CPU Temperature Fluctuating, Even When Idle?

xtobymc

Honorable
Jan 5, 2013
175
0
10,710
My CPU setup is as follows:
Intel i5 3570k overclocked to 3.7GHz
Asrock Z77 Extreme4 Motherboard
Corsair H60 Closed Loop Liquid Cooler
Two Corsair SP120 fans on the radiator in push-pull configuration

Recently I've noticed that my CPU temperature is fluctuating quite rapidly, and I'm not sure why. When I first boot my PC the idle temperature sits at around 31c and seems stable. However, the moment I start running any application the temperature suddenly goes in a pattern like this, changing every second or so: 32c, 35c, 38c, 35c, 33c, 40c, 45c, 35c, etc.

Is it normal for temperatures to change this rapidly? My GPU isn't like this at all, it goes up and down by 1 degrees every few seconds, never skipping up or down by 10c like my CPU does. I also just opened up my case and re-applied new thermal paste and cleaned up the dust in the radiator, but the problem is still there. I also noticed when downloading a game off Origin the temperature hit 50c...

Any help would be great, perhaps I should try lowering the overclock to see what happens. Although I've had 3.7GHz stable for about a year now.
 
Solution
Yes, it's normal. Especially if turbo boost is enabled which generally for most systems, it is by default. Even without it though the temps change like that. Loads shift from core to core and temps raise and lower within milliseconds. Each time a windows process kicks in or returns to the background or a call for resources occurs, it creates heat in the cores. Even at idle, there is a lot going on. GPU doesn't run a bunch of processes like the CPU. GPU generally only raises in load and temp when you're doing something related to the visual display.

How much thermal paste are you applying? An amount equal to a grain of rice is plenty for any heatsink or waterblock. More may be too much and less is definitely too little. Also, make sure...
Yes, it's normal. Especially if turbo boost is enabled which generally for most systems, it is by default. Even without it though the temps change like that. Loads shift from core to core and temps raise and lower within milliseconds. Each time a windows process kicks in or returns to the background or a call for resources occurs, it creates heat in the cores. Even at idle, there is a lot going on. GPU doesn't run a bunch of processes like the CPU. GPU generally only raises in load and temp when you're doing something related to the visual display.

How much thermal paste are you applying? An amount equal to a grain of rice is plenty for any heatsink or waterblock. More may be too much and less is definitely too little. Also, make sure you follow the tightening guidelines to the T. Too tight and you can damage the board but too loose will certainly cause a failure to efficiently transfer heat to the block. You might try pushing down on the water block slightly for thirty seconds or so while watching the temps in your monitor. If temps lower while pressing down and raise when you release pressure, you're too loose for sure.

 
Solution

xtobymc

Honorable
Jan 5, 2013
175
0
10,710


Okay, thanks, it's good to know the fluctuations are normal, it just took me by surprise because my temperatures used to stay steady and only moved up and down by 1-2 degrees every few seconds when on load.

In terms of thermal paste I usually apply a small pea sized blob, I've always done it the same way and never had problems before. The cooler bracket is pretty secure too, but I'll definitely give that technique a try of pushing it down for thirty seconds.

Edit: I tried pressing it down, but there were no noticeable changes to the temperature. It's already secured down pretty tight so I don't think that's an issue.
 
A pea sized amount may be too much. In the past the physical size of the cpus was larger which means there was more real estate to cover with TIM so a pea sized amount may have been ok depending on how big a pea we're talking. Your i5 has a smaller lid size than previous generations. A pea sized amount is probably too much, in my opinion. More is not better, and never has been.

TIM does not help with the transfer of heat, it hinders it, except to the extent that micro pockets and areas that are not true flat which become filled with TIM do increase the ability to transfer heat so long as there is not TIM in between the areas that would have been touching had there been no TIM between the contact points. Metal to metal always transfers heat more efficiently than any TIM regardless of the composition of the paste. That being said, you may be fine but take a close look at your mount. If you can visually see any TIM that has squeezed out from between the lid and the heatsink, you should clean and remount using a smaller amount of TIM.
 

davethejackal

Distinguished
May 25, 2007
24
0
18,510


That's not true. The heat spreader on the top of the chip is what you're trying to cover, not the die itself. What's more whilst the manufacturing process in 22nm this doesn't tell us anything about the die size, 22nm is the size of individual components on the chip. The die size is decided by how these are arranged and how many 100s of millions of components there are on it.
 
Regardless, the physical size of the lid is not the same as they used to be. Look at a Pentium 4, which was pretty much the standard when the "pea sized" technic became popular. It was gigantic compared to an FX, A series, or I-series CPU. If you want to use a pea sized dollop on your lid, be my guest, every CPU I've pasted with a rice sized amount, of which there have been hundreds, had full or near full coverage of the lid when it was later removed or removed to verify coverage.