[SOLVED] temps high at idle

P0tluck94

Commendable
Nov 22, 2021
264
15
1,685
i have never had any issues with idle temps, but here recently my cpu (ryzen 5 3600)and gpu (asus ko 3060ti oc 8gb) have all idled at 39-42 degrees celcius but here recently cpu is still 39-42 degrees but gpu idles at 50 degrees, i know 50 degrees isnt hot but it shouldnt idle at that temp , under load gaming it goes up to 66-70 which it never has before its always been under 65, i have no idea what happened or how to fix it

i did notice one day my gpu tweakll was on OC mode which i never set it too but none the less it was on there and i put it back on gaming mode and have since uninstalled the gpu tweak program and removed the registry entries thinking it might still be on OC mode but my Mhz are still default (1755) as well as power limit are back to default (100%) , i looked in nvidia control panel i dont have power mode to high i have it set on normal as i saw some people had high idle temps because they had the power setting in manage 3d set to high performance , ive also ran heaven benchmark on extreme and the card got hot pretty fast to the point i didnt want to run it longer (71 degrees).

things i have tried
i have tried restoring to an earlier date
cleaned out fans/case, removed side panel glass
uninstalled and freshly reinstalled newest drivers for nvidia
reinstalled/repaired chipset drivers
opened window
pointed a fan into case
set up msi afterburner fan curves
rolled back nvidia drivers to an earlier version i know i ran cooler at

My pc is as follows

MSi B550M pro vdh wifi motherboard
ryzen 5 3600/ stock cooler
32 gigs ddr4 3600 cas 16 ram
asus ko 3060ti Oc 8gb LHR
case is a cooler master master box ( dont know the exact model, 2 fans in front screen front with air vents on the edge)
i dont have any fans up top i have the stock 2 fans in the front and i have added a exhaust fan in the rear
PSu is a seasonic focus 850 gold semi modular

my ambient room temp is 75


please help i have no idea whats going on, i dont have a ton of money , im scared to use it as when i just browse things temps go up more , afraid to play games etc etc ty
 
Solution
Perfect. Good call on including the board power draw. (again because I'm being lazy) I set mine to 200W power draw using an 85% power limit in afterburner.

GPU clock 1860MHz
Mem clock 1750MHz
GPU temp 72C
Fan speed 2030rpm
GPU Voltage 0.98V
Board power draw 200W
GPU chip power draw 141W

Takeaways:
  • Again, I'll say that anything under 80C is nothing to worry about. The card will automatically thermal throttle to prevent damage at 90C.
  • Leave 0 rpm idle on. It preserves the lifespan of your GPU fans, which are statistically the first thing to die on a GPU.
  • Your core voltage seems a bit high, but not alarmingly high. You could do some tweaking (undervolting) to bring that value down, which would reduce power draw...
First off, you need to tell us fan speeds when you talk about temps. See below.

50C is a perfectly normal GPU idle temp. Modern GPUs will turn off their fans at temps below 55C. If at some point in the past, you'd overridden the factory fan curve to force the fans to always spin, then you could see temps below that.

Also, 70-75C is a typical stock/factory target temp (to where the GPU fans will adjust accordingly to maintain). Depending on the quality of the cooler (and the thermal paste application), that may be inaudible, a mild hum, a loud hum, or a vacuum cleaner.

If you start to get nearer to 80C and the GPU fans are [should be] screaming, then it's a problem.


PS. I'll try to remember to post my EVGA 3060Ti FTW temps/fan speeds when I can.
 
Last edited:

P0tluck94

Commendable
Nov 22, 2021
264
15
1,685
First off, you need to tell us fan speeds when you talk about temps. See below.

50C is a perfectly normal GPU idle temp. Modern GPUs will turn off their fans at temps below 55C. If at some point in the past, you'd overridden the factory fan curve to force the fans to always spin, then you could see temps below that.

Also, 70-75C is a typical stock/factory target temp (to where the GPU fans will adjust accordingly to maintain). Depending on the quality of the cooler (and the thermal paste application), that may be inaudible, a mild hum, a loud hum, or a vacuum cleaner.

If you start to get nearer to 80C and the GPU fans are [should be] screaming, then it's a problem.


PS. I'll try to remember to post my EVGA 3060Ti FTW temps/fan speeds when I can.
fan speds on the cpu are at 55 degrees celcius

Cpu 1350 rpm
sys fan 1 800 rpm
sys fan 2 800 rpm
Front Panel______________________________

sys fan 3 (rear) 1300 rpm

gpu idle at 50 degrees celcius ( user defined fan curve)
fan 1 275 rpm
fan 2 275 rpm

ill get fan speeds at 65 degrees if you need those as well

the issue is my temps have never been this high at idle until a few days ago and i have no idea why, i also didnt mention ive ran multiple virus scanners also to tule out viruses
 

P0tluck94

Commendable
Nov 22, 2021
264
15
1,685
275rpm GPU fans is probably a mistake. Can you visually confirm the GPU fans are spinning? Usually they don't spin at less than.....900ish rpm.
might of been a glitch in hwmonitor? it said 275 they are at 1900 at 60 degrees without user defined @ 74% , with user defined they are lower at 1250 rpm @ 60% so i have the user defined lower than default? trying to add a pic it says needs admin something
 

Zerk2012

Titan
Ambassador
might of been a glitch in hwmonitor? it said 275 they are at 1900 at 60 degrees without user defined @ 74% , with user defined they are lower at 1250 rpm @ 60% so i have the user defined lower than default? trying to add a pic it says needs admin something
70 C gaming for the video card is not hot most of the modern cards don't even cut the fan on at 50C.

EDIT if you get up around 85C then post back.
 

P0tluck94

Commendable
Nov 22, 2021
264
15
1,685
70 C gaming for the video card is not hot most of the modern cards don't even cut the fan on at 50C.

EDIT if you get up around 85C then post back.
If i wasn't worried about everything running hotter than normal or of the blue I wouldn't of posted, my system never ran warm until all a sudden the other day and still is, the Temps dont even go down with 40° air from outside which is odd, I don't have the money to replace things just to wait till it gets to 85 like you mentioned, there's something wrong for my system to just stay running hot a pc shouldn't idle at 50-52°, im trying to figure out what that may be, my computer before could run heaven benchmarks for an hour and never break 65° it runs for 15 min now and is at 70, something is wrong with my computer I don't know what it is so I'm here to try to find out.
 
I'll give you a couple voltage/frequency points to reference.

My EVGA 3060Ti FTW
CPU (Ryzen 5600G) idle
2x120mm intake fans, 1x120mm exhaust. All at 1100ish rpm
Silverstone PS07
Room temp approx. 22.5C

15+ minutes in FurMark.
1850MHz Core @ 0.89V (70% Power limit, +130MHz core clock offset)
1750MHz VRAM
66C (core) @ 1776rpm

2040Mhz Core @ 1.03V (100% Power limit, +130MHz core clock offset)
1750MHz VRAM
78C (core) @ 2260rpm

Idle
52C @ 0rpm

Happy to share more info if you want. Just too lazy to share screenshots of GPUz right now.

When something 'suddenly' changes, it's usually software-related.

Also, if you have a "GPU Hot Spot Temp" readout (GPUz/HWinfo64), that's usually* the VRAM.
 
Last edited:

P0tluck94

Commendable
Nov 22, 2021
264
15
1,685
My EVGA 3060Ti FTW
2x120mm intake fans, 1x120mm exhaust. Silverstone PS07. CPU idle
15+ minutes in FurMark.

1850MHz Core @ 0.89V (70% Power limit, +130MHz core clock offset)
1750MHz VRAM
66C (core) @ 1776rpm

2040Mhz Core @ 1.03V (100% Power limit, +130MHz core clock offset)
1750MHz VRAM
78C (core) @ 2260rpm
I have no idea what that all means, all know is something has changed, as I stated before my pc would idle at 39° it now idles at 50-52°, it never went over 60-63 in games it now goes to 70, one day it was fine the next it wasn't, I can't Think of any changes made either, I know what I'm saying isn't real technical you'll have to bare with me I had a stroke so my terms might be confusing, I just know one day the pc was nice and Cool ran smooth next it wasn't.

Is there anything off the top of your head I might be able to look at to see if say it's on oc mode or something?
 
  1. Recreate my test. Run FurMark
  2. Monitor GPU info
    1. Core clock
    2. Core voltage
    3. Core temp
    4. fan speed
  3. Report in similar format as I did so you/we can compare.
  4. For idle temps, monitor GPU usage %. It should be <5% at idle.
Just repeating that "my card used to run cooler" isn't helping us. If you can't provide any more info than some temps, then we have no choice but to start with the here and now. What it's CURRENTLY doing. That's our starting point.
 
Last edited:

P0tluck94

Commendable
Nov 22, 2021
264
15
1,685
  1. Recreate my test. Run FurMark
  2. Monitor GPU info
    1. Core clock
    2. Core voltage
    3. Core temp
    4. fan speed
  3. Report in similar format as I did so you/we can compare.
  4. For idle temps, monitor GPU usage %. It should be <5% at idle.
Just repeating that "my card used to run cooler" isn't helping us. If you can't provide any more info than "it used to run cooler", then we have no choice but to start with the here and now. What it's CURRENTLY doing. That's our starting point.
What resolution did you run this at? Anti aliasing? Did you use a preset?

Edit: running it at 2k 2560x1440
 

P0tluck94

Commendable
Nov 22, 2021
264
15
1,685
2560x1440
8x MSAA
"GPU Stress Test"
okay my rooms like an oven


15 min test ran on furmark at 2560X1440 8x MSAA

Gpu clock 1845
Memory Clock 1750
Gpu temp 68.2
Hot Spot 79.6
Fan speed 1 82% 2248 rpm
Fan Speed 2 82% 2246 rpm
Gpu load 100%
GPU Voltage 1.0180
Board power draw 200 watts
Gpu chip power draw 142.3

gpu at idle is 50.2 degrees (after test might go lower after it cools down)
3%

Edit .. forgot to add memory was at 7000 mhz if that matters
 
Perfect. Good call on including the board power draw. (again because I'm being lazy) I set mine to 200W power draw using an 85% power limit in afterburner.

GPU clock 1860MHz
Mem clock 1750MHz
GPU temp 72C
Fan speed 2030rpm
GPU Voltage 0.98V
Board power draw 200W
GPU chip power draw 141W

Takeaways:
  • Again, I'll say that anything under 80C is nothing to worry about. The card will automatically thermal throttle to prevent damage at 90C.
  • Leave 0 rpm idle on. It preserves the lifespan of your GPU fans, which are statistically the first thing to die on a GPU.
  • Your core voltage seems a bit high, but not alarmingly high. You could do some tweaking (undervolting) to bring that value down, which would reduce power draw and heat generation at the same frequency. Or more likely would allow higher frequencies (NVidia GPU Boost 3.0 is a greedy mistress). Like I said, I found (using MSI afterburner, but most use the same system) that +130MHz frequency offset was stable. That just means that, at your same 1.02V, you would now expect to see 1980MHz.
  • Your fan speeds are higher, but my card has 3 fans and yours has 2. So that's to be expected. Your temp is also 4C less than mine. So it's probably a wash.
  • The "stock" 3060Ti power limit is 200W. My card clearly has a factory expanded TDP of 230-240W. Check the dip switch next to the 8-pin power connectors labeled with "P mode" and "Q mode" to see if the card is potentially in Q mode? Not sure what that setting implies, would have to research. Q mode may be "stock", and P mode may allow higher power draw. Or maybe the difference is just a fan curve.
  • If you get really brave/desperate, you could re-apply thermal paste to your GPU. Pretty much the same as you'd do to a CPU, but adjusting quantity according to die size. I had very good results re-pasting my RX480. Bringing my load temps from 85C@ 2400rpm to 71C@1100rpm. I'd like to try that with my 3060Ti, but with the GPU stock/pricing, I can't bring myself to "risk it" at the moment. The risk of damage may be low, but never zero.
    fwebp
 
Last edited:
Solution
Apr 8, 2022
12
1
15
I've had this too figured out a way to fix it with power settings, idk if this would work for you but it does for me and some others too
Press the windows key search power & sleep settings, then additional power settings
Change to balance mode / high performance (you can run balanced mode if high performance still makes your CPU high temperature on idle, I just recommend it if you're gaming and because it doesn't really affect me on high performance) then change plan settings
Change advanced power settings
Scroll down to Processor power management, double click it
Double click Maximum processor state
Click Setting then change the percentage lower, mine is 90% right now yours could be lower or higher to solve it but usually 90-95% almost fixed it for people
Just don't go too low, I saw some people go to 60% too
Click Apply then OK and close the other window
And' youre done

Hope this helps for everyone :D
 

P0tluck94

Commendable
Nov 22, 2021
264
15
1,685
Post a couple pictures of the case most of those have poor airflow and no dust filters you could just need a good cleaning.
I cleaned the whole thing with my electric air duster before I posted this thread, it had a meshy (not very open)! front with 2 fans that have vent slots on the sides, I also removed the side glass which still didn't help the Temps, I even took the card out, used the electric air duster thing and reseated the card.
 

P0tluck94

Commendable
Nov 22, 2021
264
15
1,685
Perfect. Good call on including the board power draw. (again because I'm being lazy) I set mine to 200W power draw using an 85% power limit in afterburner.

GPU clock 1860MHz
Mem clock 1750MHz
GPU temp 72C
Fan speed 2030rpm
GPU Voltage 0.98V
Board power draw 200W
GPU chip power draw 141W

Takeaways:
  • Again, I'll say that anything under 80C is nothing to worry about. The card will automatically thermal throttle to prevent damage at 90C.
  • Leave 0 rpm idle on. It preserves the lifespan of your GPU fans, which are statistically the first thing to die on a GPU.
  • Your core voltage seems a bit high, but not alarmingly high. You could do some tweaking (undervolting) to bring that value down, which would reduce power draw and heat generation at the same frequency. Or more likely would allow higher frequencies (NVidia GPU Boost 3.0 is a greedy mistress). Like I said, I found (using MSI afterburner, but most use the same system) that +130MHz frequency offset was stable. That just means that, at your same 1.02V, you would now expect to see 1980MHz.
  • Your fan speeds are higher, but my card has 3 fans and yours has 2. So that's to be expected. Your temp is also 4C less than mine. So it's probably a wash.
  • The "stock" 3060Ti power limit is 200W. My card clearly has a factory expanded TDP of 230-240W. Check the dip switch next to the 8-pin power connectors labeled with "P mode" and "Q mode" to see if the card is potentially in Q mode? Not sure what that setting implies, would have to research. Q mode may be "stock", and P mode may allow higher power draw. Or maybe the difference is just a fan curve.
  • If you get really brave/desperate, you could re-apply thermal paste to your GPU. Pretty much the same as you'd do to a CPU, but adjusting quantity according to die size. I had very good results re-pasting my RX480. Bringing my load temps from 85C@ 2400rpm to 71C@1100rpm. I'd like to try that with my 3060Ti, but with the GPU stock/pricing, I can't bring myself to "risk it" at the moment. The risk of damage may be low, but never zero.
    fwebp
Yeah I'm not trying to do thermal paste lol I have watched people do it but I can't afford another card 😂😂.
I'm not trying to overclock the card if that's what the+130 v means that you posted I just want my Temps back and to stop freaking me out when I see them on the higher side.

Positive note is that gpus are coming down in price now that the tariffs were lifted allowing imports of chips again and the building of chip plants being done in the US.

Negative note, I still couldn't afford a new one, 😂😂

Conclusion : I know the Temps aren't extremely high I just knew how much stuff costs and right now with me not having the money at the moment to replace a gpu when I noticed the card tuning Warner than it was previously I started freaking out, I read that gpu tweak can still be running the card in oc mode even after uninstall, but we ran furmark which would of showed the card being oc'd right?
 
Yeah I'm not trying to do thermal paste lol I have watched people do it but I can't afford another card 😂😂.
Fair
I'm not trying to overclock the card if that's what the+130 v means that you posted
This setting is actually considered "undervolting", which I would argue is BETTER than stock operation. Applying less voltage [than stock] to a given frequency. The lower the voltage, the better (as long as it's not so low as to cause system crashes). In a nutshell, CPUs and GPUs are bound to a preset frequency vs voltage curve by the manufacturer. Algorithm advancements and fabrication tolerances have come a long way, but the manufacturer still needs to build in a "safety factor" of additional voltage offset along the curve to ensure stability so they don't wind up with a bunch of RMAs from customers thinking their chips are broken, when in reality they're just not getting enough voltage to retain stability. What we the consumer have the freedom to do, is choose to test/validate that the chip is stable at lesser voltage(s) for a given frequency(ies).
I just want my Temps back and to stop freaking me out when I see them on the higher side.
They're on the normal side. That's what I was up till midnight running tests on my system to show you.
ran furmark which would of showed the card being oc'd right?
Nvidia themselves implemented their black box GPUboost3.0 to auto OC every GPU within preset limits of frequency/current/power/voltage/temp. The 3060Ti has a "stock" boost clock of 1670MHz. You're seeing ~200MHz higher than that. That's not Asus' fault, that's Nvidia's GPUboost3.0. As you can see, you're running at 200W which reporting software will probably display the "PerfCap" reason as "Power Limit". As mentioned with undervolting, if you tell the card to apply less voltage at a given frequency (say 1850MHz) then your card will draw less than 200W @ 1850MHz. Nvidia's GPUboost3.0 algorithm will see that none of the boost limits (frequency/current/power/voltage/temp) are being reached, so it will automatically increase core frequency accordingly. More than likely, until it gets back to the 200W TDP (assuming you're not going to set power limit >100%). That results in a higher frequency = more performance, but at probably the same 1.02V you had before. It's free performance.

You can also play around with setting the power limit to LESS than 100% (which I run regularly). This would be akin to my earlier results [from post #8] of the "70% power limit" which limits my card to 165W power draw. I of course had my +130MHz "undervolt" applied, but you can see there that my results 1850MHz, 0.89V, 66C, 1775rpm was certainly "better" than the 1860MHz, 0.98V, 72C, 2030rpm I got without the undervolt applied. That means my undervolt saved me 35W at the same frequency. I limited what Nvidia's GPUboost3.0 can do by setting a lower power limit, but my undervolt allows me to achieve the same frequecy/performance as "stock" (again, a loose term because true stock is 1670MHz but I will be that most/all GPUs [consumers/reviewers] rarely dip that low without manual intervention or abnormal test stress [looking at you OCCT]).
ga7c8TJhC9LsC8NVjanMJD-970-80.png.webp
 
Last edited:

Zerk2012

Titan
Ambassador
I cleaned the whole thing with my electric air duster before I posted this thread, it had a meshy (not very open)! front with 2 fans that have vent slots on the sides, I also removed the side glass which still didn't help the Temps, I even took the card out, used the electric air duster thing and reseated the card.
Is the temp in the room the same as it was? Using a case with bad airflow it doesn't take a lot to raise the temps.

Edit and no reason to change any voltages your still well below a problem area.
 
Press the windows key search power & sleep settings, then additional power settings
Change advanced power settings
Scroll down to Processor power management, double click it
Double click Maximum processor state
Click Setting then change the percentage lower, mine is 90% right now yours could be lower or higher to solve it but usually 90-95% almost fixed it for people
  1. The OP is more concerned about GPU temps, which this doesn't apply to.
  2. You need to know what this is actually doing. By setting 99%, you're effectively removing the CPUs ability to "boost", thereby capping the frequency at/near the advertised "base frequency". (also, any % is being calculated off the base frequency) That's leaving "stock" performance on the table. When I've done this (to limit power/heat in specific uses, not because of high temps), I created a desktop shortcut to my power settings, so I could quickly and easily revert back to 100% when I needed it.
  3. I would say this is a more useful thing to do on laptops (while on battery, in your lap) than desktops, which is why this setting is more accessible on laptops via the charging/battery icon on the system tray.
 
Last edited:

P0tluck94

Commendable
Nov 22, 2021
264
15
1,685
Fair

This setting is actually considered "undervolting", which I would argue is BETTER than stock operation. Applying less voltage [than stock] to a given frequency. The lower the voltage, the better (as long as it's not so low as to cause system crashes). In a nutshell, CPUs and GPUs are bound to a preset frequency vs voltage curve by the manufacturer. Algorithm advancements and fabrication tolerances have come a long way, but the manufacturer still needs to build in a "safety factor" of additional voltage offset along the curve to ensure stability so they don't wind up with a bunch of RMAs from customers thinking their chips are broken, when in reality they're just not getting enough voltage to retain stability. What we the consumer have the freedom to do, is choose to test/validate that the chip is stable at lesser voltage(s) for a given frequency(ies).

They're on the normal side. That's what I was up till midnight running tests on my system to show you.

Nvidia themselves implemented their black box GPUboost3.0 to auto OC every GPU within preset limits of frequency/current/power/voltage/temp. The 3060Ti has a "stock" boost clock of 1670MHz. You're seeing ~200MHz higher than that. That's not Asus' fault, that's Nvidia's GPUboost3.0. As you can see, you're running at 200W which reporting software will probably display the "PerfCap" reason as "Power Limit". As mentioned with undervolting, if you tell the card to apply less voltage at a given frequency (say 1850MHz) then your card will draw less than 200W @ 1850MHz. Nvidia's GPUboost3.0 algorithm will see that none of the boost limits (frequency/current/power/voltage/temp) are being reached, so it will automatically increase core frequency accordingly. More than likely, until it gets back to the 200W TDP (assuming you're not going to set power limit >100%). That results in a higher frequency = more performance, but at probably the same 1.02V you had before. It's free performance.

You can also play around with setting the power limit to LESS than 100% (which I run regularly). This would be akin to my earlier results [from post #8] of the "70% power limit" which limits my card to 165W power draw. I of course had my +130MHz "undervolt" applied, but you can see there that my results 1850MHz, 0.89V, 66C, 1775rpm was certainly "better" than the 1860MHz, 0.98V, 72C, 2030rpm I got without the undervolt applied. That means my undervolt saved me 35W at the same frequency. I limited what Nvidia's GPUboost3.0 can do by setting a lower power limit, but my undervolt allows me to achieve the same frequecy/performance as "stock" (again, a loose term because true stock is 1670MHz but I will be that most/all GPUs [consumers/reviewers] rarely dip that low without manual intervention or abnormal test stress [looking at you OCCT]).
ga7c8TJhC9LsC8NVjanMJD-970-80.png.webp
So how do I do the:+130 thing or the less than 100%? Is that by simply sliding the power limit in bar to say 98% rather than 100 in msi? Like doing the cpu in processor state min/max, I got confused cause you put +130 so I thought you were adding.

I can research how to undervolt if that's the term
 
So how do I do the:+130 thing or the less than 100%? Is that by simply sliding the power limit in bar to say 98% rather than 100 in msi?
Notice my units. +130MHz
In MSI afterburner, that's the "Core Clock" offset.

And yes, the "Power Limit" slider is where you can adjust the % to reduce/increase power draw from the 100%/200W stock setting.

Make sure you hit the check mark button to apply your settings!