Very low Timespy score (RX 6800 XT, Ryzen 3900x)

[Tobyranger]

So, looking around the web, i've noticed people getting bad timespy score with a 6800xt, like 16000 graphics score or even 17000, but mine is next level super duper low

I'm getting a graphics score of 14000.

 

I don't understand.. i even overclocked it and have sam enabled

 

Rx 6800 xt amd midnight black edition

Ryzen 3900x

32 ddr4 3200mhz

Asus tuf x570 plus

1000W corsair 

[Analog]

22 minutes ago, Tobyranger said:

So, looking around the web, i've noticed people getting bad timespy score with a 6800xt, like 16000 graphics score or even 17000, but mine is next level super duper low

I'm getting a graphics score of 14000.

 

I don't understand.. i even overclocked it and have sam enabled

 

Rx 6800 xt amd midnight black edition

Ryzen 3900x

32 ddr4 3200mhz

Asus tuf x570 plus

1000W corsair 

 

If your overclocks are too aggressive, sometimes you will see a regression in scores. This is why you always need to make a reference bench run with the card at stock. Only then you start messing with it and evaluating the scores. if you see a regression this means you have gone too far. 

 

Another thing, the 6800XT reference cards do get hot so overclocking on them is a bit limited because of temperature issues. Try reverting everything back to stock. See how the scores are. Then enable rage mode and see again. 

[Tobyranger]

12 minutes ago, Analog said:

 

If your overclocks are too aggressive, sometimes you will see a regression in scores. This is why you always need to make a reference bench run with the card at stock. Only then you start messing with it and evaluating the scores. if you see a regression this means you have gone too far. 

 

Another thing, the 6800XT reference cards do get hot so overclocking on them is a bit limited because of temperature issues. Try reverting everything back to stock. See how the scores are. Then enable rage mode and see again. 

But I don't understand, my graphics score is low, yet in games I get average or above average fps. I get around 100-120 fps on warzone 4k almost everything max settings, but then i get a trash benchmark score.

 

Userbenchmark mentions my card performs way above (92nd percentile) so im confused as hell

[Mark Kaine]

39 minutes ago, Tobyranger said:

So, looking around the web, i've noticed people getting bad timespy score with a 6800xt, like 16000 graphics score or even 17000, but mine is next level super duper low

I'm getting a graphics score of 14000.

Err, whats your overall score? 

Actually if you can put a link to your results it should be easier to figure this out because  timespy shows a lot of stuff like clocks and temps, which might be the deciding factor here.

[Analog]

Just now, Tobyranger said:

But I don't understand, my graphics score is low, yet in games I get average or above average fps. I get around 100-120 fps on warzone 4k almost everything max settings, but then i get a trash benchmark score.

 

Userbenchmark mentions my card performs way above (92nd percentile) so im confused as hell

 

User benchmark is not a "benchmark." Don't use and don't look at it, it's complete garbage. There should be a special circle in hell reserved for whoever invented it and all of its derivates that have popped out. 

 

[Tobyranger]

2 minutes ago, Mark Kaine said:

Err, whats your overall score? 

Actually if you can put a link to your results it should be easier to figure this out because  timespy shows a lot of stuff like clocks and temps, which might be the deciding factor here.

I will be home in like 9 hours so I will post a link, overall score was like 13000, but clocks and temps seemed more than fine; but youll be the jidge of that when i do get it

[Tobyranger]

5 minutes ago, Mark Kaine said:

Err, whats your overall score? 

Actually if you can put a link to your results it should be easier to figure this out because  timespy shows a lot of stuff like clocks and temps, which might be the deciding factor here.

Nevermind i got it. https://www.3dmark.com/spy/20819063

Thats it

[Tobyranger]

Also, what's annoying is it says my score is bottom 2%, which annoys me a lot, like wtf am I doing wrong

[Mark Kaine]

13 minutes ago, Tobyranger said:

Nevermind i got it. https://www.3dmark.com/spy/20819063

Thats it

Yeah, that doesnt look out of the ordinary *except* the score lol.

 

I cant find the page with the graphs for temps/freqencies however - which was what i was mostly interested in, if anyone can find that, it would be great, I mean i do suspect some kind of throttling somewhere, otherwise that score doesnt make a whole lot of sense.

 

 

[Mark Kaine]

12 minutes ago, Tobyranger said:

Also, what's annoying is it says my score is bottom 2%, which annoys me a lot, like wtf am I doing wrong

Yeah, i mean something *is* wrong with that score, maybe just try again without any OCs, i mean 3dmark can be finnicky about that, also close all other programs and browsers etc, that can actually make a huge difference sometimes .

 

On the other hand at least it seems to work fine in games. You could also try Superposition benchmark, just to have some comparison.

[Tobyranger]

5 minutes ago, Mark Kaine said:

Yeah, that doesnt look out of the ordinary *except* the score lol.

 

I cant find the page with the graphs for temps/freqencies however - which was what i was mostly interested in, if anyone can find that, it would be great, I mean i do suspect some kind of throttling somewhere, otherwise that score doesnt make a whole lot of sense.

 

 

That is exactly my thought.. plus its not like im getting bad fps in games.. like mentionned i actually get very decent fps in 4k max settings on warzone (95-130) 

[tikker]

Perhaps a driver issue? Taking another random score https://www.3dmark.com/spy/19934271 the only difference I see is the driver version. Unless the extra ~400 MHz on the RAM there makes such a big difference, clocks and temps are almost exactly the same.

[Tobyranger]

23 minutes ago, tikker said:

Perhaps a driver issue? Taking another random score https://www.3dmark.com/spy/19934271 the only difference I see is the driver version. Unless the extra ~400 MHz on the RAM there makes such a big difference, clocks and temps are almost exactly the same.

This doesnt make any sense haha fuckk i dont understand.

 

I use a pci riser cable so i run on gen 3 pci, but thay shouldnt impact much at all

[Avocado Diaboli]

I'm not trying to be condescending here, honestly, but why do the scores even matter to you? If you get good performance in actual, real-world applications that compares well to other people running the same hardware, why should the score matter? Does that score improve your enjoyment while playing games? I thought playing games should be fun for their own sake and not because the framerate is below [insert your threshold here]. I mean, there's a reason why most reviewers produce graphs that show a GPU's or CPU's performance in real-world applications and not synthetic benchmarks, because that's what matters.

[Tobyranger]

10 minutes ago, Avocado Diaboli said:

I'm not trying to be condescending here, honestly, but why do the scores even matter to you? If you get good performance in actual, real-world applications that compares well to other people running the same hardware, why should the score matter? Does that score improve your enjoyment while playing games? I thought playing games should be fun for their own sake and not because the framerate is below [insert your threshold here]. I mean, there's a reason why most reviewers produce graphs that show a GPU's or CPU's performance in real-world applications and not synthetic benchmarks, because that's what matters.

Indeed, but what if it can perform a lot better than I think? 

[tikker]

28 minutes ago, Tobyranger said:

This doesnt make any sense haha fuckk i dont understand.

 

I use a pci riser cable so i run on gen 3 pci, but thay shouldnt impact much at all

I'd say (if possible)

• try the older driver
• try without the riser cable

[Avocado Diaboli]

1 minute ago, Tobyranger said:

Indeed, but what if it can perform a lot better than I think? 

You said yourself that you get average or above average framerate in games. I assume that means you've compared framerates to what others are getting with the same hardware to conclude that they're above average. And even if they were below average, if they're above whatever target you intend for them to reach and the games are playable, isn't that enough?

 

That being said, and to not be completely useless here, you mentioned that you've overclocked the card and the result you linked proves that. Have you retried it at stock speeds for everything like @Analog suggested to see if that helps? Also maybe disable SAM and retry again. Or check if you have any other software running while the benchmark is going on that might hog resources.