Will abnormal GPU usage hurt my GPU?

[Frick]

This may have been said, but have the programs you mentiom been updated?

 

[Mussels]

Wait, wouldnt the Nvidia driver power settings be involved here?

Whats this set to?
Windows power plan?

 

[Atticus_Maytrap]

Have you looked at what is actually using the GPU?

actually what i said before is not completely accurate

when i get the high GPU reaction from running the mouse over the MSI Afterburner graphs, the process that takes up the GPU usage is called "Client-Server-Laufzeitprozess" - i have a german version of windows

this translates to Client-Server Runtime Process, csrss.exe. According to the net, this is an active windows process and rather important

for all the other offending programs though - Revit, Autocad etc. - it states that them as using the GPU

This may have been said, but have the programs you mentiom been updated?

yehp, everything up to date

Wait, wouldnt the Nvidia driver power settings be involved here?

View attachment 233451

Whats this set to?
Windows power plan?

i had thought of that, i keep it set to optimal.
For a time early last year i did keep it at Prefer Max Performance as a remedy to some frequent TDR errors i was getting.
The Nvidia Driver subreddit has that as a suggestion for 10 series users - im a 1080Ti

Basically the machine was freezing everytime the GPU clock went from a high power state and wound down to a low power state. Whocrashed identified it the Nvidia driver and a TDR error
Putting on PMP kept the GPU at base clocks permanently and served as a short term remedy for the TDR errors.
Long term fix was a Windows Reinstall with the Media Creation Tool.
TDRs were gone after that and i was able to set the power management of the card back to Optimal.

Windows Power Plan is set to Balanced.

 

[Shou Miko]

@Atticus_Maytrap if you want to see what applications that uses your gpu try GPUShark and click on view--detailed mode and scroll down in the window it should lists all the current active 3d applications.

Link: https://www.geeks3d.com/20211115/gpu-shark-0-25-released/

Remember if your brower use GPU acceleration and other applications a like they show up together with windows exe files.

 

[Frick]

yehp, everything up to date

What I mean was, have the programs been recently updated? You mention specific programs, maybe the problem is with them?

@Atticus_Maytrap if you want to see what applications that uses your gpu try GPUShark and click on view--detailed mode and scroll down in the window it should lists all the current active 3d applications.

Link: https://www.geeks3d.com/20211115/gpu-shark-0-25-released/

Remember if your brower use GPU acceleration and other applications a like they show up together with windows exe files.

You can do that in Task Manager. Information -> Right click, Columns -> GPU.

 

[Atticus_Maytrap]

What I mean was, have the programs been recently updated? You mention specific programs, maybe the problem is with them?




You can do that in Task Manager. Information -> Right click, Columns -> GPU.

yeh boss i knew what you meant, programs in question are all up to date, latest versions of Revit, Afterburner etc.

while it is a GPU reaction to certain programs, i dont think its the programs themselves that are causing it, it's some background thing i cant seem to nail down
my reason for thinking this is:
1) when it wasnt happening the programs were all up to date
2) there's another scenario that causes the clock to ramp up to load levels when i know for a fact it never used to.
Specifically, when i hit win+P and turn on my second and third monitors.
In the past, this would always just cause the clock to go to base levels (1544), but never past it. That's just the way multiple monitors with different refresh rates work, so no problem there.
Whenever this bug is in play though, turning on the extra monitors boosts it up to load levels (1900+) for a period of time before it settles itself and comes back down to the base.

This right here is more or less my short-way of testing if the bug is present or not. If i turn on the monitors and it just goes to base but not past then i know my other work programs like Revit arn't going to be affected - although Autocad seemingly is always affected by it no matter what, and that is very much the least 3D intense program i use.

And this is the curious thing. It came up after a driver update, went away after a roll back, came back again after a fresh driver install and now even a rolldback won't fix it.
Strangest thing........

note: i can now add Samsung magician to the list of programs that get an odd reaction out the GPU that never used to. Does a drive monitoring program like that have any business using the GPU at all?

 

[Shou Miko]

You can do that in Task Manager. Information -> Right click, Columns -> GPU.

GPU only shows usage you need to choose GPU engine to see which is actually a 3D application. (I use Windows 11).

@Atticus_Maytrap what refresh rate is your monitor running at? if above 144Hz try to lower it to 144Hz to see if this fixes your issue it's known bug.

 

[Frick]

GPU only shows usage you need to choose GPU engine to see which is actually a 3D application. (I use Windows 11).

GPU engine also exists.

 

[Atticus_Maytrap]

GPU only shows usage you need to choose GPU engine to see which is actually a 3D application. (I use Windows 11).

@Atticus_Maytrap what refresh rate is your monitor running at? if above 144Hz try to lower it to 144Hz to see if this fixes your issue it's known bug.

when just my main monitor is running its 144Hz, that's its max, but when my 2nd and 3rd and on it gets lowered automatically to 60 to match the others

in any case the bug is there in both scenarios.

Not so much compensating, or spilling over to other apps, but rather, the initial runs were choppy and post-driver update the card manages the loads better, boosting earlier, to make sure it never gets to those dips in performance. The card has numerous ways to predict what clock it should produce but the simple gist of it is: it always wants to clock high enough to provide smooth performance, influenced by its power plan, if applicable. And drivers have been seen to change the metrics the GPU uses to determine that. Cards, for example, wouldn't always recognize 3D loads properly, especially as power saving features got introduced. I think that's what you're seeing and what caused stutter - GPU load zero is either game logic, system interrupt, or a driver bug. The whole affair regarding multi screen and high refresh monitors is of a similar nature: using the correct clock profile for the load it gets.

Note, that lower temps can also indicate the presented loads are handled better, or more efficiently, though a 5C gap kinda tells me you're just using different settings somewhere.

Other than that, I'm just working with whatever you're giving me. Its a simple fact that the GPU is working properly, boosting properly, and responds properly to a presented mix of loads, which is inherent to your setup at all times, so the only sensible idea is to be looking at:

- combinations (at moments in time) of updates/versions of software, drivers, BIOS, Windows. I find it hard to believe you've had a perfect testing bed to compare both situations and we've also seen Windows updates bork GPUs or fix them again, or requiring further TLC from GPU manufacturer.
- known changes in the driver regarding power/boost management

quick side question, what can you tell me about BUS usage?
The internet seems to claim that its "not that important", however with regard to my oddball GPU programs reactions, the BUS usage is more or less congruent with the spikes in GPU usage.
However, i just played a bit of GOTG (again, really fun game, which ran perfectly i might add) and the BUS usage basically doesnt move at all.
This is taken just after quitting out of the game:

those little dips in the usage bar were just switches between cut scenes - which again seem to line up with a slight increase in the BUS usage.

compare that with using Revit - hell not even "using" it, just running the mouse over an empty file:

BUS usage seems to match GPU usage perfectly.

Holy shit.....​

ummmm, i dont know how to say this, but the bug seems to be gone again....?
I literally just pulled the machine out of sleep mode, played about 40 minutes of GOTG (again, top game), opened up Revit and the clock and usage are where they should be.
This is what the usage for Revit looks like now:

this shit is crazy, but i'll take it.
No doubt ill come back tomorrow and it'll be all over the place again.......

 

[seth1911]

For sure the GPU think oh yeah now its 60°c im working my flow, oh no 90°c i have to hot i stop doing that

 

[Atticus_Maytrap]

Not so much compensating, or spilling over to other apps, but rather, the initial runs were choppy and post-driver update the card manages the loads better, boosting earlier, to make sure it never gets to those dips in performance. The card has numerous ways to predict what clock it should produce but the simple gist of it is: it always wants to clock high enough to provide smooth performance, influenced by its power plan, if applicable. And drivers have been seen to change the metrics the GPU uses to determine that. Cards, for example, wouldn't always recognize 3D loads properly, especially as power saving features got introduced. I think that's what you're seeing and what caused stutter - GPU load zero is either game logic, system interrupt, or a driver bug. The whole affair regarding multi screen and high refresh monitors is of a similar nature: using the correct clock profile for the load it gets.

Note, that lower temps can also indicate the presented loads are handled better, or more efficiently, though a 5C gap kinda tells me you're just using different settings somewhere.

Other than that, I'm just working with whatever you're giving me. Its a simple fact that the GPU is working properly, boosting properly, and responds properly to a presented mix of loads, which is inherent to your setup at all times, so the only sensible idea is to be looking at:

- combinations (at moments in time) of updates/versions of software, drivers, BIOS, Windows. I find it hard to believe you've had a perfect testing bed to compare both situations and we've also seen Windows updates bork GPUs or fix them again, or requiring further TLC from GPU manufacturer.
- known changes in the driver regarding power/boost management

yehp, so its back! GPU clock flying up to load levels in reaction to certain programs

i think im getting an idea of what's happening here.
It seemingly went away on its own on Saturday, all good.
Used the machine again on Tuesday evening, still everything fine.
Came back to it just now and there it is again.

The only difference between Tuesday and today is that when i shut down the computer after working with it, i had left my 2nd and 3rd monitors turned on.
So when i came back today and booted up, all 3 monitors were on, whereas on tuesday it was just the main monitor.
When this odd bug is "in play", turning on the extra 2 monitors cause the GPU clock to rise up to load levels, but then this corrects itself after a few minutes. I get that maybe the monitor setup is whats doing this, but it certainly never used to.
There was a 3-4 month period where i left the 3 monitors on permanently, even when gaming, and never saw anything like this
Strangest
fucking
thing...........

 

[Vayra86]

yehp, so its back! GPU clock flying up to load levels in reaction to certain programs

i think im getting an idea of what's happening here.
It seemingly went away on its own on Saturday, all good.
Used the machine again on Tuesday evening, still everything fine.
Came back to it just now and there it is again.

The only difference between Tuesday and today is that when i shut down the computer after working with it, i had left my 2nd and 3rd monitors turned on.
So when i came back today and booted up, all 3 monitors were on, whereas on tuesday it was just the main monitor.
When this odd bug is "in play", turning on the extra 2 monitors cause the GPU clock to rise up to load levels, but then this corrects itself after a few minutes. I get that maybe the monitor setup is whats doing this, but it certainly never used to.
There was a 3-4 month period where i left the 3 monitors on permanently, even when gaming, and never saw anything like this
Strangest
fucking
thing...........

You have the solution
Its multi monitor power management/detection, and combined 2d/3d load clocks on these screens. Give the GPU a complicated situation to work with and you can get these scenarios.

 

[Atticus_Maytrap]

You have the solution
Its multi monitor power management/detection, and combined 2d/3d load clocks on these screens. Give the GPU a complicated situation to work with and you can get these scenarios.

it certainly seems to be the case
however, after the problem seemingly receded on the weekend, when i came back to it on Tuesday evening and booted up (single monitor) i went straight into my multiple monitor setup and worked for hours and everything was just fine.
Moreover, there was a period of 3-4 months where i just left the second and third monitors on constantly and this glitch never reared its head.
......and again, why is it reacting like that only to certain programs?

Its not disimilar to what occured after i switched over to Studio Drivers.
Worked on them for a few hours, put it into sleep mode (still in multiple monitor setup) came back and the glitch was there again

is there a solution to this at all? Do i maybe have to alter the refresh rate of my main monitor?

You have the solution
Its multi monitor power management/detection, and combined 2d/3d load clocks on these screens. Give the GPU a complicated situation to work with and you can get these scenarios.

so i just switched off my extra monitors and then lowered the native refresh rate on my main monitor from 144 down to 60Mhz
rebooted, back in desktop now, and the bug is gone again

good grief!! After all this time i was so sure it was software causing the trouble - the driver updates, the roll backs, the specific program reactions, it all pointed to a software issue
This is incredible.

It can still be an odd reaction from the driver when confronted with the situation through, right?

 

[Vayra86]

it certainly seems to be the case
however, after the problem seemingly receded on the weekend, when i came back to it on Tuesday evening and booted up (single monitor) i went straight into my multiple monitor setup and worked for hours and everything was just fine.
Moreover, there was a period of 3-4 months where i just left the second and third monitors on constantly and this glitch never reared its head.
......and again, why is it reacting like that only to certain programs?

Its not disimilar to what occured after i switched over to Studio Drivers.
Worked on them for a few hours, put it into sleep mode (still in multiple monitor setup) came back and the glitch was there again

is there a solution to this at all? Do i maybe have to alter the refresh rate of my main monitor?


so i just switched off my extra monitors and then lowered the native refresh rate on my main monitor from 144 down to 60Mhz
rebooted, back in desktop now, and the bug is gone again

good grief!! After all this time i was so sure it was software causing the trouble - the driver updates, the roll backs, the specific program reactions, it all pointed to a software issue
This is incredible.

It can still be an odd reaction from the driver when confronted with the situation through, right?

The driver predicts and reads load. I really don't know what the conditions are for one to take precedence over the other, or why that happens.

Its a similar thing in Windows with a lot of services, I know of a lot of things why they might be useful or might happen, but I really stopped trying to figure out exactly what occurs in every odd situation. I've learned from that and Nvidia cards that my humble end-user situation is exactly that. The complexity is so high nowadays, with so many elements each doing their part... At that point it is far more useful to resort to 'rules of thumb' and check whether the situation follows such rules.

 

[Atticus_Maytrap]

The driver predicts and reads load. I really don't know what the conditions are for one to take precedence over the other, or why that happens.

Its a similar thing in Windows with a lot of services, I know of a lot of things why they might be useful or might happen, but I really stopped trying to figure out exactly what occurs in every odd situation. I've learned from that and Nvidia cards that my humble end-user situation is exactly that. The complexity is so high nowadays, with so many elements each doing their part... At that point it is far more useful to resort to 'rules of thumb' and check whether the situation follows such rules.

yeh i that i can understand, i shudder to think how complex a driver coding job would be.
i feel like a broken record saying this, but i only came here looking for answers because i knew it to be odd behaviour and i didnt want anything to break on me right now
if it weren't for my constant eye on MSI Afterburner i probably would never have even noticed, system performance and temp were never an issue.

Thanks again for your help and suggestions my man, they were very much appreciated

 

[Atticus_Maytrap]

The driver predicts and reads load. I really don't know what the conditions are for one to take precedence over the other, or why that happens.

Its a similar thing in Windows with a lot of services, I know of a lot of things why they might be useful or might happen, but I really stopped trying to figure out exactly what occurs in every odd situation. I've learned from that and Nvidia cards that my humble end-user situation is exactly that. The complexity is so high nowadays, with so many elements each doing their part... At that point it is far more useful to resort to 'rules of thumb' and check whether the situation follows such rules.

....just turned the machine on.

Single screen.
Refresh rate lowered to 60Mhz
GPU upclockig bug is back again

god damn it.......

The driver predicts and reads load. I really don't know what the conditions are for one to take precedence over the other, or why that happens.

Its a similar thing in Windows with a lot of services, I know of a lot of things why they might be useful or might happen, but I really stopped trying to figure out exactly what occurs in every odd situation. I've learned from that and Nvidia cards that my humble end-user situation is exactly that. The complexity is so high nowadays, with so many elements each doing their part... At that point it is far more useful to resort to 'rules of thumb' and check whether the situation follows such rules.

.....played a round of BF, restarted, and its gone again

 

[Atticus_Maytrap]

The driver predicts and reads load. I really don't know what the conditions are for one to take precedence over the other, or why that happens.

Its a similar thing in Windows with a lot of services, I know of a lot of things why they might be useful or might happen, but I really stopped trying to figure out exactly what occurs in every odd situation. I've learned from that and Nvidia cards that my humble end-user situation is exactly that. The complexity is so high nowadays, with so many elements each doing their part... At that point it is far more useful to resort to 'rules of thumb' and check whether the situation follows such rules.

Hey there man, just to provide an update, i decided to just get on with it and ignore the upclocking bug if and when it came about.
The temps never got out of control, it wasn't hurting performance - if anything games performed better when it was in effect.

The on-again off-again nature of it persisted though, sometimes i turn the machine on and it's there, sometimes not, although as time went on over the past month it leaned more towards not, happy days.

However, over the weekend i turned the machine on, bug not present, however now im getting GPU usage drops while gaming. Games that were once using 95-99% of the card are now around 70-80%, with some games like Guardians of the Galaxy experiencing regular drops all the way to 0%, producing stuttering and basically making the game unplayable.

This was something i had dealt with before in late December, i DDU'd the driver at the time out and fresh installed the latest (511.23), which fixed that problem but also brought on the excessive GPU usage in my desktop work apps which started this whole thread.

This current GPU usage drop is on the same driver (511.23), it was working absolutely fine up until Saturday. Ive tried DDUing it out and clean installing the latest, but the problem persists.
The only thing i can think of that changed on the system in between was a Steam client update

Any ideas here?

 

[Atticus_Maytrap]

The driver predicts and reads load. I really don't know what the conditions are for one to take precedence over the other, or why that happens.

Its a similar thing in Windows with a lot of services, I know of a lot of things why they might be useful or might happen, but I really stopped trying to figure out exactly what occurs in every odd situation. I've learned from that and Nvidia cards that my humble end-user situation is exactly that. The complexity is so high nowadays, with so many elements each doing their part... At that point it is far more useful to resort to 'rules of thumb' and check whether the situation follows such rules.

Hey there man, i think i just solved this problem once and for all and it was mind numbingly simple

Im 99% sure it was the Geforce Experience app all along.

In the periods where the "bug" was absent, the geforce app wasnt recognising / optimising any apps or games, and hence the GPU clock was behaving normally.

I've been using the machine with driver 511.79 for weeks now, and the clock has been behaving as expected with regard to my work apps, but my games all seem to have taken a hit performance wise.
All during this, the geforce app has not been optimising anything.

I started up the machine today, worked for a ferw hours, then got the alert for the new driver.
Opened GFE, and it went through an app update (not the driver update) and momentary black screen, and then i noticed that all the games and apps were being optimised once again.

My work apps functioned still fine, held off on updating the driver (always give it a few days) put it into sleep mode for lunch, came back and BAM, GPU is upclocking on my work programs again.

I removed the drive where i keep my work apps from GFEs optimisation list, restarted, and everything is back to the way it should be.

This HAS to be it, right? I mean it just HAS to be???

 

[pavle]

Probably; in the future, just nvcleanstall the drivers you're putting on and remove GFexperience and Telemetry from the get go and you'll be fine.

 

[Atticus_Maytrap]

Probably; in the future, just nvcleanstall the drivers you're putting on and remove GFexperience and Telemetry from the get go and you'll be fine.

i've been doing nothing but clean installs since around Feb last year when things started getting iffy - TDR errors and what not.
Before that i always just did express installs the day of, but now im way more careful about it, probably more than i need to be

is GFE that bad of an app? I do see a lot of people on the Nvidia forums say they avoid it like the plague?

 

[eidairaman1]

Must be phoning home (telemetry)

 

[pavle]

...is GFE that bad of an app? I do see a lot of people on the Nvidia forums say they avoid it like the plague?

It seems unnecessary to me to have additional TSR (terminate and stay resident) running in OS session, application profiles are updated in driver releases and are frequent enough to not cause any major concerns.

 

[xtreemchaos]

hey bud your gpu is more acting normal than me, nothing to worry about in my view. my rigg dos the same with multy displays with diff refresh rates.

 

[Mussels]

Hey there man, i think i just solved this problem once and for all and it was mind numbingly simple

Im 99% sure it was the Geforce Experience app all along.

In the periods where the "bug" was absent, the geforce app wasnt recognising / optimising any apps or games, and hence the GPU clock was behaving normally.

I've been using the machine with driver 511.79 for weeks now, and the clock has been behaving as expected with regard to my work apps, but my games all seem to have taken a hit performance wise.
All during this, the geforce app has not been optimising anything.

I started up the machine today, worked for a ferw hours, then got the alert for the new driver.
Opened GFE, and it went through an app update (not the driver update) and momentary black screen, and then i noticed that all the games and apps were being optimised once again.

My work apps functioned still fine, held off on updating the driver (always give it a few days) put it into sleep mode for lunch, came back and BAM, GPU is upclocking on my work programs again.

I removed the drive where i keep my work apps from GFEs optimisation list, restarted, and everything is back to the way it should be.

This HAS to be it, right? I mean it just HAS to be???

thats not what GFE does, all it does is goes and edits the games settings - changing resolutions, lowering details from ultra to medium and so on

I use GFE but not that function of it, i set games how i want them - don't need a program screwing with them afterwards

 

[Atticus_Maytrap]

thats not what GFE does, all it does is goes and edits the games settings - changing resolutions, lowering details from ultra to medium and so on

I use GFE but not that function of it, i set games how i want them - don't need a program screwing with them afterwards

okay, so GFE has no impact whatsoever on GPU usage with regard to games and apps then?

the programs that were being affected by the abnormal GPU usage were showing up in the list there next to my games as being "optimised". Further to that, one of the GFE sub headings shown when i click on the programs in the list is "hardware acceleration", which was turned on.

My work apps have a Hardware acceleration option internal to the programs themselves, which i keep on and should only be making the GPU use about 15-20% at any give time, perfectly normal.
When it's turned off, this drops down to around 2-5% and navigating in the 3D windows becomes very sluggish.

When the GFE HA is also turned on, the GPU usage treats the program like a game and i get load level clocks usage up to around 70-80%, just from even hovering the mouse over some linework, don't even need to be rotating around 3D objects.

As soon as i removed them from the GFE optimisation list, this upclocking "bug" associated with said programs has since gone - and for the period there where the bug was absent by itself, none of my games or apps were being optimised ("info could not be retrieved, please try again later") and i was getting terrible performance issues with a lot of my games

is there really no way that's at all connected?