Question GPU is always idle, even while heavy gaming, is there a way to fix this?

Status
Not open for further replies.

KyaraM

Admirable
Mar 11, 2022
1,494
661
6,790
First, what GPU is that? What PSU do you use?

Second, it's not idle. It goes up to 1995MHz, which is where my 3070Ti usually sits under load, and the average clock rate (last column) states an average clock of 1171.9 MHz, I assume you never resetted the readings so it's a mix between low desktop clocks and actual gaming clocks.. That's not idle. Core load also went up to 100%. When you close the game, the GPU clocks down to somewhere in the 200MHz range since those hich clocks aren't needed on the desktop. It will then give you that reading you highlighted. It has absolutely 0 relevance, and you can actually see that this state is only applied to the current actual idle mide on the desktop, not high loads.. If you want to actually monitor gaming performance, either make an overlay with the help of RTSS, or momentarily tap out of the game to check. Or use tools like the Nvidia GeForce Experience overlay. Also, reset readings when starting to game for more accurate averages under load and check right after leaving the game.

What is your CPU? If it's an older model with low core count/no HC, it can lead to stuttering in games. You should also try to deactivate the Windows Fullscreen Optimization for the game(s) you got issues with. That helped me with similar issues recently. Right-click on the game exe, go to compatibility, and check the box to deactivate. It should improve performance, ironically enough. If not, just activate again. Nothing lost. If not the issue, report back here for further suggestions.
 

ShwaBdudle

Reputable
Aug 10, 2019
125
3
4,595
First, what GPU is that? What PSU do you use?

Second, it's not idle. It goes up to 1995MHz, which is where my 3070Ti usually sits under load, and the average clock rate (last column) states an average clock of 1171.9 MHz, I assume you never resetted the readings so it's a mix between low desktop clocks and actual gaming clocks.. That's not idle. Core load also went up to 100%. When you close the game, the GPU clocks down to somewhere in the 200MHz range since those hich clocks aren't needed on the desktop. It will then give you that reading you highlighted. It has absolutely 0 relevance, and you can actually see that this state is only applied to the current actual idle mide on the desktop, not high loads.. If you want to actually monitor gaming performance, either make an overlay with the help of RTSS, or momentarily tap out of the game to check. Or use tools like the Nvidia GeForce Experience overlay. Also, reset readings when starting to game for more accurate averages under load and check right after leaving the game.

What is your CPU? If it's an older model with low core count/no HC, it can lead to stuttering in games. You should also try to deactivate the Windows Fullscreen Optimization for the game(s) you got issues with. That helped me with similar issues recently. Right-click on the game exe, go to compatibility, and check the box to deactivate. It should improve performance, ironically enough. If not, just activate again. Nothing lost. If not the issue, report back here for further suggestions.
still despite what you have said hwinfo64 still says that the GPU (Performance limit - Utilization permanently on yes) is in idle mode. GPU is RTX 2060 CPU is i7-10700k. PSU is EVGA SuperNOVA 650 G5, 80 PLUS Gold 650W.
 

KyaraM

Admirable
Mar 11, 2022
1,494
661
6,790
still despite what you have said hwinfo64 still says that the GPU (Performance limit - Utilization permanently on yes) is in idle mode. GPU is RTX 2060 CPU is i7-10700k. PSU is EVGA SuperNOVA 650 G5, 80 PLUS Gold 650W.
If I read this article right, your PSU is low-priority B-Tier, so I would replace it some time.
https://forums.tomshardware.com/thr...er-list-rev-14-8-final-update-jul-21.3624094/
What does the 12V rail reading say? Does it dip below 12V? The reading should be above the clock rate reading. What are the GPU and CPU temps, and the resolution? But again, I got heavy doubts about the GPU idling with the readings you posted so far. That's simply not the case. I once again suggesting using an ingame overlay to check. It's also not "permanently on yes", or else I hallucinate that "no" in the second row.

Did you check ingame and after resetting readings, after entering the game and while in the game right now? Else, as stated above, forget that reading, it's not the issue and has no relevance. Did you try disabling the setting yet?
 

ShwaBdudle

Reputable
Aug 10, 2019
125
3
4,595
If I read this article right, your PSU is low-priority B-Tier, so I would replace it some time.
https://forums.tomshardware.com/thr...er-list-rev-14-8-final-update-jul-21.3624094/
What does the 12V rail reading say? Does it dip below 12V? The reading should be above the clock rate reading. What are the GPU and CPU temps, and the resolution? But again, I got heavy doubts about the GPU idling with the readings you posted so far. That's simply not the case. I once again suggesting using an ingame overlay to check. It's also not "permanently on yes", or else I hallucinate that "no" in the second row.

Did you check ingame and after resetting readings, after entering the game and while in the game right now? Else, as stated above, forget that reading, it's not the issue and has no relevance. Did you try disabling the setting yet?
Well it's a decent PSU and I don't see any valid signs that it needs replacing of, and it's only been 2 years. What are the clock rate readings? GPU clock? I will check tomorrow about the 12v rail, but a few days ago I checked (via a graph) every single GPU statistic in hwinfo64 and nothing stood out of the ordinary, there were no sudden drops when the lag occured, nothing, everything was stable yet there still is a problem without the cause. CPU temps don't go above 80 GPU temps don't go above 70. 1080p 144 Hz. I use MSI in-game overlay which for me shows frame rate and frame time. The column is pretty much always Yes, that No appeared maybe for 1 second. Even when gaming and even when sitting on my desktop it still says yes. But maybe it means something else. Also what setting do you mean?(the one you asked if I had tried disabling it)
 

ShwaBdudle

Reputable
Aug 10, 2019
125
3
4,595
I'm talking about this:
https://www.minitool.com/news/disable-fullscreen-optimizations.html

Fullscreen optimization is a standard in Windows 10 and 11, but sometimes, it leads to stuttering and framerate drops so it might be worth a shot. Nothing lost trying something so easy I think, and if it isn't the issue, at least it can be reverted back easily. But it certainly helped me.
This is the hwinfo64 info on the GPU that I collected after 20-30 minutes of game time in borderlands 3:
here it is
on average the CPU temperature doesn't go above 70 degrees. sometimes in some games, the CPU temperature average can be 80-85. but most of the time the CPU reaches 80 degrees only for a second, as can be sometimes seen in the sudden spikes in MSI afterburner.
 

ShwaBdudle

Reputable
Aug 10, 2019
125
3
4,595
CPU temps close to 80C concern me. I try to keep CPU below 65C. Does it happen when accessing HDD?
on average the CPU temperature doesn't go above 70 degrees. sometimes in some games, the CPU temperature average can be 80-85. but most of the time the CPU reaches 80 degrees only for a second, as can be sometimes seen in the sudden spikes in MSI afterburner.
I don't have anything installed on my HDD. I have a brand new Samsung 980 500 GB nvme. windows 11 is installed on it and so are my games
 

KyaraM

Admirable
Mar 11, 2022
1,494
661
6,790
This is the hwinfo64 info on the GPU that I collected after 20-30 minutes of game time in borderlands 3:
here it is
on average the CPU temperature doesn't go above 70 degrees. sometimes in some games, the CPU temperature average can be 80-85. but most of the time the CPU reaches 80 degrees only for a second, as can be sometimes seen in the sudden spikes in MSI afterburner.
That GPU hotspot temperature in your screenshot got awfully hot, holy <Mod Edit>! Over 100°C o.o
I never get past 80, and 80 is quite high already for my card. Most of the time, I keep it around 10°C higher than the core temperature. You got a discrepancy of 40°C, which feels very high to me. I have to admit, I never looked much into the RTX 20 series since I had a great card with the 1070 already and I think nobody anticipated the current situation. So I can't tell you exactly if that's normal for this generation. It doesn't feel normal to me, though. Also, I'm aware that GDDR6X runs hotter than regular VRAM. However, on my card, it's very similar to hotspot temperatures and while the sensor is unfortunately missing in your card (probs all previous generations since my 1070 doesn't have it, either), it should be roughly the same. Also, my GTX 1070 has similar offsets between core and hotspot temperatures, so 70°C-78°C max on hotspot, and that card was running almost every day for over 5 years with only basic cleaning. I would honestly look into this. If your VRAM is overheating, that might be the reason for your stuttering. Heck, even that hotspot temperature...
 
Last edited by a moderator:

ShwaBdudle

Reputable
Aug 10, 2019
125
3
4,595
That GPU hotspot temperature in your screenshot got awfully hot, holy <Mod Edit>! Over 100°C o.o
I never get past 80, and 80 is quite high already for my card. Most of the time, I keep it around 10°C higher than the core temperature. You got a discrepancy of 40°C, which feels very high to me. I have to admit, I never looked much into the RTX 20 series since I had a great card with the 1070 already and I think nobody anticipated the current situation. So I can't tell you exactly if that's normal for this generation. It doesn't feel normal to me, though. Also, I'm aware that GDDR6X runs hotter than regular VRAM. However, on my card, it's very similar to hotspot temperatures and while the sensor is unfortunately missing in your card (probs all previous generations since my 1070 doesn't have it, either), it should be roughly the same. Also, my GTX 1070 has similar offsets between core and hotspot temperatures, so 70°C-78°C max on hotspot, and that card was running almost every day for over 5 years with only basic cleaning. I would honestly look into this. If your VRAM is overheating, that might be the reason for your stuttering. Heck, even that hotspot temperature...
well, the lag occurs even when the hotspot temperature is below 80 degrees. plus as it says it's unconfirmed temp. when my card's temperature reaches 60 degrees fans go to around 100%. there's nothing more that I can do, the GPU is not thermal throttling, I cant buy a new GPU, and plus as I said above the lag occurs even when the GPUs hotspot temp is below 80. is there any way I can check for sure what the VRAM temp is?
 

ShwaBdudle

Reputable
Aug 10, 2019
125
3
4,595
That GPU hotspot temperature in your screenshot got awfully hot, holy <Mod Edit>! Over 100°C o.o
I never get past 80, and 80 is quite high already for my card. Most of the time, I keep it around 10°C higher than the core temperature. You got a discrepancy of 40°C, which feels very high to me. I have to admit, I never looked much into the RTX 20 series since I had a great card with the 1070 already and I think nobody anticipated the current situation. So I can't tell you exactly if that's normal for this generation. It doesn't feel normal to me, though. Also, I'm aware that GDDR6X runs hotter than regular VRAM. However, on my card, it's very similar to hotspot temperatures and while the sensor is unfortunately missing in your card (probs all previous generations since my 1070 doesn't have it, either), it should be roughly the same. Also, my GTX 1070 has similar offsets between core and hotspot temperatures, so 70°C-78°C max on hotspot, and that card was running almost every day for over 5 years with only basic cleaning. I would honestly look into this. If your VRAM is overheating, that might be the reason for your stuttering. Heck, even that hotspot temperature...
in this thread posted on Reddit, a user has the same hotspot temperatures as I do. most of the people say not to worry about it
 

ShwaBdudle

Reputable
Aug 10, 2019
125
3
4,595
We need to figure out what's going on when the lag happens. High temps on the CPU, GPU, board/VRMs, etc. Disk access? Internet drop? etc.

Looking at your pic I'm not sure how much I believe that. After gaming for 30min your GPU is at sub 30c?
around 10 minutes have passed after the gaming session before I took the picture, the GPU probably has cooled by then. before when my games were installed on my old HDD I thought that this was because of disk access (evident by the spikes on the disk graph in task manager) but turns out it wasn't the cause, at least not the only cause. whenever I look at resource monitor or task manager the graphs are always pretty smooth and there are no spikes in the graphs when the stuttering occurs.
 

Satan-IR

Splendid
Ambassador
Have you tried leaving GPU-Z running and checking min/max values for GPU clock, memory clock and GPU load and slot and 8-pin voltagse power etc after gaming for a while?

That 80C for CPU is too high. Also how can a gamig GPU be sitting at 28C after gaming? They're usually higher than that at idle considering they ambeint is over around 25C (room temperature) and not in a freezer. Is this immediately after gaming session ends? If there's a gap and GPU fan going at almost 1400RPM those temps could be right.

Also hot spot getting around 104C is too high. Assuming the reading is from the card sesors and it's somehow accurate the 12V from the PCIE slot dropped (at least once - we don't know how long) from 12.109 to 11.913, This I think is still inside ATX regulation but if it goes any lower momnetarily while you're not minitoring that can cause instability too.

Stettring happens without any FPS drop? Just happens in games? No other stuttering? Video playback is always OK?
 

ShwaBdudle

Reputable
Aug 10, 2019
125
3
4,595
Have you tried leaving GPU-Z running and checking min/max values for GPU clock, memory clock and GPU load and slot and 8-pin voltagse power etc after gaming for a while?

That 80C for CPU is too high. Also how can a gamig GPU be sitting at 28C after gaming? They're usually higher than that at idle considering they ambeint is over around 25C (room temperature) and not in a freezer. Is this immediately after gaming session ends? If there's a gap and GPU fan going at almost 1400RPM those temps could be right.

Also hot spot getting around 104C is too high. Assuming the reading is from the card sesors and it's somehow accurate the 12V from the PCIE slot dropped (at least once - we don't know how long) from 12.109 to 11.913, This I think is still inside ATX regulation but if it goes any lower momnetarily while you're not minitoring that can cause instability too.

Stettring happens without any FPS drop? Just happens in games? No other stuttering? Video playback is always OK?
I have not tried running GPU-Z in the background. my GPU fans are at 40% when GPU is idle which helps it get cooled down quicker after a gaming session. but shouldn't HWINFO64 catch momentary drops and put it in the minimum column? whenever I start my PC I always open hwinfo64. stuttering happens with minimal fps drop, just happens in games. but what I did notice is that when the lag occurs everything else lags as well(i.e discord voice playback).
 

Satan-IR

Splendid
Ambassador
As 4745454b said above, you need to expand the scope of your focus. It might be internet lag if it's online play, it might be storage working hard and causing a moments lag in loading stuff.

Anyhow those numbers in the HWINFO somehow don't add up. GPU total board input is 185W. What's that about? Off the top of my head the other power draws don't add up I think.

You can do test with GPU-Z and see if the results (temps, power draw) are consistent with HWINFO.
 

ShwaBdudle

Reputable
Aug 10, 2019
125
3
4,595
As 4745454b said above, you need to expand the scope of your focus. It might be internet lag if it's online play, it might be storage working hard and causing a moments lag in loading stuff.

Anyhow those numbers in the HWINFO somehow don't add up. GPU total board input is 185W. What's that about? Off the top of my head the other power draws don't add up I think.
it's a single-player game running on a brand new Samsung 980 nvme. what do you mean the numbers in hwinfo64 don't add up?
 

Satan-IR

Splendid
Ambassador
it's a single-player game running on a brand new Samsung 980 nvme. what do you mean the numbers in hwinfo64 don't add up?
I mean maybe at idle it shouldn't draw what it is drawing? Hav you checked what they draw at idle in benchmarks? Also what is nvidia control panel power settings? Is it as 'Optimal' or 'Adaptive' or 'Prefer Maximum Performance'?

Also there might be something else overwehlsmig the storage with load at certan moments that causes it to lag for a short while. Driver bug or firmware issue that needs a patch. Again you should look at storge performance too during lags/stutters to see if there's something going on.
 

ShwaBdudle

Reputable
Aug 10, 2019
125
3
4,595
I mean maybe at idle it shouldn't draw what it is drawing? Hav you checked what they draw at idle in benchmarks? Also what is nvidia control panel power settings? Is it as 'Optimal' or 'Adaptive' or 'Prefer Maximum Performance'?

Also there might be something else overwehlsmig the storage with load at certan moments that causes it to lag for a short while. Driver bug or firmware issue that needs a patch. Again you should look at storge performance too during lags/stutters to see if there's something going on.
power settings are set to optimal. maybe the GPU is drawing that much power because the fans are constantly spinning?
 

Satan-IR

Splendid
Ambassador
Don't think the card would need to draw almost18W to make fans turn, even at full speed. Unless at idle with fans at 0 RPM it draws around 13-14W.

Accurately calculating it would depens on number of fans and their maximum amperage and all but I don't think the difference between card at idle with fan at 0RPM and full speed would be more than 4-6 watts. I've not tested this myself or seen any figures for it and of course it would vary depending on the card and number and max speed of fans etc.

So IF your card fans at almost 1400RPM are at around 30-40% duty cycle I don't think they should draw more than 1-2W than when at 0 RPM. This is all just guesses as I don't know the exact specs of the fans and their numbers.
 
Last edited:

4745454b

Titan
Moderator
It honestly depends on the fan. I've seen some that claim to pull 5W, and others that claim 20W. I've just never seen a GPU temp sub 30C. So I suspect the program isn't reading sensors correctly. Might be only the temp ones but not sure. Unless it's way cold in your room.
 
  • Like
Reactions: Satan-IR
Status
Not open for further replies.