CPU core temperature spikes

[Moogs]

Hey fellas,

 

I've consistently had problems setting a fan curve for my CPU because the core temperatures are constantly spiking all over the place whenever I do practically anything on my computer.

I've used all sorts of temperature monitoring software to look into this, and they always look about the same. I've attached a screenshot - you can see random gigantic spikes in temperature, which only last for a second or even milliseconds at a time. I have cores 0 through 5 displayed on this screenshot, and the huge increase in baseline temperature is when I started Prime95.

Is this possibly just a known issue with this series of CPUs? Or perhaps cores read temperature in a really finnicky, jumpy way? Regardless, it sucks setting a fan curve based on the core temperatures because simply opening a tab in chrome will cause the fans to ramp up crazy high and then ramp back down all within a couple seconds.

I would use a more stable and generic sensor like the overall CPU temp but for most programs that seems to be measuring socket temperature, which never gets nearly as hot as the actual cores, and also has a pretty lengthy lag between the cores heating up and the socket heating up.

 

Does anyone else have this problem with a decent solution? Right now I'm basically just manually changing my CPU cooler's fan when I go from idling to gaming. I've also tried setting curves in the BIOS but I don't get nearly as much control, and I still have the same problem with the fan ramping up and down rapidly.

 

Side note: How accurate are temperature monitoring software supposed to be? Core Temp and HWInfo both read similar core temps, which are about 10-15 degrees warmer than what Speedfan reads for core temps.

 

Specs:

CPU: i7 8700k

Mobo: ASUS Prime Z370-a

Cooler: Cryorig R1 Universal

 

[Haro]

21 minutes ago, Moogs said:

Hey fellas,

 

I've consistently had problems setting a fan curve for my CPU because the core temperatures are constantly spiking all over the place whenever I do practically anything on my computer.

I've used all sorts of temperature monitoring software to look into this, and they always look about the same. I've attached a screenshot - you can see random gigantic spikes in temperature, which only last for a second or even milliseconds at a time. I have cores 0 through 5 displayed on this screenshot, and the huge increase in baseline temperature is when I started Prime95.

Is this possibly just a known issue with this series of CPUs? Or perhaps cores read temperature in a really finnicky, jumpy way? Regardless, it sucks setting a fan curve based on the core temperatures because simply opening a tab in chrome will cause the fans to ramp up crazy high and then ramp back down all within a couple seconds.

I would use a more stable and generic sensor like the overall CPU temp but for most programs that seems to be measuring socket temperature, which never gets nearly as hot as the actual cores, and also has a pretty lengthy lag between the cores heating up and the socket heating up.

 

Does anyone else have this problem with a decent solution? Right now I'm basically just manually changing my CPU cooler's fan when I go from idling to gaming. I've also tried setting curves in the BIOS but I don't get nearly as much control, and I still have the same problem with the fan ramping up and down rapidly.

 

Side note: How accurate are temperature monitoring software supposed to be? Core Temp and HWInfo both read similar core temps, which are about 10-15 degrees warmer than what Speedfan reads for core temps.

 

Specs:

CPU: i7 8700k

Mobo: ASUS Prime Z370-a

Cooler: Cryorig R1 Universal

 

do you have the latest UEFI installed ? 

[Moogs]

20 minutes ago, TofuHaroto said:

do you have the latest UEFI installed ? 

I do, yes. Ended up updating it recently due to some unrelated issues. The issue is the same pre- and post UEFI update.

[Haro]

34 minutes ago, Moogs said:

I do, yes. Ended up updating it recently due to some unrelated issues. The issue is the same pre- and post UEFI update.

What voltage do you have and do you have mce

And can you tell us what your PSU model is?

[Moogs]

I'm afraid I don't know what MCE is. As for voltage I haven't touched anything, the system is just set to the stock OC through the UEFI.

When I was having to RMA my GPU recently, the tech had me check my 12v rail which was within normal limits. I'm afraid I don't know really anything about voltage.

My PSU is a Corsair RM750X.

 

If it helps, I've included a picture of the voltages that HWInfo lists.

[Haro]

9 hours ago, Moogs said:

system is just set to the stock OC through the UEFI.

That's mce basically,and it  does change the voltage so you can check in the bios what your settings are

[Moogs]

Probably too late to reply but I've been hella busy...

 

Core temperatures are still totally whack, with all the huge spikes all over the place, with MCE and all the Mobo enhancements turned off. In fact, the CPU actually idles at a higher temperature for some reason with it all disabled. Voltages in the UEFI are all the same as HWInfo showed.

[Guest]

There is several reasons why you get "spikes" when running Prime 95 for example.

 

First, we'll touch on the actual physical aspect of the processor. The stock TIM under the IHS plate is pure garbage. replacing it will reduce your temps greatly (15c'ish)

 

Then, we'll look at the software. Prime95 is designed to Ramp up usage when it implements AVX instructions, and that's where you see the big tower leap in temps.

 

You can have some fun experimenting with using OCCT while this program has AVX feature selections of AVX, AVX2, and AVX 512 (for older AMD) and disabled. You'll get different load temps testing with these settings. 

 

But all in all, I don't see any real issue here. You won't have any issues until your loaded at around 100c with that chip. I like to be 85c or less on air cooling loaded personally.

 

Quote

How accurate are temperature monitoring software supposed to be? Core Temp and HWInfo both read similar core temps, which are about 10-15 degrees warmer than what Speedfan reads for core temps.

Was interesting to see SpeedFan. 

Try Hwinfo64 maybe, use the sensors.

 

 

hwi_624.exe

[Moogs]

I'm afraid I'm not remotely knowledgeable on any of these subjects to understand a lick of what you posted, ShrimpBrime.

But, the spiking occurs during regular usage. When Prime95 is running at full stress the temperatures sit stable at temperatures that are actually kind of scary.

I was using the blended test for Prime95 earlier and didn't wait long enough for it to get to the high stress portion - HWInfo reports some pretty nasty temperatures on high stress, although the temps are a bit strange. A few cores will sit right around 95-99 degrees C which is obviously too close to 100 degrees for my comfort. Other cores will sit around 60 C.

 

My cooler is a Cryorig R1 Universal which I understand is a pretty powerful air cooler. Full load temps of 99 C seems unreasonable for such a cooler running at 100%.

As well, I touched the heat sinks on the cooler while running this high stress test for about 10 minutes, and the sinks are not hot - not even warm. I'd say they felt a few degrees above room temperature at best. Is it possible that the cooler isn't interfacing with the CPU snugly enough? Or maybe the thermal grease isn't doing its job?

 

Sorry that my issue has kind of taken a right turn here but everything seems to change the more I look into this.

 

Edit: HWI64 works nicely, I trust the temperatures it's giving me.

[Guest]

7 minutes ago, Moogs said:

I'm afraid I'm not remotely knowledgeable on any of these subjects to understand a lick of what you posted, ShrimpBrime.

But, the spiking occurs during regular usage. When Prime95 is running at full stress the temperatures sit stable at temperatures that are actually kind of scary.

I was using the blended test for Prime95 earlier and didn't wait long enough for it to get to the high stress portion - HWInfo reports some pretty nasty temperatures on high stress, although the temps are a bit strange. A few cores will sit right around 95-99 degrees C which is obviously too close to 100 degrees for my comfort. Other cores will sit around 60 C.

 

My cooler is a Cryorig R1 Universal which I understand is a pretty powerful air cooler. Full load temps of 99 C seems unreasonable for such a cooler running at 100%.

As well, I touched the heat sinks on the cooler while running this high stress test for about 10 minutes, and the sinks are not hot - not even warm. I'd say they felt a few degrees above room temperature at best. Is it possible that the cooler isn't interfacing with the CPU snugly enough? Or maybe the thermal grease isn't doing its job?

 

Sorry that my issue has kind of taken a right turn here but everything seems to change the more I look into this.

 

Edit: HWI64 works nicely, I trust the temperatures it's giving me.

Lol, yea its a bit to take in if you dont know a lot about it. Its ok.

 

90c that cooler would be hot at the base and warm near the top if loaded up heavy. 

 

You could try and reseat the cooler. 

If you do, take note how the paste had spread. Too much? Not enough? Maybe looks like uneven spread or air bubbles?

 

Yea Hwinfo has everything. Its a good application. Hwmonitor is useful for older tech like AMD Phenom chips.

[Moogs]

43 minutes ago, ShrimpBrime said:

Lol, yea its a bit to take in if you dont know a lot about it. Its ok.

 

90c that cooler would be hot at the base and warm near the top if loaded up heavy. 

 

You could try and reseat the cooler. 

If you do, take note how the paste had spread. Too much? Not enough? Maybe looks like uneven spread or air bubbles?

 

Yea Hwinfo has everything. Its a good application. Hwmonitor is useful for older tech like AMD Phenom chips.

I think I have a tube of thermal paste left over from when I first installed it, I think I'll take a crack at reapplying it. When I did it in the first place, it was my first time applying thermal paste, and after hearing all the horror stories about too much or too little paste I was probably a bit anxious.

 

Thanks for the help!