I can throw at it.

Have you downloaded Intel's own stability testing software, Intel Extreme Tuning? If not, do so and stress test your cpu for at least 6 Hrs (die hards say at least 24 hrs).

I'm not sure you'll be cool at 4.4 Ghz but you might.

From everything I've read the 5960X maxes at 4.2 to 4.5 Ghz. Remember EIGHT cores SIXTEEN threads. That's a ton of heat.

BTW, I always buy the Intel Protection Plan insurance for my overclocked cpus.

Finally, I know the feeling of the excitement of having the latest and greatest cpu etc and how you want to push it to the limit. However, if it isn't stable and running relatively safe from a temp/voltage standpoint the "extra" bit form OCing too far is not worth it.

Good luck.
 

james1701

Golden Member
Sep 14, 2007
1,791
34
91
The only thing you missing is a few points on a benchmark and maybe a few seconds if you encoding.

Otherwise, just run it and enjoy it.