Did I get Sold a Broken 2080 Ti?

I don't see the issue, you say you get 30 fps on 4k ultra with DLSS on quality, the guy get's 23 average
No he gets 23 with RTX on. I don't have RTX on. He gets 45 fps with RTX OFF and Ultra settings with DLSS
 
Cyberpunk 2077 isn't a good benchmark to use. First of all, the game has known performance issues and has had many patches. I haven't played it in months but I assume I will get different performance now than I did when I played. Second, you're not going to replicate the exact scene which could explain FPS variance.

If you're within a few frame rates in most stable games compared to an average of benchmarks, that is generally normal.
 
Something to think about is he may have a dud for a card. I have had many multiples of the exact same card and some be ched better.than others. Also to note is how was this card run in the past? Heavy load even just gaming can make a winner into a dud. Personally I would be happy with its performance.
 
Something to think about is he may have a dud for a card. I have had many multiples of the exact same card and some be ched better.than others. Also to note is how was this card run in the past? Heavy load even just gaming can make a winner into a dud. Personally I would be happy with its performance.
its performing in the normal range for 3dmark, the card is fine. hes comparing two very different systems and expecting exact same performance. he also cant provide consistent tests of his own. so yeah feeling like someone elses threads now....
 
its performing in the normal range for 3dmark, the card is fine. hes comparing two very different systems and expecting exact same performance. he also cant provide consistent tests of his own. so yeah feeling like someone elses threads now....
Hopefully he comes back after trying some other games. And preferably ones that have an in-game benchmark of some kind so he can look at a more apples to apples comparison.
 
Scores look like they're in the ballpark, so I think the card is ok.

There are a ton of things that can affect performance. Two you could try are turning off hyperthreading (the i7-9700k beat the i9-9900k on one of the benchmarks the OP posted... both are 8 cores and only the i9 has hyperthreading support) and setting your CPU back to stock clocks. OCing doesn't always make everything faster and can occasionally make some things slower. The caveat with both of these is it's probably not worth it even if it makes your Time Spy score go up.
 
Specs in sig..... Shunt Modded 2080ti ftw3 Ultra Gaming, Custom Liquid Cooling.... Need to tune the curve more... Your scores look inline with a stock 2080ti setup
 

Attachments

  • 3dmark.png
    3dmark.png
    697.8 KB · Views: 0
Use two separate power cord to two separate connectors on your PSU

My RX 5700 would never run full speed with his Y type hook up going on and my card is a 220watt ref board and it likes two single 12v rails and the same time .
 
I pulled my video card and dropped a new RTX 3070 in and used both 12v leads from my 2 year old Corsair CX650M and it's been just as stable as the RX 5700 was for the last 2 years .
 
I'm happy to report that after running in game benchmarks of titles in strix 2080 Ti reviews I saw on several sites, the card performs on par or slightly better at stock. It seems synthetic benchmarks like 3DMark and Superposition and others are not a good way to test how if your card is up to snuff or not?

I'm confused tbh.
 
I'm happy to report that after running in game benchmarks of titles in strix 2080 Ti reviews I saw on several sites, the card performs on par or slightly better at stock. It seems synthetic benchmarks like 3DMark and Superposition and others are not a good way to test how if your card is up to snuff or not?

I'm confused tbh.
Synthetic benchmarks are affected by everything in your system. Expecting exactly the same number as someone else is unrealistic.
What they do is let you tune your system and see the effects, positive or negative. They don't work well for comparing system A to system B.
They get used in reviews because they are easy, but you will not replicate the results, and shouldn't expect to be able to.

Your scores were very close to your target, like 0.5% variance. That seems to prove your card is up to snuff. If you want higher scores start overclocking.
 
I can't believe your 3dmark score is like 500 points lower than that one review with a totally different setup! Did you make sure to open your window blinds while running the benchmark? Your room was probably too dark so your 2080 Ti wasn't fully awake yet to run the benchmark properly; maybe pour some starbucks on it and try again.
 
Hi guys, I got a Asus Strix 2080 Ti from reddit user. Paid through Paypal invoice.
My rig:
10850K @ 5Ghz
Crucial Ballistix RGB 3600 MHz
Corsair RM 750W
Windows 10 64 bit 20H2
Strix 2080 Ti stock clocks On Nvida driver 452.06

My scores in benchmarks seem to be below the average:

View attachment 408227

View attachment 408228

In comparison to these:
https://www.pctestbench.com/asus-rog-strix-rtx-2080-ti-review/4/
https://www.guru3d.com/articles_pages/asus_geforce_rtx_2080_ti_rog_strix_preview,22.html
https://www.guru3d.com/articles_pages/asus_geforce_rtx_2080_ti_rog_strix_preview,23.html


And my performance in CBP 2077 is so poor. I get 30 fps with DLSS and Ultra settings on 4K. While these guys get 45 fps and sometimes 60 fps in CBP 2077:



First of all, the performance is in the normal range for the GPU, but we all want to get the best performance we can. That being a Strix, I can see why you would expect it to be on the higher end. Strix should use a binned GPU (higher performing part), but the silicon lottery has its part to play. I am not super familiar with that card, it might just have better cooling and power delivery, which help but do not guarantee any performance as the GPU silicon quality is a bigger factor. If it comes with a factory OC, it's likely a binned GPU.
-make sure the card is seated properly in the PCI-E Slot
-Connect an approved PSU on the proper rails into the duel 8-pin connectors.
-Check bios to see what is being reported and that all bios settings are properly adhered to for a 2080ti class gpu
-Post a screenshot of what GPUz is reporting for your card at stock
-Do a clean uninstall - full wipe of any video drivers currently on your system (ask for help or do a forum search for best practices in ensuring you get a full reset)
-Download and install the most current Nvidia Drivers
-Update the GPU bios from the vendor website only, with the most current build. - Do this ONLY if you find it has a non-standard/mining Bios. Be sure you fully understand the process, as well as make a backup of its' current Bios, also have a spare video card on hand in case you need to flash it back, before proceeding.
Check the above, but also go into the nVidia control panel, and make sure the FPS cap is not enabled. Your LCD's refresh rate can affect performance results if Adaptive sync is enabled like G-sync or Freesync, and an FPS cap (which can improve smoothness and playability, so a good thing, but not while running benchmarks) could be all this is. If you find the FPS cap is enabled in the nVidia control panel, turn it off and save the settings, then reboot, then try again.
Also, did you install Afterburner and do any overlocking? You can bet that those who score on the higher side of that benchmark graph, overclocked their GPU's.

So, a few things you can check, but I think it's a normal result.

As far as Cyberpunk 2077 performance goes, set
Cascaded Shadows Range to Low
Cascaded Shadows Resolution to Medium
Distant Shadow Resolution to Low
and performance will be fine. I have a 3090 and can play at 3440x1440 with Psycho Raytracing with DLSS set to Off, and it's fine. Fast and looks great. Your card being 2xxx, you might need DLSS turned on, but I'd recommend setting it to "Quality".

1642296505946.png
 
Back
Top