[SOLVED] 2060 - bad performance in performance test, good performance in game

[Mephis39]

• CPU - Ryzen 5 3600
• Mobo - prime b450m micro atx
• RAM - Crucial Ballistix 3200MHz DDR4 (XMP enabled)
• PSU - Corsair CV450 Bronze (450W)
• GPU - RTX 2060

One of the first things I did upon building the PC and installing windows was run UserBenchmark to get a quick, general idea of the health of the components. Everything was fine except my GPU, which scored in the 18th percentile.

I then ran TimeSpy in 3Dmark and scored a graphics score of 6982. Here's a link to the results from one of the Time Spy runs.

I wracked my brains for what might be wrong, and here I've written a list of what I've thought of/checked/fixed:

• I have updated the bios
• I am running the latest drivers
• my motherboard has a PCIe x16 slot, which the card is plugged into.
• My card was bought on Amazon, and is a legit 2060.
• My card runs at 100% usage happily
• My GPU core and memory clocks are correct
• I haven't noticed any throttling and I use the afterburner overlay to keep a constant check

Here's an even stranger thing. I ran some in-game benchmarks, and in Shadow of the Tomb Raider I scored within the margin of error above another guy's benchmark on an equivalent system who I was comparing with. The performance seemed fine to me. It was maxed out, at 2560x1440, no ray tracing, and getting between 60 and 80 fps.

Then I ran a benchmark in Deus Ex Mankind Divided, where I was achieving 10 fewer frames that the one I was comparing against, and in Hitman 2 I seemed to be on par.

Is my issue my power supply? Again, I haven't noticed any throttling, which I have taken to mean my components are not starving for power.

Thanks in advance for any help.

 

[cryoburner]

Do you have a link to the userbenchmark run, and have you tried running the test a second time, without other applications open?

Being in the "18th percentile" might not sound great, but that just means that 82% of other 2060s benchmarked there scored better. The average benchmark score might only be a few percent higher though, and not anything to be particularly concerned about. They report a lot of different numbers at that site, and some of them are not particularly relevant.

For example, the average score for a 2060 is about an 84% there. Being in the "18th percentile" would put your score somewhere around 80% though, so within about 5% of the average performance of a typical 2060. That's not exactly a big difference. As to why there is that small difference, it could potentially be due to something like the temperature the card is running at. Or maybe you just had some other background processes running at the time you ran the benchmark, affecting the score a bit. Being able to see the benchmark page might potentially indicate what it could be.

 

[Mephis39]

Do you have a link to the userbenchmark run, and have you tried running the test a second time, without other applications open?

Being in the "18th percentile" might not sound great, but that just means that 82% of other 2060s benchmarked there scored better. The average benchmark score might only be a few percent higher though, and not anything to be particularly concerned about. They report a lot of different numbers at that site, and some of them are not particularly relevant.

For example, the average score for a 2060 is about an 84% there. Being in the "18th percentile" would put your score somewhere around 80% though, so within about 5% of the average performance of a typical 2060. That's not exactly a big difference. As to why there is that small difference, it could potentially be due to something like the temperature the card is running at. Or maybe you just had some other background processes running at the time you ran the benchmark, affecting the score a bit. Being able to see the benchmark page might potentially indicate what it could be.

Hey, sorry for the late reply. I got a motherboard replacement and was without for a few days.

https://www.userbenchmark.com/UserRun/29594359 Here's a link to my UserBenchmark results, I hope you can see something I can't!

Edit:
I've taken a closer look, and I never noticed until now that mousing over the bars in the bar chart for each component reveals the percentage score. The most common score is 85%. Whereas mine is 79.9% (let's say 80). Would you say that that kind of performance differential does not indicate that something's wrong?

 

[cryoburner]

That result does seem to be more or less within reason. I did notice that most of your results for the DX9 and DX10 graphics tests shown to the right are actually above-average for a 2060, with the exception of the "MRender" benchmark, which is performing lower than what's typical and throwing off the score. I'm not certain what would be affecting that one test though, or even how much of an effect that would have in real-world scenarios. Userbenchmark does have a tendency of over-representing some numbers that are not necessarily all that relevant to real-world performance, though I can't say whether or not that multi-rendering test might be one of them. If you look at the average of the three DX9 tests, your card performs several percent above average for a 2060, while it performs several percent below average in the DX10 tests due to that weaker MRender performance.

It might just be that your particular model of 2060 has those characteristics, as I know there are some different revisions of the standard 2060 6GB design. Or maybe it could be something like a power-delivery or heat issue, as it's possible that test might push the card's power use and heat output more than some of the others. You could always verify your temperatures under load with Afterburner to make sure they're staying within reason. I did notice some other 2060s in the list of results that get similarly lower MRender test results though, so it might just be normal for some versions of the card. So long as performance in games seems fine, it's probably not a big concern.