ValvE's VR performance test - is your PC ready for VR?

[dimitriianghelov]

3 minutes ago, Humbug said:

It cannot be marketing for Nvidia. It's not judging according to the name of the GPU. It's actually running the sequence and then giving the result.

So if by some miracle a 750ti aces it then it will say that that too is ready for VR. Often two people with similar GPUs will get different experiences due to drivers versions or something in the rest of their system etc.

one person with the same gpu will get different results

[VinZie]

21 hours ago, zMeul said:

source: http://store.steampowered.com/app/323910/

via: http://www.kitguru.net/desktop-pc/gaming-rig/matthew-wilson/valve-launches-steamvr-portal-performance-test-demo/

 

 

I ran the test on my rusty old Q9550 with a GTX970 and amazingly, it's VR ready 

so yeah ,, this test is absolutely flawed - unless ValvE will "force" every VR title launched on their platform to use the Source⁽²⁾ engine

  Reveal hidden contents

 

ps: to run the test on a multi-GPU system, add -multigpu to the command line

When I went through the files and the directories I found some thing leading to Half Life 3, and Left for Dead 3... I recommend you all look through the directories and check out what is in there.

[Humbug]

Not sure you guys noticed but one interesting thing about this sequence is that there are no graphics settings. The source 2 engine is dynamically varying the graphical fidelity in order to target 90fps. i.e. if the previous frame took more than ~11ms to render then it starts dropping the graphics in order to maintain 90fps. The result is less framerate variations but more graphical variations (if your system cannot handle it). This is the opposite of what we usually have (wild framerate variations and fixed graphical target).

https://www.reddit.com/r/oculus/comments/471lnh/source_2_engine_in_steamvr_performance_tool/

 

Also in order to have a closer look at the assets I forced it out of stereoscopic mode and into my native resolution fullscreen (1080p). The textures seem to be very high res and there is a lot of geometry detail in the scene. Still when you take it out of stereoscopy it runs at almost 300fps because it's a small room in a good engine and has been designed to run at high framerate.

 

 

 

 

 

 

[MEC-777]

All set.

 

[Humbug]

3 hours ago, Nisco3000 said:

Yeah look at this huge list. https://www.reddit.com/r/Vive/wiki/systems

apart from occasional outliers the chips seem to be falling as expected. 980ti is king of the hill, followed by Fury X and so on...

 

the GTX 780 I noticed seems to be underperforming. Only one of them has scraped into the high fidelity bracket, the rest are in medium. Which is a shame because the 290/390/970 are doing well.

 

it looks like this benchmark agrees with recommended specs provided by Oculus a few months ago.

[suicidalfranco]

loving my VM setup

[spellmanuk]

lol 

[Thony]

3 hours ago, Humbug said:

Does the counter work properly for this?

It's rendering in stereoscopy and in windowed mode.

It came up and changed values. Maybe it shows total fps (2 eyes added together).

[NumLock21]

 

 

[Sharkyx1]

On 22/02/2016 at 4:40 PM, ElfenSky said:

I don't particularly care, because most likely I'll be getting the PlayStation VR kit and not the Oculus/Vive, but none the less, my laptop can run the test somewhat competently, averaging around 75.

 

Also it runs at the above mentioned fps, but does not detect my Nvidia GPU and says that my system is not capable of rendering even low fidelity VR. After running at around 75 average fps. Fuck logic, I guess.

 

 

and

 

 

 

It doesn't base it's  tests on model numbers, it goes by performance, it will sometimes read you Intel graphics by accident but it is still using the frame  rate in the test 75 isn't nearly enough, it considers 90 minimum 

[elfensky]

23 minutes ago, Sharkyx1 said:

It doesn't base it's  tests on model numbers, it goes by performance, it will sometimes read you Intel graphics by accident but it is still using the frame  rate in the test 75 isn't nearly enough, it considers 90 minimum 

Oooooh ok.

[Sharkyx1]

43 minutes ago, ElfenSky said:

Oooooh ok.

Yeah it gave me a score of 0.3 and I think we have a similar gpu

 

I think maxwell is better in this bench though because it says 960 is passable and it's only 5-10% faster 

[ashtaroth]

Hmm without multigpu will enabling it make a difference?

[Humbug]

16 hours ago, Sharkyx1 said:

Yeah it gave me a score of 0.3 and I think we have a similar gpu

 

I think maxwell is better in this bench though because it says 960 is passable and it's only 5-10% faster 

Maxwell seems to be way more suited to virtual reality than Kepler, at least according to this benchmark. There is a compiled result list on reddit and the GTX 970 is doing a lot better than the 780 and 780ti.

[Sharkyx1]

1 hour ago, Humbug said:

Maxwell seems to be way more suited to virtual reality than Kepler, at least according to this benchmark. There is a compiled result list on reddit and the GTX 970 is doing a lot better than the 780 and 780ti.

Do you think there are some inherent advantages in maxwell?

Like sure it can tesselate better and Witcher 3 perfromance is nothing to scoff at, but this is just rendering two offset windows in my eyes. shouldn't give up that much performance to maxwell

[callum92]

23 hours ago, VinZie said:

When I went through the files and the directories I found some thing leading to Half Life 3, and Left for Dead 3... I recommend you all look through the directories and check out what is in there.

Which files are they exactly? If you are going to make such wild claims at least back them up with something.

[Humbug]

15 minutes ago, callum92 said:

Which files are they exactly? If you are going to make such wild claims at least back them up with something.

he's not lying. It's nothing new. Valve has left lots of HL3 stuff lying around.

But it doesn't mean that they have anything close to releasing or anything concrete.

It just means that they are or have worked on it at some point.

[adiobe]

Not a problem! #sweet

[VinZie]

23 minutes ago, Humbug said:

he's not lying. It's nothing new. Valve has left lots of HL3 stuff lying around.

But it doesn't mean that they have anything close to releasing or anything concrete.

It just means that they are or have worked on it at some point.

But they are getting somewhere with Half Life 3.

[DXMember]

yeah.... no comment

from the posted results it really looks like the tool makes the conclusion based on the model numbers instead of the performance

[Nowak]

 

I don't know what I was expecting.

[Humbug]

26 minutes ago, DXMember said:

yeah.... no comment

from the posted results it really looks like the tool makes the conclusion based on the model numbers instead of the performance

nope, it's going according to the performance and rendering results. It dynamically adjusts image quality to try to stay above 90fps.

People are getting better results by overclocking.

And there are a couple cases where it has wrongly displayed the name of the crappy onboard GPU but said the system was VR ready (since it got good performance results by rendering on the discrete GPU). So it's not going by the name.

[DXMember]

3 hours ago, Humbug said:

nope, it's going according to the performance and rendering results. It dynamically adjusts image quality to try to stay above 90fps.

People are getting better results by overclocking.

And there are a couple cases where it has wrongly displayed the name of the crappy onboard GPU but said the system was VR ready (since it got good performance results by rendering on the discrete GPU). So it's not going by the name.

okay I guess... but why does it reduce IQ if there are no frames below 90?

also how do you explain 680's Failing but 770's Passing?

[Thony]

Does anyone know why the test runs in very small window ?! Mine was little less than 1/4 of 1080p....

 

Not a good test IMO.

[VinZie]

7 hours ago, Thony said:

Does anyone know why the test runs in very small window ?! Mine was little less than 1/4 of 1080p....

 

Not a good test IMO.

Got dat k70 tho  I do too!