(Solved) Gigabyte RTX 2070 Super fans ramping / surging / spiking to 4000 rpm. (Formerly titled: Can you force case fans to ignore CPU utilization?)

[Sizeable_lad]

(This text is now obsolete, read the solution)

 

My case fans like to ramp up to 100% briefly when my cpu has usage spikes. This is common in a lot of modern games and is exceptionally annoying.

 

is there a setting in bios or with a controller that will prevent this? 
 

P.s. my temps are fine, i have a 360ml rad and my 3900x rarely gets up over 60 c even on a hot day.

Edited May 3, 2022 by Sizeable_lad

[TetraSky]

Yes, most bios's power management/hardware monitor option allows for fan curve to be put in. You can decrease the curve to make it run quieter in exchange for more heat.

 

For example, 
https://forum-en.msi.com/index.php?threads/guide-how-to-set-up-a-fan-curve-in-the-bios.369711/

[GuiltySpark_]

This will depend on where these case fans are plugged in, but if we assume for a moment that its to your motherboard, it would be as simple as getting into the fan control section of your BIOS and adjusting the curve to do what you want. Either less aggressive on the ramp up or just staying static. 

[Sizeable_lad]

4 hours ago, GuiltySpark_ said:

This will depend on where these case fans are plugged in, but if we assume for a moment that its to your motherboard, it would be as simple as getting into the fan control section of your BIOS and adjusting the curve to do what you want. Either less aggressive on the ramp up or just staying static. 

Pressed the solution button accidentally, whoops.

 

anyway suppose these fans were connected to an rgb controller, could this have an effect? Such as their rpm not being logged in hwinfo? Because the spikes arent getting logged in hwinfo. 

[Sizeable_lad]

4 hours ago, TetraSky said:

Yes, most bios's power management/hardware monitor option allows for fan curve to be put in. You can decrease the curve to make it run quieter in exchange for more heat.

 

For example, 
https://forum-en.msi.com/index.php?threads/guide-how-to-set-up-a-fan-curve-in-the-bios.369711/

Ive tried implementing fan curves using the bios, and argus monitor, niether have helped, like i said to the other guy the fans are also plugged into a rgb controller (coolermaster) would this affect anything?

[GuiltySpark_]

7 minutes ago, Sizeable_lad said:

Pressed the solution button accidentally, whoops.

 

anyway suppose these fans were connected to an rgb controller, could this have an effect? Such as their rpm not being logged in hwinfo? Because the spikes arent getting logged in hwinfo. 

An RGB controller is for.. RGB, not fan control. Does this device do both? What exactly is it? Help us help you.

[Sizeable_lad]

55 minutes ago, GuiltySpark_ said:

An RGB controller is for.. RGB, not fan control. Does this device do both? What exactly is it? Help us help you.

I really wish i knew more but heres what i do know:

its a coolermaster argb controller that came along with my ml360 rad and 3 fans, i cant remember precisely how it hooks into the motherboard but im pretty sure one lead from the fans goes to power on the mb and the other goes to the rgb controller which connects via sata to the psu.

 

thats all i really know.

 

also i attempted to set the fan step up/step down in my bios settings, to hopefully make the fans ignore short lived spikes in usage and temp but it doesnt appear to have worked, current step up for the cpu fans is 4 seconds, should i go higher?

[Sizeable_lad]

1 hour ago, GuiltySpark_ said:

An RGB controller is for.. RGB, not fan control. Does this device do both? What exactly is it? Help us help you.

Apologies for the confusion but in doing my own research into the matter i think i was being quite blind to what the issue was. It now seems obvious but ill take you through it.

 

fan spikes despite nominal cpu temps

bios settings having no effect

monitors not detecting the spikes.

i foolishly never thought to look at my gpu fans. I believe i have now identified the issue as a defect or problem with my 2070 super from gigabyte as many other forums have reported similar issues. 
 

sorry to waste your time with faulty information.

[Valso]

7 hours ago, Sizeable_lad said:

My case fans like to ramp up to 100% briefly when my cpu has usage spikes. This is common in a lot of modern games and is exceptionally annoying.

 

is there a setting in bios or with a controller that will prevent this? 
 

P.s. my temps are fine, i have a 360ml rad and my 3900x rarely gets up over 60 c even on a hot day.

There is a setting in general but IDK whether it will work in your case or not. Go to "H/W Monitor" and set each case fan to either silent or performance, depending on what you need them to do. I've set mine to "performance" because I don't want any hot air to stay in the case but my fans have never cared about the utilization of the CPU and have always worked the way I've set them - constant speed 24/7/365.

Thinking about it now, you may have plugged your fans to one or a few of the pins for CPU fans. If I'm right and you did so, no wonder they behave according to the CPU behavior. You gotta plug them to pins that read "Case_fan" or something like that, just not "CPU_Fan".

[Sizeable_lad]

PSA BOYS AND GIRLS: Adaptive power mode is your friend.

 

In short if you are someone reading this thread in the future who owns a 20 series rtx card probably manufactured by gigabyte heres the deal:

 

no clue why but if you have your power management mode set to “prefer maximum performance” gigabyte cards (according to me and about 7 other forum posters from across the internet) will randomly ramp up to 4000 rpm. 
 

for me this issue had no bearing on temperature, and odd things like crafting items in mount and blade 2 bannerlord, tabbing out of furmark while it is running, opening the server browser for holdfast: nations at war. Or DARING to play the witcher 3 would trigger it.

 

I have no clue if switching to adaptive power mode is the optimal fix for this issue, or if it occuring in the first place points to a fault in my hardware or setup, but i dont care, my pc no longer randomly screams at me every 2 minutes. 
 

Thanks LTT forums and sorry for wasting your time.

[Mark Kaine]

21 minutes ago, Sizeable_lad said:

no clue why but if you have your power management mode set to “prefer maximum performance” gigabyte cards [...] will randomly ramp up

actually thats a known Nvidia driver bug, but ok.

 

21 minutes ago, Sizeable_lad said:

I have no clue if switching to adaptive power mode is the optimal fix for this issu

There is *no* good reason to ever set a Nvidia gpu to "prefer maximum performance" all it does is heat up the card unnecessarily and therefore *lowers* performance due to how Nvidia gpu "boost" works (higher temps  = lower frequencies) 

 

Nope, you're not the first person to be confused about that, I just thought Id clarify the situation.  : )

 

 

[Sizeable_lad]

21 minutes ago, Mark Kaine said:

actually thats a known Nvidia driver bug, but ok.

 

There is *no* good reason to ever set a Nvidia gpu to "prefer maximum performance" all it does is heat up the card unnecessarily and therefore *lowers* performance due to how Nvidia gpu "boost" works (higher temps  = lower frequencies) 

 

Nope, you're not the first person to be confused about that, I just thought Id clarify the situation.  : )

 

 

Thanks for your info, the problem happens less now but still occurs. Adaptive power has helped but i wish it would stop entirely. If its a known bug are there known fixes or am i forced to endure my fans screaming at me every 2 minutes when playing intensive games.

[Mark Kaine]

6 minutes ago, Sizeable_lad said:

Thanks for your info, the problem happens less now but still occurs. Adaptive power has helped but i wish it would stop entirely. If its a known bug are there known fixes or am i forced to endure my fans screaming at me every 2 minutes when playing intensive games.

no problem,  but i actually thought that its fixed, if its still doing this with adaptive performance (they did rename this recently i think) then thats a bit weird... If you can set it to "normal" then do that... it wont affect performance,  and if you think about it, any usage changes could trigger the card boosting with "adaptive" more than without,  it kinda works as designed (except its useless, a gpu *will* boost as much as needed, if needed anyways) 

 

(They actually did change this a couple of times,  but if you have "normal" set it to that)

 

 

If that still doesn't fix it, DDU your drivers and reinstall  - thats the normal way of fixing this bug anyways  (it really is a bug in fact, just never really gets fixed by Nvidia) 

 

 

 

[Sizeable_lad]

17 minutes ago, Mark Kaine said:

no problem,  but i actually thought that its fixed, if its still doing this with adaptive performance (they did rename this recently i think) then thats a bit weird... If you can set it to "normal" then do that... it wont affect performance,  and if you think about it, any usage changes could trigger the card boosting with "adaptive" more than without,  it kinda works as designed (except its useless, a gpu *will* boost as much as needed, if needed anyways) 

 

(They actually did change this a couple of times,  but if you have "normal" set it to that)

 

 

If that still doesn't fix it, DDU your drivers and reinstall  - thats the normal way of fixing this bug anyways  (it really is a bug in fact, just never really gets fixed by Nvidia) 

 

 

 

Ok thanks, ill give that a shot, no option for “normal” though just adaptive, optimal and prefer performance, ill try optimal first.

 

I’ve seen a few threads out there claiming that RMAing or buying a new card is the only fix which is stressful so i sincerely hope DDU-ing is going to work. 
 

anyway if worst comes to worst i guess i can just sell my left kidney for a 30 series card   ¯\_(ツ)_/¯

[Mark Kaine]

20 minutes ago, Sizeable_lad said:

I’ve seen a few threads out there claiming that RMAing or buying a new card is the only fix which is stressful so i sincerely hope DDU-ing is going to work. 

haha, nah, i mean thats unlikely at least, or maybe it was an actual manufacturing issue... but its far more likely a settings issue,  or this odd bug.

 

Thats also a thing, did you set a custom fancurve? that could also show the same behavior tbh.

 

 

24 minutes ago, Sizeable_lad said:

Ok thanks, ill give that a shot, no option for “normal” though just adaptive, optimal and prefer performance, ill try optimal first.

yeah, its odd, i had that too, now with a different driver its back to "normal" and "high performance " nothing else lol.

 

Really try DDU, in safe mode, that "should" fix it.

 

 

26 minutes ago, Sizeable_lad said:

anyway if worst comes to worst i guess i can just sell my left kidney for a 30 series card

I bought mine with paypal credit, originally card was 700 euros, with paypal its 750... pretty ok imo - especially because  2000 series cards weren't exactly any cheaper (german prices are messed up) ¯\_(ツ)_/¯

 

I really wanted an igame vulcan oc though,  but they cost over 1000 from ali express..  too unsafe and too expensive lmao.

 

 

 

 

[GuiltySpark_]

Its not entirely uncommon for the fan controller on GPU's to just freak out and start going nuts for no apparent reason. If once this is all settled and its continuing, if you can provide graphs of GPU usage, clockspeed, power draw and the fan RPM's, like you can do with HWInfo or MSI Afterburner, this might be easier to diagnose. Again, once you've done a DDU wipe and its still happening, I mean.

 

These are all the graphs I watch for my GPU at all times on another display, for example. Note the two bottom graphs are my GPU's two fans. 

 

[Sizeable_lad]

8 minutes ago, Mark Kaine said:

haha, nah, i mean thats unlikely at least, or maybe it was an actual manufacturing issue... but its far more likely a settings issue,  or this odd bug.

 

Thats also a thing, did you set a custom fancurve? that could also show the same behavior tbh.

 

 

yeah, its odd, i had that too, now with a different driver its back to "normal" and "high performance " nothing else lol.

 

Really try DDU, in safe mode, that "should" fix it.

 

 

I bought mine with paypal credit, originally card was 700 euros, with paypal its 750... pretty ok imo - especially because  2000 series cards weren't exactly any cheaper (german prices are messed up) ¯\_(ツ)_/¯

 

I really wanted an igame vulcan oc though,  but they cost over 1000 from ali express..  too unsafe and too expensive lmao.

 

 

 

 

Ok so i did the ddu and installed the latest drivers. 
 

first i tried furmark as i could consistently trigger the ramping by clicking into the render and out onto my second monitor and back in. 
 

With Adaptive on no ramping occured so that was promising.

 

then i booted up “holdfast nations at war” a game that seemed to trigger the ramping after around 15 minutes of play consistently and to my disappointment my fans began to wail.

 

to give some extra context i bought this gpu a few weeks short of 2 years ago and the ramping problems started this year as far as i can remember.

 

 

 

[Mark Kaine]

21 minutes ago, Sizeable_lad said:

Ok so i did the ddu and installed the latest drivers. 
 

first i tried furmark as i could consistently trigger the ramping by clicking into the render and out onto my second monitor and back in. 
 

With Adaptive on no ramping occured so that was promising.

 

then i booted up “holdfast nations at war” a game that seemed to trigger the ramping after around 15 minutes of play consistently and to my disappointment my fans began to wail.

 

to give some extra context i bought this gpu a few weeks short of 2 years ago and the ramping problems started this year as far as i can remember.

 

 

 

Hmm, i actually heard 2 monitors can trigger this too.

 

So maybe try the same thing except with only 1 monitor connected  - requires a reboot also.

 

And have you tried "optimal" i think thats supposed to replace normal though not sure... 

 

Also any custom fancurves, overclocks etc?

 

 

Also I agree at this point we need temps, usage, etc info... since you said it just started this year maybe the card just needs some maintenance 

 

43 minutes ago, GuiltySpark_ said:

if you can provide graphs of GPU usage, clockspeed, power draw and the fan RPM's, like you can do with HWInfo or MSI Afterburner

 

[Sizeable_lad]

12 minutes ago, Mark Kaine said:

Hmm, i actually heard 2 monitors can trigger this too.

 

So maybe try the same thing except with only 1 monitor connected  - requires a reboot also.

 

And have you tried "optimal" i think thats supposed to replace normal though not sure... 

 

Also any custom fancurves, overclocks etc?

 

 

Also I agree at this point we need temps, usage, etc info... since you said it just started this year maybe the card just needs some maintenance 

 

 

Alrighty i've done some data collection for you guys.

 

Thanks for your help by the way, I really do appreciate it a bunch

 

Anyway, here are some graphs from HWinfo

 

As far as i can tell The spikes in fan speed do not appear to coincide with spikes elsewhere, so i'm stumped, otherwise, is everything else in order? or is something critically wrong here?

[Mark Kaine]

12 minutes ago, Sizeable_lad said:

As far as i can tell The spikes in fan speed do not appear to coincide with spikes elsewhere, so i'm stumped, otherwise, is everything else in order? or is something critically wrong here?

I doubt something is super wrong... but its hard to say... is the card dusty, etc?

 

Also read my previous post i think u should set it to optimal and try just 1 monitor .

what about custom fancurves, overclocks...

 

And can you do a run with GPUZ? Its hard for me to interpret the hwinfo64 stuff... 

 

In GPUZ set it to show max and play or run something for 15 minutes or so (until it spikes i guess)... 

 

 

(also 4000rpm? just odd to me my card maxes out at like ~2500 o.o) 

 

 

 

[Sizeable_lad]

1 minute ago, Mark Kaine said:

I doubt something is super wrong... but its hard to say... is the card dusty, etc?

 

Also read my previous post i think u should set it to optimal and try just 1 monitor .

what about custom fancurves, overclocks...

 

And can you do a run with GPUZ? Its hard for me to interpret the hwinfo64 stuff... 

 

In GPUZ set it to show max and play or run something for 15 minutes or so (until it spikes i guess)... 

 

 

(also 4000rpm? just odd to me my card maxes out at like 2300 o.o) 

 

 

 

The mention of 4000 RPM appears consistently across multiple threads mentioning gigabyte cards specifically.

 

I have not enabled any custom fan curves for my gpu and i set my casefans to "silent" in the bios with some smoothing enabled, though this has little bearing on anything as my Temps appear to be within normal operating limits.

 

As far as overclocking goes... I wouldnt know how, other than enabling xmp for my ram and slightly raising my DRAM Voltage to 1.35 from 1.2  after following a youtube guide,. both of which afaik have nothing to do with the gpu, i havent touched anything.

 

Im pretty sure the card is decently clean as i blew the whole pc out with some compressed air earlier on this year in an inital attempt to fix this ramping issue (as i had assumed my temps were just high or something)

 

Though i suppose if all else fails i could pull the card out and give it a clean.

 

Already set to optimal, have been for a few hours and it appears to be basically identical to adaptive, ill give one screen a crack now and i suppose one more Hardware monitoring program among the dozen ive already installed trying to fix this thing wont hurt.

 

Though, I would be rather dissapointed if i had to give up on having a second monitor, its really quite handy, hell i was thinking about getting a third not too long ago.

 

Oh and the thought occurred to me to reinstall Asus GPU tweak as i uninstalled it months ago due to a nasty bug that would cause all but the most basic of programs to crash on startup, no clue if it will affect anything.

 

Im going to run one test now with both screens using GPU-Z for you, then ill do another with one screen and report.

 

Thanks Again

 

 

[Sizeable_lad]

Ok, Heres test 1, 2 monitors, 4 spikes in a 15 minute session, Im no good at interpereting this data so have at it

 

 

 

I'll now shut down, unplug my second monitor and do another test.

[Mark Kaine]

16 minutes ago, Sizeable_lad said:

The mention of 4000 RPM appears consistently across multiple threads mentioning gigabyte cards specifically.

yeah, other card models will be different,  Im just saying i don't think my cards fans can go this high... p sure they max out at 2500.

 

i mean maybe its a bios bug causing the spikes, but probably it just stands out because it so damn *high* lol, because as said otherwise sometimes erratic fan behavior isnt out of the ordinary (although it obviously shouldn't do that) 

 

Also manufacturer software can indeed cause things like that also... sometimes hard to diagnose,  especially because people are often reluctant to uninstall this stuff. : D

 

You also said you have something else installed... cant exclude that possibility of a bug there either.  I mean its not like they do it on purpose but manufacturer software is usually pretty bad, they are hardware manufacturers after all, software is *not* their fortune, otherwise they would do that! xD

 

 

23 minutes ago, Sizeable_lad said:

Though, I would be rather dissapointed if i had to give up on having a second monitor, its really quite handy, hell i was thinking about getting a third not too long ago.

 

well its just for testing purposes,  and its indeed another Nvidia bug, with similar symptoms,  basically 2 + monitors triggers the card to run at kinda max performance mode, even when it shouldn't...

 

 

25 minutes ago, Sizeable_lad said:

Already set to optimal, have been for a few hours and it appears to be basically identical to adaptive, ill give one screen a crack now and i suppose one more Hardware monitoring program among the dozen ive already installed trying to fix this thing wont hurt.

yeah, ok because its not "normal" lol... I was very happy when my driver reverted to that option,  trust me... (even though i didn't have that bug, it still annoyed the heck out of me...)

 

 

26 minutes ago, Sizeable_lad said:

Im going to run one test now with both screens using GPU-Z for you, then ill do another with one screen and report.

 

Thanks Again

 

ok cool, lets hope it does show something conclusive... it really should,  i can't really imagine fans ramping up that much out of basically nowhere  (not that I doubt it happening) 

 

GPU Z is really good for this stuff, shows p much everything,  powerdraw temps, fans, usage, etc.......

[Mark Kaine]

18 minutes ago, Sizeable_lad said:

Ok, Heres test 1, 2 monitors, 4 spikes in a 15 minute session, Im no good at interpereting this data so have at it

 

 

 

I'll now shut down, unplug my second monitor and do another test.

Oh okay... but now it didn't actually spike that much?

(i get its not easy to replicate this stuff sometimes) 

 

BUT, I already see the issue i think...

 

Thermal throttling... Hotspot 100C... its not surprising at all that the fans would freak out at this point (imho - I dont actually know, because my card runs way cooler so this never happens, thankfully)  imo thats very very likely  (i see some small fan spikes too... although  its not 4000 lol )

 

The card didn't actually throttle..  thats good, but i suppose its super on the edge of doing so.

 

Ah, well, now lets see what the second test does - but to me its pretty clear already... Btw, to fix this will be tricky (new thermal paste, etc) but additionally a custom fancurve can help also...

 

 

I suppose the card is still under warranty?  I would consider RMA in this case, depends if you want to risk fixing it yourself or not... its not super difficult,  but still tricky (Id consider it a 6/10 lolz) 

 

But yeah... this is just what it looks to me rn, the card is super hot... : o 

[Sizeable_lad]

2 minutes ago, Mark Kaine said:

Oh okay... but now it didn't actually spike that much?

(i get its not easy to replicate this stuff sometimes) 

 

BUT, I already see the issue i think...

 

Thermal throttling... Hotspot 100C... its not surprising at all that the fans would freak out at this point (imho - I dont actually know, because card runs way cooler so this never happens, thankfully)  imo thats very very likely  (i see some small fan spikes too... although  its not 4000 lol )

 

The card didn't actually throttle..  thats good, but i suppose its super on the edge of doing so.

 

Ah, well, now lets see what the second test does - but to me its pretty clear already... Btw, to fix this will be tricky (new thermal paste, etc) but additionally a custom fancurve can help also...

 

 

I suppose the card is still under warranty?  I would consider RMA in this case, depends if you want to risk fixing it yourself or not... its not super difficult,  but still tricky (Id consider it a 6/10 lolz) 

 

But yeah... this is just what it looks to me rn, the card is super hot... : o 

Right, so my card is defective or has become so after a few years of use.

 

I dont really like the idea of taking the thing apart and reapplying the thermal paste, is this all that would be involved or would other things need doing?

 

ive heard bad things about gigabytes customer support so rma-ing doesnt sound great either.

 

on another note i currently have the side off of the case and the intake filter removed to improve airflow, its been like this for a few hours so i guess we can rule out my case setup being bad for temps.

 

Oh and the second test didnt change anything. Same spikes at the same points.

 

i suppose i get to choose one of three options:

Buy a new card (probably a 3060 ti) 

RMA

or DIY