SteamVR Performance Test: 16 GPUs Compared

[William Schnatz]

Your score with the R9 380 seems way off. I just tested my MSI R9 380 Gaming 4G model that I have clocked at 1120 MHz and it got a fidelity score of 5.1.

 

[blackbit75]

I don't know why, NVidia cards, with much lower specs than Radeon, performs better in those tests.
Normally, in not Steam games, performance is more balanced between the two brands.
My neurotic mind, says that there is a benefit in optimization for NVidia in Steam.
[Yes, truly, I'm a fanboy of Radeon ]
Cheers,

 

[KPalm The Wise]

I get 10.7 with my r9 295x2, pre Crimson drivers...
0% frames under 90
0% frames CPU bound

Seems stupid to run this test with Fraps on, that's effecting the score

 

[KPalm The Wise]

Using the pre-crimson driver my r9 295x2 gets 10.7... No manual crossfire button, just ran it and get consistent 10.7... 0% frames under 0% 0% of frames CPU bound...

Wouldn't be surprised if Fraps is botching the comparison, fucking stupid to run the tests with it on

 

[kcarbotte]

I get 10.7 with my r9 295x2, pre Crimson drivers...
0% frames under 90
0% frames CPU bound

Seems stupid to run this test with Fraps on, that's effecting the score

running fraps for framerate and frametime logging has no effect on the performance. This is how we test all GPUs.
Fraps only has a performance hit when recording video, which we were not doing for these tests.


As stated, there was something not working correctly with the crossfire performance in our testing. That could have been isolated to the board, or a driver. On an X99 system, the same card had significantly better performance, more in line with what you'd expect from that card.

Also, if you rerun the test, you'll likely get a different number. From my experience, the result the test spits out varies each time.

 

[KPalm The Wise]

I get 10.7 with my r9 295x2, pre Crimson drivers...
0% frames under 90
0% frames CPU bound

Seems stupid to run this test with Fraps on, that's effecting the score

running fraps for framerate and frametime logging has no effect on the performance. This is how we test all GPUs.
Fraps only has a performance hit when recording video, which we were not doing for these tests.


As stated, there was something not working correctly with the crossfire performance in our testing. That could have been isolated to the board, or a driver. On an X99 system, the same card had significantly better performance, more in line with what you'd expect from that card.

Also, if you rerun the test, you'll likely get a different number. From my experience, the result the test spits out varies each time.

Have run the test multiple times, get exactly 10.7 every time... If you're getting significant changes your testing procedure is not valid

 

[Elmz]

Strange scores... I have a GTX 980 and mine is 9.6 with Xeon W3690

 

[deuce_23]

im researching which VR to get im leaning towards vive but how can the ps4 run a vr headset with a HD 7870

 

[kcarbotte]

im researching which VR to get im leaning towards vive but how can the ps4 run a vr headset with a HD 7870

Lower fidelity graphics, lower resolution, and lower refresh rate.
most PSVR games will actually run at 60 FPS and use some sort of reprojection technology to double it to 120 Hz