New tool for VR benchmarking

[karoliens]

I found this on reddit and thought I should share
What I like is that now I could compare GPU performance in VR and now it just confirms that 5700 XT is not doing well in VR compared for e.g. with 1080ti.
I've always found it confusing when people with 1080ti were achieving better performance in VR sims than me with 5700 XT while using the screen it performs about the same.
It looks like VR favours GPUs with larger memory size.
I guess it is time to reconsider swapping to 1080ti then.

 

[MarcG]

It's not the first as he claims, nVidia have FCAT Software that I've been using for years to Benchmark VR!

 

[pattikins]

I found this on reddit and thought I should share
What I like is that now I could compare GPU performance in VR and now it just confirms that 5700 XT is not doing well in VR compared for e.g. with 1080ti.
I've always found it confusing when people with 1080ti were achieving better performance in VR sims than me with 5700 XT while using the screen it performs about the same.
It looks like VR favours GPUs with larger memory size.
I guess it is time to reconsider swapping to 1080ti then.

I ran the test twice & on both occasions my 1080ti performed between 80-85%. No thread on my AMD 3600x cpu was even remotely taxed under the benchmark. I'm therefore skeptical as to what the benchmark proves. Nor does it change my opinion that the vast majority of VR sim games are limited by the CPU - ACC being the exception, but also unplayable in VR.

 

[karoliens]

I ran the test twice & on both occasions my 1080ti performed between 80-85%. No thread on my AMD 3600x cpu was even remotely taxed under the benchmark. I'm therefore skeptical as to what the benchmark proves. Nor does it change my opinion that the vast majority of VR sim games are limited by the CPU - ACC being the exception, but also unplayable in VR.

Would you mind posting the results? I guess the average FPS is around 47? If that so that just proves the fact that 5700 XT is quite bad when it comes to VR gaming as I am getting 32fps.
P.S. CPU is 3600.

 

[pattikins]

 

[karoliens]

59fps makes sense as you're using CV1 and if I check the leaderboard for oculus rift s 1080 ti it's around 47fps as I said before.
So to compare accurately one needs to compare results for the same headset.

 

[pattikins]

Would you mind posting the results? I guess the average FPS is around 47? If that so that just proves the fact that 5700 XT is quite bad when it comes to VR gaming as I am getting 32fps.
P.S. CPU is 3600.

Is the AMD 5700XT running flat out during the benchmark test. Prior to buying my 1080ti, I owned a Vega 56 for six months (sold it during the bitcoin frenzy for the same price as a 1080ti...) & it was fine in VR.

 

[pattikins]

.... if I check the leaderboard for oculus rift s 1080 ti it's around 47fps as I said before.

Where? Anyhow, I don't see any merit in a GPU benchmark that can't deliver 100% gpu performance as a guide between different video cards.

 

[karoliens]

Is the AMD 5700XT running flat out during the benchmark test. Prior to buying my 1080ti, I owned a Vega 56 for six months (sold it during the bitcoin frenzy for the same price as a 1080ti...) & it was fine in VR.

It is running flat out (see attached screenshot). But from what I read previously Vega is even worse that Navi when it comes to VR sim racing. Regular VR games I am pretty sure it's fine.

 

[karoliens]

Where? Anyhow, I don't see any merit in a GPU benchmark that can't deliver 100% gpu performance as a guide between different video cards.

You need to buy DLC for this software to be able to see leaderboard.

 

[Vintagedriver01]

Its only benchmarking on a video and not the game specific, so to me it appears your just creating a ranking. I still think the best benchmark is the specific game itself. Here you can check all impact adjustments have on the frametime, how flat this frametime is and what the influences is on the game view improvements itself.
I always did use FCAT VR for this but after installing a new rig I can not get this software going. Steams fpsVR could be an alternative, as it now can log history.
Another remark the 5700TX can be compared to an GTX1080 or RTX2070 and not the 1080ti

 

[Deleted member 197115]

Interesting tool, kudos to author, but what synthetic benchmark is good for when every title is so different?
FpsVR so far was the best for me measiring real games performance.

 

[karoliens]

Correct me if I'm wrong but in general 5700 XT is around 10% slower than 1080ti/2070S on a pancake. So what surprised me is that using that benchmark I am seeing a difference of 32fps vs 47fps (32%?).

 

[karoliens]

Interesting tool, kudos to author, but what synthetic benchmark is good for when every title is so different?
FpsVR so far was the best for me measiring real games performance.

Fair enough. But let's say I am deciding to buy a new GPU for VR and I am choosing between for e.g. 5700 XT and 1080 ti. How should I know which one would performs better in general for VR if I do not have both of them at hand to run fpsVR. In addition fpsVR works only when launching a game in Steam VR mode. I'm on Oculus Rift S so I assume not running in Oculus VR mode robs performance as well?
Anyway personally I found that tool pretty useful it's a shame too late as I already have 5700 XT. Should have bought used 1080ti instead looking at the results.

 

[Deleted member 197115]

Fair enough. But let's say I am deciding to buy a new GPU for VR and I am choosing between for e.g. 5700 XT and 1080 ti.

1080Ti, you don't need a tool to know that Nvidia beats AMD all the times.

On a more serious note, VR performance is directly proportional pancake, generic flatscreen benchmark will give you the same reference points, there is nothing magical in VR rendering pipe, except that it does it twice (once per eye) and had to do it quicker to finish rendering cycle shared with CPU under 11.1ms for 90 fps.