What has caused my idle CPU temps to go up 20 degrees?

Hi all, I'm hoping for some help figuring out why my idle CPU temps have increased drastically over the past couple of weeks. In the past, when I had a few tabs open in a browser and maybe Microsoft Word or another program open, my CPU temps would hover around 30 degrees C, and the Task Manager would indicate that my CPU usage was hovering between 1% and 4%. Now, the CPU temps hover around 50 degrees C in the same circumstances, and the Task Manager indicates that CPU usage is still consistently between 1% and 4%. Nothing has changed that I'm aware of.

Any ideas on what could be causing this, and how to fix it?

My laptop is an MSI GS43VR with an Intel Skylake i7-6700HQ CPU, nVidia Pascal 1060 GPU with 6GB VRAM, 16GB RAM, Intel Wireless 8260, and a 500GB Samsung 850 Evo SSD. The CPU and GPU were both pasted with IC Diamond Thermal Compound by the reseller when I purchased it back in August 2016.

Thanks for any help you can offer!

 

It's probably the power plan changed from Balanced to High Performance, or some other update / change is causing the CPU to not down clock when idle. Check your power plan, make sure it is set to Balanced, and go into Advanced and set the CPU to 0%/100% to get the full range of CPU frequency cycling.

You did mention paste, so it could also be paste drying out..., but look for OS changes / updates, power plan (slider?) changes first.

 

No change in load but a change in temps sounds like the year and a half old paste is going bad

Load temps should be through the roof compared to where they used to be if it's paste. If it's just a setting issue 100% load should be no hotter than before

Paste drying out and getting hotter is a chicken and egg thing so it'll keep getting worse if that's what's going on.

 

I think its Win 10 scheduled tasks for OS servicing. Use Dism++ to cleanup older SxS files incl. WinSxs temps and backups too.

 

Read OP's post one more time....

"my CPU temps would hover around 30 degrees C, and the Task Manager would indicate that my CPU usage was hovering between 1% and 4%. Now, the CPU temps hover around 50 degrees C in the same circumstances, and the Task Manager indicates that CPU usage is still consistently between 1% and 4%. Nothing has changed that I'm aware of."

OP should apply new thermal paste!!

 

W10 now can run apps/tasks in background with lowest priority.
Probably repaste is required.

 

1 up to 4% with apps/tasks running in the background is normal. If Win 10 run scheduled tasks the processor load in % will increase a lot more than this. So yeah, a repaste is needed.

 

Thanks, everyone. There may be a few contributing factors: for some reason my power plan was changed to "Movie," and after changing it back to "Balanced", and setting the CPU to 1-100% per @hmscott 's suggestion, my CPU does seem to run about 5 C cooler, but that's still about 15 degrees above what it used to be. I'm going to look into doing a repaste--I've never done one before, but I'm willing to take the plunge if it means a cooler and longer-lived CPU.

 

Did you undervolt yet?, if you did before the Windows Update, your undervolt may need to be refreshed - XTU doesn't like MS's Windows 10 SoftShutdown and thinks a crash or other non-standard shutdown was done - and coming out of Sleep the undervolt might not stay applied.

Most CPU's can take a -100mV undervolt, so I suggest trying that first. That's usually enough to get 100% load temps down 10c.

If you can do more, as some have gone as high as -220mV - average I've seen is -140mV.

If -100mV isn't stable, try -50mV and go up / down from there.

That may not help much with Idle, but it will help with load temps.

I hate to suggest people pull apart their laptop, but you already did it once, so it's probably just time to refresh the paste. So consider changing pastes, and technique of application.

Generally letting air in is bad, so keep the spill over to a minimum, I usually use an Xacto knife to clean an edge around the CPU so pressure won't gish out paste open to the air.

Too thick pads can make it so you can't tighten down evenly and an edge can be open to air.

Good luck

 

Sorry for resurrecting this thread, but I finally was able to repaste the CPU (and GPU), and it's still having the same problem: high CPU idle temps, roughly 15-20 degrees C higher than they used to be. I cleaned off the old, dry IC Diamond and replaced it with a thin layer of Kryonaut Thermal Grizzly. I also dusted out the fans and around the heat pipes (though there wasn't much dust).

Even after this, I'm still seeing idle CPU temps in the 45 - 50 C range when they used to hover around 30 C. Again, there has been no change in CPU load (1% to 4% while idle). Also, I didn't mention it in my original post, but my CPU Cache and Core are undervolted -169.9 mV, iGPU and System Agent undervolted -125 mV using ThrottleStop (and they have been undervolted since around the time I purchased the notebook in August 2016). I've updated the graphics drivers for both the iGPU and the dGPU, I'm current on Windows Updates, and I even updated the BIOS, hoping for a fix - still the same issue.

And strangely, when the system is not idle and I'm running something like Firestrike or Unigine, it typically has respectable CPU temps in the mid-50s C. (If anything, the repaste might actually have slightly brought down my gaming temps... but it didn't change my idle temps)

Any other thoughts on what this might be, or what I might be able to do to troubleshoot? Thank you!

 

Are you living in a hotter environment? And how are the CPU clocks and voltage, are they dropping when idle or running full-bore?

 

Can you hear your CPU fan running at idle?

 

Sometimes, yes. The CPU fan will kick on when the CPU temp exceeds 50C, which never used to happen on idle, but it happens quite a bit now.

#1 Nope, I've been living in the same environment.

#2 As for CPU clocks and voltage, I just ran HWinfo64 for 10 minutes while idling (nothing open in the foreground except HWinfo64 itself). Here are some pertinent results:



So the clock speeds aren't full bore (which would be 2.6 GHz base, 3.5 GHz turbo for my CPU), but I'm not sure if the reported ~1.2 GHz is considered normal or too high when idle... any thoughts?

 

Looks to me like something is hogging CPU resources when you are idle.
Check task manager to see what process is using CPU cycles.
If it's something like svchost or something, try to find out what is running in the background.
Stuff like windows experience index, volume shadow copy or any number of cancer things can do background scans.

Run Throttlestop 8.60 and look at the c0% tab. When you are trully fully idle, it should not ever exceed 0.3% on ANY thread. And if you are using deeper c-states, like deeper than C3 (c6 or deeper) it can drop to 0.1%. Check to make sure you have C-states enabled. Disabling C-states can cause 10C higher temperatures, even if the CPU is downclocking. (Speed Shift can also downclock the CPU).

 

Thanks. The Task Manager hasn't been very illuminating, unfortunately; rarely does it say that the CPU is exceeding 4% usage, and I haven't been able to find any particular process that has been sucking CPU resources.

Here is what ThrottleStop is showing (while no other programs are open in the foreground):



C0% hovers around the values shown here (including the noticeably higher value in the third row, hovering around 6.0). Overall C0% hovers around 1.7. This does look quite a bit higher than the 0.3 you mentioned.

I've never played around with C-states before. Is there somewhere in ThrottleStop I can check to see if C-states are enabled/disabled? Or do I need to head into the BIOS or somewhere else for that? (You also mentioned Speed Shift; do you mean it would possibly help if I enabled the "Speed Shift - EEP" option in TS?)

Thanks!

 

What does it look like when you click C10 at the bottom?

 

Something is using up background processes.
I have c states totally disabled and here is what I have. And you're doing worse with cstates enabled. You need to clean your OS.

 

If you don't mind me asking, when was the last time you checked your OS for Malware?
If its been a while, I recommend downloading Malwarebytes free and use it to scan the system.

Considering you already cleaned out the laptop from dust and re-pasted, it might be a change on a software level.

Out of curiosity which thermal paste did you use and how did you apply it?

Also, is it possible that your previous temperatures were reported incorrectly and that the temps you are reading now are pretty much what one would expect?
40-45 degrees C while idling is considered to be within tolerance levels (or, acceptable).
I think for your particular CPU, 40 degrees is what the reviews stated the temps would be around.

Also, if you aren't experiencing thermal throttling and high temperatures when the CPU and GPU are fully stressed or gaming, then I don't think you'd need to worry about it too much (plus undervolting the cpu might be beneficial).

Otherwise, if you still think the temps are too high and that something might be hogging your resources without you knowing about it (after you did the Malware scan and ascertained there's none on your system), you could try doing a clean Windows install as a last resort.

 

It's definitely not the thermal paste.
He shouldn't be hitting 1.4% or 6.3% at idle like that. If fans are running at auto speed and anything hits the CPU, it's going to cause the temps to rise. And the load is randomly placed on each thread also.

 

I use Malwarebytes and the scans come back clean.

I used Thermal Grizzly Kryonaut and spread a line layer out using the spatula that came with it.

I don't think so. My idle temps were around 30 C for over a year. Now they hover around 50 C, and the CPU fans will come on whenever the temps exceed that temperature, which has become frequent now.

True, but I'm concerned that the heat while idle is unnecessarily shortening the life of my components. Also, the CPU fans kicking on regularly when I'm browsing/word processing/etc. can be pretty irritating given how loud and whiny they are. (I've undervolted the CPU as well.)

I was wondering whether that might do the trick... I'm willing to give it a shot if it'll help!

Here's a screenshot of that:

Thanks, that's very helpful. I suppose I can try a clean Windows 10 install and go from there.

--
New data point: I used my notebook for about 4 hours today without it being plugged in (which I don't usually do). My idle temps were fine! Back to normal - hovering around 30 C. Just like when it's plugged in, the power plan was turned to "Balanced." But unlike when it's plugged in, "Battery Saver" was turned on. And when I turned off "Battery Saver," or plugged the notebook back in, the CPU temp went up ~15 C.
So now I'm wondering--is something going awry with the power settings? Or maybe power saver is restricting some CPU-hogging background processes, and this is just further evidence that I need to clean the OS...

 

Try to find out what processes are using CPU resources. You can look in taskmanager and I believe there is also a third party task manager you can use as well.

 

Well the only thing is that Task Manager doesn't show any processes as using a lot of CPU resources. Here's a typical picture of what Task Manager reports while CPU temps are near 50 C (here, I have two Firefox tabs open, and everything else is a background process):



So nothing really jumps out. (I wish something did, it would sure make figuring this out a lot easier!)

 

Firefox eating up 1.9% is enough. And the killer network service isnt helping. Is that service even required?
You're supposed to install only the base drivers, not the software suite. What is DSAsuite?
And CTF Loader, ipoint and itype...you have so much stuff in your processes....

 

Yeah, the Killer Network Service is bothering me as well. I only have the driver installed - my notebook came preinstalled with the software suite, and I got rid of it a while back. Yet the process remains.

As for the others... good questions. I don't know. I'll do some research and see if I can get rid of them. I didn't think these processes would be responsible given their small CPU usage, but I'll try to remove whatever is causing them. Fingers crossed...

Edit: iType and iPoint are for my wireless mouse and keyboard. DSAsuite is for the Intel Driver Update Utility. CTF Loader is for MS Office.

 

How thin do you think you spread the line you made? Did you spread the Kryonaut so the entire IHS is almost covered or is just a 2-3 mm wide strip about .1 mm thick?

If coverage is sparse, you may want to find a video of the X Cross method or just simply lay down a line to cover 80% down the middle, without using the spatula, and tighten your heat sink in a criss-cross method or in a manner specified by any numbering on the heat sink.


(Edited "X Cross Method" to include a link to a @Phoenix post)

 

If you have a CPU running at 1.2W but idling at 50C, you have some seriously poor cooling (A lot of ICs I've played with can even dissipate 3W of heat without needing an additional heatsink.) I think your C states, given that you have so much garbage in the background, is actually quite reasonable. But I'm completely unsure as to why your CPU is idling at 50C yet drawing 1.2W. You must be living in the Sahara. Are your load temperatures any different?

For reference, my 8250U, previous 6500U and 6200U idle at around 33C. My friend's 6700HQ idles lower than 30C. A laptop idling at 50C is so 2012.

Given that you managed to 'solve' the issue by just going to battery power, can you give us screenshots of ThrottleStop on all the power plans you currently have on both power modes (AC/DC)?

I read somewhere that CPU temperatures are not accurate when reporting low temperatures, (so I'd assume can be influenced by voltage differences (I'll assume your AC Adapter outputs 20V and your battery is either ~11v or ~14v depending on cell count) but I think a differenc of 20C is a little extreme.

 

Outside on a warm afternoon, idling with 30-40 tabs of chrome open, using ~5W, my i7-4800mq is at 40C.

 

I'm idling at 45-47.5 degrees C along with light browsing in Google chrome (which is using 3.6% of CPU resources and 7GB RAM - yeah I have A LOT of tabs open).
But my CPU is desktop 1700 in a laptop (with 32GB RAM)

 

The load temps are a higher, but not by much. Gaming temps reach mid-to-high 50s, which is about where they've always been, including back in the good old days when my idle temps hovered around 30 C.

Absolutely. Here's an image of everything "solved" - battery power, "Balanced" power plan with "Battery Saver" turned on. No programs are running in the foreground except Throttlestop itself and Task Manager.




And here's the same thing, but with how it looks like while plugged in - Balanced power plan without "Battery Saver" turned on (the power slider is just set to the default "Better Performance")

 

Well that leaves me speechless. My best guess would be to just ignore these things because the only thing I can think of is a faulty sensor. Your C states are normal, package power is normal, and task manager looks normal. Could there be a short-circuit within your computer which is mucking something up? Does your laptop warm up noticeably (as in, does the case feel warmer) when you plug in the power?

It also looks like your dGPU is off(which is good) so THAT can't be causing the extra heat. There is no other component I know of in a laptop, that can generate enough thermal heat to raise the temperature of a CPU by 10C

 

Nope. Thanks for taking time to help me diagnose this, though.

And thanks to everyone else in this thread as well for all of your insight. Even though we haven't gotten down to the bottom of this, it's been a great learning experience for me, and I appreciate everyone's help. I'm going to try a completely fresh install of Windows next to see if, for some reason, that helps - and then be very observant and I reinstall programs. I'll report back on how that goes.