Question GPU ISSUE - STUTTERING AND TOO HIGH GPU USAGE

[swigon96]

Hello guys,

These are my components:

Intel core i5 12600 KF
RTX 3070 TI OC Gigabyte version (Overclocked in Msi burner)
16 GB ram DDR 4 3600 MHz
Motherboard: MSI Pro Z690- A DDR4
PSU: 850 W GOLD BE QUIET
WINDOWS 11

It feels like GPU is not working as it should:

1. [GPU] Comparing other YT videos with the same GPU - usage of mine GPU is much higher. For example I have stutters and lags while playing Elden Ring on 2K with medium mode.
2. [GPU] Temperature is quite high, it goes drastically from 40 Cels to 82-83 in two or three minutes.
3. I was thinking about UV but once stress testing GPU i've faced weird situation when Voltage was changing all the time from 0.906 to 0.925 - this should not be fixed value?
4. In most of the games i am experiencing stutters even in games that should not have any issues ( Dark Souls 3)

I hope that you will be able to help me with that.

Thanks,
Jakub

 

[swigon96]

+ i've just tested once again and Voltage sometimes is 1.06

 

[whatdoitypehere9937982739]

Does it do the same without overclock? Are you using a driver meant for a previous card/having out of date drivers?

 

[drivinfast247]

Undo the overclock. Use DDU to remove the current drivers. Install the latest drivers.

 

[Phaaze88]

It feels like GPU is not working as it should:

1. [GPU] Comparing other YT videos with the same GPU - usage of mine GPU is much higher. For example I have stutters and lags while playing Elden Ring on 2K with medium mode.
2. [GPU] Temperature is quite high, it goes drastically from 40 Cels to 82-83 in two or three minutes.
3. I was thinking about UV but once stress testing GPU i've faced weird situation when Voltage was changing all the time from 0.906 to 0.925 - this should not be fixed value?
4. In most of the games i am experiencing stutters even in games that should not have any issues ( Dark Souls 3)

1)Could simply be from the differences in software that is, or isn't installed between PCs, or programs/services running in the background - that could be a lot of ground to cover.
Could also be that you need a quality monitor cable.

2)Cooler not being fed enough air, cooler's waste(heat) not being removed fast enough...
Typical range between the gpu core and hot spot under load for gpus out there is ~20C or lower. IF it's higher, then look at a repaste.

3)It's not supposed to be fixed. Nvidia changed all that when they introduced their dynamic boost/overclocking algorithm, literally called Gpu Boost.
The card monitors its own parameters and will make its own adjustments.

4)See #1.

 

[swigon96]

Thank you guys for that. All drivers are fresh and up to date. I just reinstalled the whole Windows to verify if it still going to occur. Without OC - same results.

@Phaaze88 thanks for that. I have a question here. All components in the PC are quite new (2-3 years). However, i am still using my good old Noctua NH-15 to cool the CPU. This cooler is 11 years old actually. I was told that those are the best on the market and there is no point in changing that. My case is Be quiet! Pure Base 500D. This noctua cooler however is extremely big and is taking up a lot of space in this midi tower. Maybe i should change the cooler for something newer/smaller or maybe AIO?
Regarding the quality monitor cable - I am using it quite expensive but maybe it is important to add that I am playing on 65" TV.

Thank you very much,
Jakub

​

 

[swigon96]

Here i've done some furmark tests. I would highly appreciate if you guys can take a look at see if this behavior is normal.

1: 2K: https://www.gpumagick.com/scores/show.php?id=220258
2: 4K: https://www.gpumagick.com/scores/show.php?id=220265

 

[Phaaze88]

Did you mean Pure Base 500DX?

This noctua cooler however is extremely big and is taking up a lot of space in this midi tower. Maybe i should change the cooler for something newer/smaller or maybe AIO?

I wouldn't, but if you've gotten tired of looking at it, that's your call to make.

Regarding the quality monitor cable - I am using it quite expensive but maybe it is important to add that I am playing on 65" TV.

Does the big ol' TV have any kind of built in variable refresh tech(which help reduce the visual effects from stutters/tearing when frame rate dips or spikes too much above or below the screen's refresh rate?
If not, do you have a variable refresh screen - be it g-sync, g-sync compatible, freesync, etc - lying around that you can test with and compare what you feel compared to the TV?

Here i've done some furmark tests. I would highly appreciate if you guys can take a look at see if this behavior is normal.

1: 2K: https://www.gpumagick.com/scores/show.php?id=220258
2: 4K: https://www.gpumagick.com/scores/show.php?id=220265

Can't make heads or tails of that, someone else likely can, but I don't think it's applicable here. Furmark has only like one use: testing the gpu cooler(temperatures).
You've not mentioned what the gap between gpu core and hot spot is like. You need to use either hwinfo or gpu-z to see that.

 

[swigon96]

@Phaaze88
Hey, during game - 82 Cels/ hotSpot 106 Cels. Fans are working extremely loud - 3800 RPM on each. I don't know why.

 

[Phaaze88]

@Phaaze88
Hey, during game - 82 Cels/ hotSpot 106 Cels. Fans are working extremely loud - 3800 RPM on each. I don't know why.

It's happening because the hot spot is reaching 100C+.
The gap is 24C, just above what's typical.
This card likely needs a repaste, but can we see a picture of the fan setup? [Please remove the glass panel beforehand, due to reflections and blurriness.]

 

[swigon96]

Hey,

Thank you very much for your willingness to help appreciate that a lot.

Here is a photo of my case. There are two Noctua fans for the CPU and two bequite on the left and right side.

CASE

In addition, I've run some tests to verify Temperature + Hotpoint. Screen below.

gpu-z

This temperature was captured while playing a game that did not even use 100% GPU and I still had stutters (all the time FPS dropped from 60 to 59 which drives me crazy and makes no sense at all)

Game test

Thank you,
Jakub

 

[Phaaze88]

Here is a photo of my case. There are two Noctua fans for the CPU and two bequite on the left and right side.

CASE

That's the NH-D14, not 15.
Also, the fans' positions are backwards - the round fan is supposed to be in the center, with the little one at the front.

You could definitely do with some more fans - an extra 140mm exhaust up top, above the D14... or, fill up the roof of the case with exhaust fans to try to get gpu waste heat out as fast as possible, but at the cost of warmer cpu temperatures. Balanced vs extreme option, choice is yours...
Can't forget putting a couple more fans at the front too.

In addition, I've run some tests to verify Temperature + Hotpoint. Screen below.

gpu-z

~20C gap in this screenshot. That's normal - combined with the shot of the case interior, those temperatures should go down with some more fans.

This temperature was captured while playing a game that did not even use 100% GPU and I still had stutters (all the time FPS dropped from 60 to 59 which drives me crazy and makes no sense at all)

Game test

The stutters aren't related to gpu temperature or gpu usage. That doesn't narrow it down much - as I posted earlier, that can be caused by software or hardware.
Got a screen with some form of variable refresh that the PC can be plugged into instead?