Question Worried my 2080 Super is underperforming

[Cyber_Akuma]

CPU: 11700K
RAM: 2x32GB DDR4-3600
GPU: EVGA RTX 2080 SUPER Black

I posted before about the temperatures of my GPU and people said it was fine, so I started doing benchmarks on it to see how much of an improvement it is over my previous GPU, and it didn't seem like it was performing as it should.

First was the result in Cinebench 2024:

https://imgur.com/j6X7rQe

View: https://i.imgur.com/j6X7rQe.png


It barely scored higher than a 2070 Super, but I am new to C24 and it's GPU benchmark so I wasn't sure of the granularity of the scores and if that was normal, could not find any other 2080 Supers that were not Ti models to compare.

So then I tried Userbench. Yes I know I know, I don't trust them either, but from my understanding the one thing that can be accurate is comparing results of the same hardware on their site.

And it said that my GPU was under-performing... yet scored a 130?

https://www.userbenchmark.com/UserRun/64822614

https://www.userbenchmark.com/UserRun/64822793

Went through the gauntlet of 3D Mark benchmarks as well, and after that when I compared my scores of the more demanding benchmarks with ones who had the same CPU and GPU as me it was around the bottom of the list for every one. I know there are overclocked cards in those lists, but unless over 90% of the results were from overclocked cards my card was performing worse than the others.

Here's some of the results:
https://www.3dmark.com/fs/30796362
https://www.3dmark.com/fs/30796374
https://www.3dmark.com/fs/30796393
https://www.3dmark.com/spy/42290803
https://www.3dmark.com/spy/42290914
https://www.3dmark.com/pr/2595122
https://www.3dmark.com/sw/885008

When I tried to compare these to reviews of the card that had run some of the same benchmarks, my score was lower by a few hundred.

Any suggestions on what might be wrong or how I can start looking into this?

 

[hotaru.hino]

It would help to know a timeline of its clock speed and temperatures while you're running these tests. Though the more information you can provide, like power consumption and performance limit reason would also be helpful.

You can use HWiNFO, GPU-z, or MSI Afterburner to get all this.

 

[Cyber_Akuma]

What is that feature called to post that info? I am not aware that HWiNFO or Afterburner had any sort of benchmarks, and GPU-Z only has a basic benchmark.

 

[hotaru.hino]

I'm not asking for benchmarks. Those apps pull data from the card like clock speed, temperatures, etc. Sometimes the values they report could lead to something obvious. Like if your card is running hot or the clock speeds aren't where they should be.

 

[Cyber_Akuma]

Huh, I did record video of MSI Afterburner's in-game overlay running through a playthrough of the Atomic Heart demo a few days ago, would that help?

View: https://www.youtube.com/watch?v=SU_hRk9HJLg


I also asked here a while ago about if my card seemed to me running hot, and was told that it appeared to be normal, although it felt like it was running a bit hot to me:

Question - Would it be a good idea to just replace the paste on used RTX 2000 series cards? Or the pads?

Probably vastly over-thinking this but, I have two 2018 era GPUS that I bought used off eBay. One is a... (who came up with this complicated name?) EVGA RTX 2060 Super SC Black ACX 3.0 (model number 08G-P4-3062-KR) and the other is a EVGA 2080 Super Black (Model number 08G-P4-3081-KR)...

forums.tomshardware.com

 

[Phaaze88]

Open Gpu-Z and click on the 'sensors' tab. Click the 3 bar icon at the top-right of the window.
In the new window, click the 'sensors' tab, and next to 'sensor display mode', click 'highest', and then 'ok'.
There should be a bunch of 'MAX' in green font now. Vertically stretch the old window as needed so all gpu parameters are visible.
With Gpu-Z running in the background, play one of the problematic games for at least a few minutes, then click the camera icon at the top-right of the Gpu-Z window and upload it here.

 

[Bob.B]

CPU: 11700K
RAM: 2x32GB DDR4-3600
GPU: EVGA RTX 2080 SUPER Black

I posted before about the temperatures of my GPU and people said it was fine, so I started doing benchmarks on it to see how much of an improvement it is over my previous GPU, and it didn't seem like it was performing as it should.

First was the result in Cinebench 2024:

https://imgur.com/j6X7rQe

View: https://i.imgur.com/j6X7rQe.png


It barely scored higher than a 2070 Super, but I am new to C24 and it's GPU benchmark so I wasn't sure of the granularity of the scores and if that was normal, could not find any other 2080 Supers that were not Ti models to compare.

So then I tried Userbench. Yes I know I know, I don't trust them either, but from my understanding the one thing that can be accurate is comparing results of the same hardware on their site.

And it said that my GPU was under-performing... yet scored a 130?

https://www.userbenchmark.com/UserRun/64822614

https://www.userbenchmark.com/UserRun/64822793

Went through the gauntlet of 3D Mark benchmarks as well, and after that when I compared my scores of the more demanding benchmarks with ones who had the same CPU and GPU as me it was around the bottom of the list for every one. I know there are overclocked cards in those lists, but unless over 90% of the results were from overclocked cards my card was performing worse than the others.

Here's some of the results:
https://www.3dmark.com/fs/30796362
https://www.3dmark.com/fs/30796374
https://www.3dmark.com/fs/30796393
https://www.3dmark.com/spy/42290803
https://www.3dmark.com/spy/42290914
https://www.3dmark.com/pr/2595122
https://www.3dmark.com/sw/885008

When I tried to compare these to reviews of the card that had run some of the same benchmarks, my score was lower by a few hundred.

Any suggestions on what might be wrong or how I can start looking into this?

You might want to update the bios looks like your a little behind also check the chipset driver level.

Gpu driver looks a little behind.


Clean off another 15GB from the OS disk.....too full.

Clean off another 50GB from the wd blue....too full.

 

[Cyber_Akuma]

Open Gpu-Z and click on the 'sensors' tab. Click the 3 bar icon at the top-right of the window.
In the new window, click the 'sensors' tab, and next to 'sensor display mode', click 'highest', and then 'ok'.
There should be a bunch of 'MAX' in green font now. Vertically stretch the old window as needed so all gpu parameters are visible.
With Gpu-Z running in the background, play one of the problematic games for at least a few minutes, then click the camera icon at the top-right of the Gpu-Z window and upload it here.

It's not a problematic game, but benchmarks where my GPU seems to be scoring at the bottom in regards to others who benchmarked the same GPU.

You might want to update the bios looks like your a little behind also check the chipset driver level.

Gpu driver looks a little behind.


Clean off another 15GB from the OS disk.....too full.

Clean off another 50GB from the wd blue....too full.

All the BIOS updates do according to the changelog is enable TPM by default and fix a vulnerability, nothing to do with PCIe or GPU or memory compatibility or anything that would effect GPU performance.

 

[Bob.B]

It's not a problematic game, but benchmarks where my GPU seems to be scoring at the bottom in regards to others who benchmarked the same GPU.



All the BIOS updates do according to the changelog is enable TPM by default and fix a vulnerability, nothing to do with PCIe or GPU or memory compatibility or anything that would effect GPU performance.

There may be more going on than what the change log is showing.
It may have no affect on your problem but it gets it out of the mix just like those other things I mentioned......your call.

 

[Phaaze88]

It's not a problematic game, but benchmarks where my GPU seems to be scoring at the bottom in regards to others who benchmarked the same GPU.

Here's some of the results:
https://www.3dmark.com/fs/30796362
https://www.3dmark.com/fs/30796374
https://www.3dmark.com/fs/30796393
https://www.3dmark.com/spy/42290803
https://www.3dmark.com/spy/42290914
https://www.3dmark.com/pr/2595122
https://www.3dmark.com/sw/885008

With the exception of that one oddity - Firestrike normal graphics vs average graphics - every one of the others is within a few % of its respective average.

You have to remember that there is no default 2080 Super to compare to - yours is being compared among other 2080 Supers, in all kinds of conditions(good and poor). The average has kept changing(not always positively), ever since the person who first uploaded their 2080 Super results.
Same deal with UBM.

Huh... I guess that means these numbers don't mean jack. Unless you're seeing major performance issues in games, consider ignoring all that stuff.

 

[Joseph_138]

Have you tried changing the paste? A lot of issues on GPU's a few years old can be solved by applying new thermal paste, of a better quality than what the manufacturer used.