CPU core temps jumping around under stress test - Natural?

[wbynum]

I just put together a i5-3570k system today with the Asrock z77 Extreme4 MB. I am running Aida64 stress test. CPU oc'ed to 4.4ghz at 1.25v fixed Vcore. Max temps on each core are around 70c. One thing I noticed though is the temps on each core seem to jump from ~60c to ~70c and anywhere in between. Is this normal? I would have guessed the temps should rise to a max value and stay roughly around that value until the stress test was turned off. Thanks.

 

[ShintaiDK]

Yes its normal.

 

[dma0991]

It's normal for some fluctuation in temps. Mine would fluctuate +-5C or slightly more but it will rarely dip to more than 10C difference.

 

[ShintaiDK]

Less cores used in the stresstest the higher fluctuation basicly. Since the thread(s) will jump around.

 

[wbynum]

Thanks for the info. This was with all 4 cores enabled under the Aida64 stress test.

 

[SheHateMe]

Can I ask what cooler you're using + TIM?

I have mine running at Stock and the temps get that high under load. Im trying to figure out the problem before I bother overclocking.

 

[Ferzerp]

Any stress test that tests a variety of functions (or have any variation in the load produced actually) will product fluctuating temperatures. Processors drop in temp rapidly when not in use, and so any time they are waiting (or even doing a different instruction), they will change temps.

 

[wbynum]

Can I ask what cooler you're using + TIM?

I have mine running at Stock and the temps get that high under load. Im trying to figure out the problem before I bother overclocking.

I'm using my old Hyper 212+ HSF from my previous original i7 build. Picked up a tube of Céramique 2 grease from MicroCenter for $5.

Before I overclocked I believe my stress test temps were in the 50c - 60c range. So ~10c higher with the overclock.

 

[wbynum]

Found out about voltage offset and load line calibration today. So I ditched the 1.25v fixed volts and changed it to +0.005v offset and a LLC level of 2. Vcore under load is a bit higher than when Vcore was fixed at 1.25v. I tried some negative offsets and even though the Vcore under load was about the same I got bootup problems. Temps are still in the same 60c - 70c range under load so I am happy. Low Vcore at idle is nice.

 

[tweakboy]

Intel recommends 50's c and 60's c . Soo if your 70 and above, ur really hurting the long life it can have.

 

[Phynaz]

Intel recommends 50's c and 60's c . Soo if your 70 and above, ur really hurting the long life it can have.

Link?

 

[TakeNoPrisoners]

Intel recommends 50's c and 60's c . Soo if your 70 and above, ur really hurting the long life it can have.

Unless you have a link I will assume you are talking about AMD processors, in that case you are correct.

 

[2is]

Link?

Here you go.

http://forums.anandtech.com/showpost.php?p=33888429&postcount=10

 

[wbynum]

I spoke to soon on things being stable with voltage offset. Started getting random app crashes. So I am back to fixed voltage. A Vcore of 1.220v in the bios and 1.200v under IBT load, reported by CPU-Z, seems stable. Passed a 5 run Very High IBT test. Doing a 10 run test currently. Max temp on the hottest core is 87c. Under Aida64 and OCCT the temps don't get near that high. Around a max of 74c.

How many passes is recommended for IBT? What temps should I be seeing on this 3570k at 4.4ghz using the above voltages when running Very High IBT runs? I have a Hyper 212+ HSF on the cpu. Thanks.

 

[wbynum]

It just passed the 10 run IBT Very high test. Hit 88c max core temp on the hottest core. I tried the Maximum test but that just into the 90c+ range pretty quickly so I aborted.

 

[atomjuice]

heatsinks
I've used many different heatsinks...some exotic..expensive...on and on...both with amd and intel...the largest heatsinks are not always the best...particularly I would like to point out a propblem with many copper-pipe heatsinks: some of the 'direct contact' copper pipes will bend (they are hollow in many..or too thinly made..so the copper pipes bend inward on the surface when pressed against the cpu die)...this bent area in the copper pipes cause an area of depression which makes little or no contact with the die surface at all...even on expensive heatsinks that look amazing, this can be the case. I have found a few aftermarket heatsinks which have very solidly built copper pipes...the temps drop dramatically with these well made heatsinks....CHECK the strength and weight of the copper piping before buying that georgous heatsink (which is really a glossy piece of junk)...the heavier the piping, the better the contact..no depression from pressure...and excellent conductive quality...something I learned the hard way to check for!

 

[ShintaiDK]

Intel recommends 50's c and 60's c . Soo if your 70 and above, ur really hurting the long life it can have.

You couldnt be more wrong.

Intel recommends running IB below 105C to put it shortly. The Tjmax is there for a reason.