What causes abnormal CPU heat spikes?

benjii

benjii

benjii

Soldato
Joined
26 Apr 2013
Posts
5,014
Location
Plymouth
My 6700k is OC'd to 4.7Ghz @ 1.344v under load; motherboard is a Asus Maximus VIII Hero Alpha. During stress testing, the temperature never goes above 72c. However, I've been playing some GW2 today, which is fairly CPU intensive and usually keeps the CPU around 55-60c, I noticed that I had a max temp spike of 81c.

Is that caused by slow mobo reads and subsequently not ramping the fans quickly enough? If so, what can I do about that? I realise 81c for a quick spike is not the end of the world, I'm just curious as to what causes it and how to mitigate it.
 
Different things stress cpus in different ways. That's why P95 is nowadays not commonly recommended because it put's undue stress on cpus.

I would suspect that your spike is coming from the cpu either having it's strings pulled by GW2 in a different way than other games have previously. OR it was caused by windows doing something else in the background for a moment putting more load on the cpu.
 
I had erratic spikes on my 6700k @4.6ghz. I had a Corsair GTX110i .. after I replaced the stock thermal paste the spikes were no longer there and the max temp dropped by around 2 to 4 degrees .. I also know that some stress testing can cause different temperatures .. if your only getting these max temps while playing could it be to do with temperatures within the case? How long does the spike actually go for? perhaps post a screenshot of the temp graph
 
timbo2410 said:
I had erratic spikes on my 6700k @4.6ghz. I had a Corsair GTX110i .. after I replaced the stock thermal paste the spikes were no longer there and the max temp dropped by around 2 to 4 degrees .. I also know that some stress testing can cause different temperatures .. if your only getting these max temps while playing could it be to do with temperatures within the case? How long does the spike actually go for? perhaps post a screenshot of the temp graph
Click to expand...

I'm using thermal grizzly and temperatures within the case are low. I don't have a graph of a spike to hand unfortunately, but I know that the spikes last for only a few seconds. That's why I'm wondering if it's related to slow motherboard readings not ramping the fans fast enough. That would also explain why stress testing sits at 72, because it's a consistent load.
 
if the spikes are that fast then fans are not going to be able to cope with it any way. For interests sake, find a few different apps to stress the CPU and see what the results are and if you can duplicate the spikes under stress. If the spikes are not present during any of these then we can rule out any sort of issue with cooling (ie thermal paste, cooler itself etc). Then track down what is different between the stress testing and when you are running GW2
 
As it goes, I've been refining my OC lately so I've done quite a few 10 hour stress tests, with a number of applications. On one occasion, with RealBench, I spiked to 74, but it stabilised on 72 and never went higher.

I guess without maintaining a higher RPM for 55-65c range I may not be able to prevent the spikes.
 
Don't forget that your internal air temp will be hotter due to the heat from the GPU.

Plus it's possible that the heat spike doesn't actually occur, just a software anomaly.
 
Hi,

I have the same query as the OP. My 6700k is at 1.344v @4.7ghz. My cooling is a custom loop with 2x360 and 1x240 rads.
When i am gaming, my temps will fluctuate between 30 and 55 degrees. But my max temps will be around 65. When i look at the graphs i see that every now and again my temps spike for a second or two. So it seems 95% of the time it is below 55, then on occasion spikes by what i consider a lot.
As i am on water with a lot of rad, i was expecting 55 to be the max. Which seems so until the spikes.
Now, after a little investigating....
I have found whilst using OCCT, that cpu: OCCT acts the same. Temps seems to fluctuate a lot whilst the cpu is loaded. It will fluctaute mainly between 45-60 with spikes of 1-2s at 75.
Cpu: Linpack seems to work how i expect. Temp quickly rises to about 58 and steadily over 30 mins levels out at 63 no spikes and variance of only a couple of degrees (versus the OCCT variance of ~20degrees).

So what confuses me is, what is causing these spikes. Is this normal with todays cpu's?
Also why do i get temp spikes during gaming that are higher than my OCCT cpu: linpack max temps after hours of stress test?
 
Processes can cause a slight fluctuation in the temps depending on what core is being used. Windows updating or something loading in the back, is check the core usage % and refer that to the core that's doing the spiking. I don't the core is idle when the temp shoots up, more likely it is processing something, and probably around 50% usage going by your temp ( just a guess)

Hwmonitor is perfect for that
 
Could it be a sensor/reporting issue.
Just thinking out loud, if you're spiking to 90 for example, and your nominally 'under load temps' are say 60 or 70, the coolers not going to be able to shave off 20c in less than a second, surely you'd see it creep down in temp from the spike over about, I dunno, at least 30 seconds to a minute? Not instantly?
 
chilipeppersc said:
Processes can cause a slight fluctuation in the temps depending on what core is being used. Windows updating or something loading in the back, is check the core usage % and refer that to the core that's doing the spiking. I don't the core is idle when the temp shoots up, more likely it is processing something, and probably around 50% usage going by your temp ( just a guess)

Hwmonitor is perfect for that
Click to expand...

For my example, why would a process like windows update cause temp fluctuations 10-15 degrees higher than the linpack stress test.
The temp behaviour is very different to what I have had in the past.

Normally, I would do occt and linpack. Linpack and Intel burn test would give me my highest temps. They would get to the max temp and stay there, with maybe 1 degree variance. Example 70degrees.
If I gamed, or used my pc normally, the highest I would see would be 55-60.

Today, if I use occt or ibt. My temps will hit 75 for 1 second then back down to around 55-60. This during 100% load.
Linpack behaves as I expect, it shoots to about 58 within a minute, and stays there without spikes or variance. Gradually creeping up to 65 after an hour and my water has hit equilibrium.
If I game, or use my pc normally, the highest I would see will be 65-70. Higher than linpack at 100% after one hour.
This just seems odd compared to all previous experiences. Also from around the Web, people with a similar setup report highest temps whilst gaming as 55-60, mine is this on average with max 70.

Is it possible, someone with a 6700k could post a screenshot of a graph with core temps over time during a cpu: occt test? It will give me a better way to see how mine correlates. Thanks.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom