[SOLVED] Low GPU usage on RTX 3090 on 1440P

[Ulzgan]

Hello all,

I need your help to undertand why my GPU usage is lower on same games at 1440p.

For example, on Hunt:Showdown at 1440p, my GPU usage is only 70%/75% it doesn't reach 99%.
I saw some youtube videos of others 3090 owners playing Hunt:Showdown and they have 100% GPU usage on 1440p.

My PC specs are the following:
Ryzen 9 3900x
32GB RAm 3200Mhz
ASUS ROG STRIX X570-E Gaming
250GB SDD Sata
250GB SDD Sata
1Tb SSD M.2 2280
512GB SSD M.2 2280
RTX 3090 MSI gaming X trio
Corsair HX850i 80 plus Platinium

On Control I'm able to get 97% of GPU Usage but if turn on DLSS the GPU usage goes down to 60%/70%.

And this behaviour doesn't seem ok.

On baldurs gate I have the same issue, on 1440p the GPU usage is low, 50%60% and this doesn't seem right.

If I put it on 4K, the GPU usage goes up to 97% / 98%. And this behavior seems to be some kind of bottleneck on 1440p...

My score on Time spy is 19777, but 3D guru was able to achive 20500 with the same card, but they did it with the old drivers,
If I try the 4K benchmarks, my results are equal or superior to some of the other 3090 cards.

Could you please help me undertand this behaviour?

Thank you

 

[rgd1101]

and the cpu usage?

 

[sizzling]

Are you using a frame limiter or V-Sync?

 

[Ulzgan]

and the cpu usage?

Also Low, 20% 30/ max

 

[Ulzgan]

Are you using a frame limiter or V-Sync?

No V-Sync used or G-Sync

 

[Phaaze88]

Can you link a userbenchmark run?

 

[Ulzgan]

Can you link a userbenchmark run?

Of course.

UserBenchmarks:
CPU: AMD Ryzen 9 3900X - 93.7%
GPU: Nvidia RTX 3090 - 167.8%
SSD: WD Black SN750 NVMe PCIe M.2 1TB (2019) - 309.9%
SSD: Samsung 850 Evo 250GB - 106.8%
SSD: Samsung 850 Evo 250GB - 105.9%
SSD: Samsung 970 Evo NVMe PCIe M.2 500GB - 309.3%
HDD: WD Black 1TB (2013) - 80.5%
RAM: G.SKILL F4 DDR4 3200 C16 4x8GB - 109.8%
MBD: Asus ROG STRIX X570-E GAMING

I got the following error on userbenchmarks
Benchmarks - missing GPU

 

[Ulzgan]

Of course.

UserBenchmarks:
CPU: AMD Ryzen 9 3900X - 93.7%
GPU: Nvidia RTX 3090 - 167.8%
SSD: WD Black SN750 NVMe PCIe M.2 1TB (2019) - 309.9%
SSD: Samsung 850 Evo 250GB - 106.8%
SSD: Samsung 850 Evo 250GB - 105.9%
SSD: Samsung 970 Evo NVMe PCIe M.2 500GB - 309.3%
HDD: WD Black 1TB (2013) - 80.5%
RAM: G.SKILL F4 DDR4 3200 C16 4x8GB - 109.8%
MBD: Asus ROG STRIX X570-E GAMING

I got the following error on userbenchmarks
Benchmarks - missing GPU

More info:
Please enter the exact model for your Nvidia RTX 3090.
e.g. GV-N1060G1GAMING-6GD, Gigabyte G1 Gaming
rtx 3090 msi gaming x trio

Nvidia RTX 3090-€1,290
2,139 User benchmarks, average bench 212%
MSI(1462 3884) ≥ 4GB; rtx 3090 msi gaming x trio
CLim: 2115 MHz, MLim: 4875 MHz, Ram: 24GB, Driver: 456.71
Relative performance n/a - benchmarks incomplete

 

[Ulzgan]

More info:
Please enter the exact model for your Nvidia RTX 3090.
e.g. GV-N1060G1GAMING-6GD, Gigabyte G1 Gaming
rtx 3090 msi gaming x trio

Nvidia RTX 3090-€1,290
2,139 User benchmarks, average bench 212%
MSI(1462 3884) ≥ 4GB; rtx 3090 msi gaming x trio
CLim: 2115 MHz, MLim: 4875 MHz, Ram: 24GB, Driver: 456.71
Relative performance n/a - benchmarks incomplete

For some reason it's failing the last benchmark for the GPU, the 3D DX11

 

[Phaaze88]

UserBenchmarks:
CPU: AMD Ryzen 9 3900X - 93.7%
GPU: Nvidia RTX 3090 - 167.8%
SSD: WD Black SN750 NVMe PCIe M.2 1TB (2019) - 309.9%
SSD: Samsung 850 Evo 250GB - 106.8%
SSD: Samsung 850 Evo 250GB - 105.9%
SSD: Samsung 970 Evo NVMe PCIe M.2 500GB - 309.3%
HDD: WD Black 1TB (2013) - 80.5%
RAM: G.SKILL F4 DDR4 3200 C16 4x8GB - 109.8%
MBD: Asus ROG STRIX X570-E GAMING

The actual link, please. Those numbers don't mean anything by themselves.

 

[Ulzgan]

The actual link, please. Those numbers don't mean anything by themselves.

I'm sorry, link below:
https://www.userbenchmark.com/UserRun/34701251

 

[Ulzgan]

The actual link, please. Those numbers don't mean anything by themselves.

I was able to fix the userbenchmark issue.
Below the new resut
https://www.userbenchmark.com/UserRun/34702822

 

[Phaaze88]

How did you fix the missing gpu result from earlier?

 

[Ulzgan]

How did you fix the missing gpu result from earlier?

I uninstalled Sonic radar sw.
After that the userbenchmark didn't show any error.

I saw this tip on another forum and it worked for me

 

[Phaaze88]

Sonic radar

What other needless software apps have you - perhaps - installed?
That's probably what's separating your system from others you've seen from the videos.
Userbenchmark isn't a reliable performance benchmark, but it is useful in aiding to find possible conflicts... and going by the most recent link, there's nothing problematic sticking its ugly head out.

It's most likely some 3rd party software putting a damper on things. The fewer of these you rely on, the smoother your PC will run overall.
They all add up, and some don't get along with others. Some examples:
A)Having too many security apps, or more than one of the same.
Windows Defender is a pretty good AV by itself, but having an anti-malware, or anti-spyware isn't a bad idea, IMO, because not everything classifies as a virus. Though installing other AVs alongside Defender is pretty redundant.

B)Bloatware. This one's pretty broad, but they're all basically apps the user doesn't really need.
Asus AI Suite
Msi Dragon Center
Corsair iCUE: too bad the jerks made it mandatory to use with their hybrid coolers.
NZXT CAM: ditto with iCUE
Steve at Gamers' Nexus did a review of sorts; a bloatware benchmark. The performance impact from 1 or 2, while small, still adds up.
Some of these things constantly run in the background, or accessing the company's servers for whatever reason, and all it costs is some of your own resources - internet, cpu, memory, etc.

Let's not forget some browser extensions that also want to run in the background.
I'm reminded of a very recent Angry Joe news video where they announced bloody Honey as a sponsor... UGH.
"It's free!"
[No it isn't. It has to run in the background, use my system resources, and likely sends sells my shopping info to at least one advertising company so they can annoy me further with ads I don't give 2 cents about. Yeah... 'free'.]

"It can find good deals!"
[Maybe, but I lose some peace of mind in the process. Because of the demographic I fall in - male, between 30-40 years - they constantly cram beer/alcohol ads down my throat, and I can't change a thing about it.
I was a light drinker, plus the medication I'm on makes that crap a no-no, so buzz off already! Can't you get the message when I'm constantly trying to spam the skip button on them? 'Ads catered to you' also doesn't work correctly - and companies wonder why people use ad blockers...
Dang it, I started ranting...]

 

[Ulzgan]

What other needless software apps have you - perhaps - installed?
That's probably what's separating your system from others you've seen from the videos.
Userbenchmark isn't a reliable performance benchmark, but it is useful in aiding to find possible conflicts... and going by the most recent link, there's nothing problematic sticking its ugly head out.

It's most likely some 3rd party software putting a damper on things. The fewer of these you rely on, the smoother your PC will run overall.
They all add up, and some don't get along with others. Some examples:
A)Having too many security apps, or more than one of the same.
Windows Defender is a pretty good AV by itself, but having an anti-malware, or anti-spyware isn't a bad idea, IMO, because not everything classifies as a virus. Though installing other AVs alongside Defender is pretty redundant.

B)Bloatware. This one's pretty broad, but they're all basically apps the user doesn't really need.
Asus AI Suite
Msi Dragon Center
Corsair iCUE: too bad the jerks made it mandatory to use with their hybrid coolers.
NZXT CAM: ditto with iCUE
Steve at Gamers' Nexus did a review of sorts; a bloatware benchmark. The performance impact from 1 or 2, while small, still adds up.
Some of these things constantly run in the background, or accessing the company's servers for whatever reason, and all it costs is some of your own resources - internet, cpu, memory, etc.

Let's not forget some browser extensions that also want to run in the background.
I'm reminded of a very recent Angry Joe news video where they announced bloody Honey as a sponsor... UGH.
"It's free!"
[No it isn't. It has to run in the background, use my system resources, and likely sends sells my shopping info to at least one advertising company so they can annoy me further with ads I don't give 2 cents about. Yeah... 'free'.]

"It can find good deals!"
[Maybe, but I lose some peace of mind in the process. Because of the demographic I fall in - male, between 30-40 years - they constantly cram beer/alcohol ads down my throat, and I can't change a thing about it.
I was a light drinker, plus the medication I'm on makes that crap a no-no, so buzz off already! Can't you get the message when I'm constantly trying to spam the skip button on them? 'Ads catered to you' also doesn't work correctly - and companies wonder why people use ad blockers...
Dang it, I started ranting...]

Hello ,

Thank you for your reply.

But I don't think it's a program that is causing this or maybe it is... The only programs that I have are a vpn, malwarebytes premium, and NZXT CAM.

Today I enter Hunt:Showdown all seeting on Max, 1440p, and for my surprise the GPU load was 92/95% and my FPS were amazing, 220/240 fps.

But then, the same behaviours started, and the GPU load started to fluctuate between 70% and 90%.
Shouldn't GPU load be steady at 95%/100%?

When it's on 70%, my FPS are 160/170, and when it reach 90% of GPU load it's 220/240, even high then the youtuber that I saw on a I9 10900k oc 5.1Ghz.

And this is the behaviour that I don't understand and wants help to solve it .


Because I'm losing performance due to this .

Thank you for your undertanding.

 

[Phaaze88]

The only programs that I have are a vpn, malwarebytes premium, and NZXT CAM.

VPNs: can slow down your internet - especially when the server is under heavy load. For games that require constant access, this can be a potential problem.


Malwarebytes Premium: I use it

NZXT CAM

Can be observed communicating with NZXT's servers on your internet connection quite often for whatever reason...

Now that you know the above, you should know which games suffer the hardest.
Control shouldn't require a steady connection, so performance should be fine.
But Hunt and Baldurs do require a steady connection, don't they?
Compare with other games you may have on whether or not the issue really is your internet.

Shouldn't GPU load be steady at 95%/100%?

No.
It's normal for some areas to be easier to render than others.

 

[Solidjake]

How did you fix the missing gpu result from earlier?

When I did the benchmark test a few months it wouldn't be complete as the OP had. I had to turn off MSI Afterburner since the rivatool? was preventing the test from being completed .

 

[Phaaze88]

When I did the benchmark test a few months it wouldn't be complete as the OP had. I had to turn off MSI Afterburner since the rivatool? was preventing the test from being completed .

I've heard about Rivatuner causing issues before.
It didn't sit right with me that it had to 'stay on'. Never installed it again.

Good call, though.