Question GPU fan going above 100% on its own ?

[Dreadbeard]

im looking for some advice for my gpu - when gaming it gets up to around 82c which is fine - but at the point it hits that my gpu fans go above the 100% mark - 100% is around 3600 rpm - but they will jump up to around 4100 rpm at about 82c and they make a jet engine sound - i have tried everything i can think of, such as using user made curves for the fans, but it doesnt seem to help , it still bumps the fans up to 4100rpm - anyone got any advice?

 

[Dreadbeard]

So add to this, i have found "i think" a way to stop it happening - in msi afterburner under the fan area i have it set to limit 80c / 91% power - is this a bad idea

? is it going to effect the frame rate i get?

 

[Phaaze88]

is this a bad idea? is it going to effect the frame rate i get?

1)No.
2)Test using the games you have installed while comparing the custom profile to the default one, then report back with your findings.

 

[XxDarkMario20xX]

I think doing power limit will just make GPU use more power to get more peformance and think that will make gpu work harder and temps will just go up abit. When i play games normaly set the fan speed to 50-60 and my gpu still runs smoothly and below 80

 

[Dreadbeard]

1)No.
2)Test using the games you have installed while comparing the custom profile to the default one, then report back with your findings.

So far in testing the big thing i have seen is a micro stutter when the temp gets to around 81c it stutters a tad and drops temp down - however the frames still seem to be exactly the same "in the two games ive tested so far" and the fans havent jumped up to 4100rpm so far
I will continue to test and report back if anything changes or seems different

 

[Dreadbeard]

Updating with a weird thing ive noticed - my gpu is stated in all its specs to be 1770mhz boost clock - so how is mine running up to 1900? Could this be whats causing the extra heat? "the framerate here is limited in port i beleive"

 

[Phaaze88]

Updating with a weird thing ive noticed - my gpu is stated in all its specs to be 1770mhz boost clock - so how is mine running up to 1900? Could this be whats causing the extra heat?

Because they overclock themselves dynamically, boosting as high as they can go until they run into one of several walls.
You can see what those walls are with apps like hwinfo or gpu-z.

 

[Dreadbeard]

Because they overclock themselves dynamically, boosting as high as they can go until they run into one of several walls.
You can see what those walls are with apps like hwinfo or gpu-z.

This is a screenshot from hwinfo and gpu-z

 

[Phaaze88]

This is a screenshot from hwinfo and gpu-z

See at the bottom, where the Performance tab is? Power, temperature, R. volt, Op. volt, utility?
Those are the walls I was referring to.

 

[Dreadbeard]

See at the bottom, where the Performance tab is? Power, temperature, R. volt, Op. volt, utility?
Those are the walls I was referring to.

Okay i see, and how do i know the limits as they are just 0?

 

[Phaaze88]

What they mean:
Power: ran into the power limit. You can see your gpu's board power limit in Gpu-Z's advanced tab(click nvidia bios in the drop-down box).

Temperature limit is self explanatory. Temperature limit can be seen in the same tabs mentioned above.

Reliable voltage. The limit of the silicon on your gpu has been reached.

Op voltage. The voltage controller on the card can't go any further.

Utility. Self explanatory.

Okay i see, and how do i know the limits as they are just 0?

Gpu-Z has a real-time graph for performance cap reasons in the sensors tab, and hwinfo can pull up a graph if you double-click on the parameter.
Hwmonitor doesn't let you do either, does it?

 

[Dreadbeard]

What they mean:
Power: ran into the power limit. You can see your gpu's board power limit in Gpu-Z's advanced tab(click nvidia bios in the drop-down box).

Temperature limit is self explanatory. Temperature limit can be seen in the same tabs mentioned above.

Reliable voltage. The limit of the silicon on your gpu has been reached.

Op voltage. The voltage controller on the card can't go any further.

Utility. Self explanatory.


Gpu-Z has a real-time graph for performance cap reasons in the sensors tab, and hwinfo can pull up a graph if you double-click on the parameter.
Hwmonitor doesn't let you do either, does it?

I have no idea tbh, this is all fairly high above my level ^^
I can’t say I understand fully what all this means, but what I have done is set a limit of 81c on my gpu, and since then the fans don’t go above the 3600 rpm 100% now, which is good, I’m just not sure if it’s going to impact the performance

 

[Phaaze88]

Whoops.

I’m just not sure if it’s going to impact the performance

Won't know until you've tried it out; it's all application dependent, so there's no solid yes/no answer to that.

 

[Dreadbeard]

Whoops.


Won't know until you've tried it out; it's all application dependent, so there's no solid yes/no answer to that.

That I understand okay well I’ll keep testing things and see what happens, so far in world of warships and ready or not it seems like I still get normal frames, I think war thunder had a micro stutter but then the game isn’t exactly optimised the best, so I’ll keep trying things and see how it goes

 

[Dreadbeard]

I have a quick follow up to this - was limiting the temperature a good way to go? I’ve heard people say undercoat it, but with my lack of knowledge on how it all works I don’t trust myself to do that - so is what I did a good idea?

 

[Phaaze88]

I have a quick follow up to this - was limiting the temperature a good way to go? I’ve heard people say undercoat it, but with my lack of knowledge on how it all works I don’t trust myself to do that - so is what I did a good idea?

Ehh, I seldom hear about lowering temperature limit, so I can't say for sure.
Undervolting gets tossed about very much, and there's also lowering the power limit, which effectively does the same thing.
The default for the core is 83C, with a max of 90C. I don't see the harm with lowering the max a couple C.

All that said, feeding the gpu cooler with enough air to handle the 200-ish watts of draw that card can do is probably a challenge; the front panel looks quite restrictive.
Instead of lowering the temperature limit, lowering just the power limit should also stop the fans going berserk.

 

[Dreadbeard]

Ehh, I seldom hear about lowering temperature limit, so I can't say for sure.
Undervolting gets tossed about very much, and there's also lowering the power limit, which effectively does the same thing.
The default for the core is 83C, with a max of 90C. I don't see the harm with lowering the max a couple C.

All that said, feeding the gpu cooler with enough air to handle the 200-ish watts of draw that card can do is probably a challenge; the front panel looks quite restrictive.
Instead of lowering the temperature limit, lowering just the power limit should also stop the fans going berserk.

Okay understood I’ll play around a lil bit and see how things go, I have been considering getting a new case, as the one I have has the front panel blocked out with fans on the side instead, I’m sure this doesn’t help the temps