Valve Releases SteamVR Performance Test - Is Your Rig Ready?

Subject: Graphics Cards | February 22, 2016 - 06:03 PM |
Tagged: vive, valve, steamvr, steam, rift, performance test, Oculus, htc

Though I am away from my stacks of hardware at the office attending Mobile World Congress in Barcelona, Valve dropped a bomb on us today in the form of a new hardware performance test that gamers can use to determine if they are ready for the SteamVR revolution. The aptly named "SteamVR Performance Test" is a free title available through Steam that any user can download and run to get a report card on their installed hardware. No VR headset required!

And unlike the Oculus Compatibility Checker, the application from Valve runs actual game content to measure your system. Oculus' app only looks at the hardware on your system for certification, not taking into account the performance of your system in any way. (Overclockers and users with Ivy Bridge Core i7 processors have been reporting failed results on the Oculus test for some time.)

View Full Size

The SteamVR Performance Test runs a set of scenes from the Aperture Science Robot Repair demo, an experience developed directly for the HTC Vive and one that I was able to run through during CES last month. Valve is using a very interesting new feature called "dynamic fidelity" that adjusts image quality of the game in a way to avoid dropped frames and frame rates under 90 FPS in order to maintain a smooth and comfortable experience for the VR user. Though it is the first time I have seen it used, it sounds similar to what John Carmack did with the id Tech 5 engine, attempting to balance performance on hardware while maintaining a targeted frame rate.

The technology could be a perfect match for VR content where frame rates above or at the 90 FPS target are more important than visual fidelity (in nearly all cases). I am curious to see how Valve may or may not pursue and push this technology in its own games and for the Vive / Rift in general. I have some questions pending with them, so we'll see what they come back with.

View Full Size

A result for a Radeon R9 Fury provided by AMD

Valve's test offers a very simple three tiered breakdown for your system: Not Ready, Capable and Ready. For a more detailed explanation you can expand on the data to see metrics like the number of frames you are CPU bound on, frames below the very important 90 FPS mark and how many frames were tested in the run. The Average Fidelity metric is the number that we are reporting below and essentially tells us "how much quality" the test estimates you can run at while maintaining that 90 FPS mark. What else that fidelity result means is still unknown - but again we are trying to find out. The short answer is that the higher that number goes, the better off you are, and the more demanding game content you'll be able to run at acceptable performance levels. At least, according to Valve.

View Full Size

Because I am not at the office to run my own tests, I decided to write up this story using results from a third part. That third party is AMD - let the complaining begin. Obviously this does NOT count as independent testing but, in truth, it would be hard to cheat on these results unless you go WAY out of your way to change control panel settings, etc. The demo is self run and AMD detailed the hardware and drivers used in the results.

  • Intel i7-6700K
  • 2x4GB DDR4-2666 RAM
  • Z170 motherboard
  • Radeon Software 16.1.1
  • NVIDIA driver 361.91
  • Win10 64-bit

GPU Score
2x Radeon R9 Nano 11.0
GeForce GTX 980 Ti 11.0
Radeon R9 Fury X 9.6
Radeon R9 Fury 9.2
GeForce GTX 980 8.1
Radeon R9 Nano 8.0
Radeon R9 390X 7.8
Radeon R9 390 7.0
GeForce GTX 970 6.5

These results were provided by AMD in an email to the media. Take that for what you will until we can run our own tests.

First, the GeForce GTX 980 Ti is the highest performing single GPU tested, with a score of 11 - because of course it goes to 11. The same score is reported on the multi-GPU configuration with two Radeon R9 Nanos so clearly we are seeing a ceiling of this version of the SteamVR Performance Test. With a single GPU score of 9.2, that is only a 19% scaling rate, but I think we are limited by the test in this case. Either way, it's great news to see that AMD has affinity multi-GPU up and running, utilizing one GPU for each eye's rendering. (AMD pointed out that users that want to test the multi-GPU implementation will need to add the -multigpu launch option.) I still need to confirm if GeForce cards scale accordingly. UPDATE: Ken at the office ran a quick check with a pair of GeForce GTX 970 cards with the same -multigpu option and saw no scaling improvements. It appears NVIDIA has work to do here.

Moving down the stack, its clear why AMD was so excited to send out these early results. The R9 Fury X and R9 Fury both come out ahead of the GeForce GTX 980 while the R9 Nano, R9 390X and R9 390 result in better scores than NVIDIA's GeForce GTX 970. This comes as no surprise - AMD's Radeon parts tend to offer better performance per dollar when it comes to benchmarks and many games.

View Full Size

There is obviously a lot more to consider than the results this SteamVR Performance Test provides when picking hardware for a VR system, but we are glad to see Valve out in front of the many, many questions that are flooding forums across the web. Is your system ready??

Source: Valve

February 22, 2016 | 06:21 PM - Posted by Anonymous (not verified)

Please start including links in your articles, - http://store.steampowered.com/app/323910/

February 22, 2016 | 06:43 PM - Posted by Ryan Shrout

It's linked at the Source section, but I'll add it higher up.

February 22, 2016 | 09:00 PM - Posted by Anonymous (not verified)

Scaling for the test should be 0% since the test is running in windowed mode. Can you check to see if the Nvidia card had any load? It absolutely wont use two of my 7979ghz in windowed mode since the second card is still in ULPS. This is normal for windowed mode in general.

Notice how mine only shows one card?

http://s8.postimg.org/ou8ttw6n9/Untitled.jpg

February 22, 2016 | 09:02 PM - Posted by Anonymous (not verified)

Im stupid and failed to read the entire blog. Sorry -multigpu launch option

February 22, 2016 | 09:12 PM - Posted by Anonymous (not verified)

After adding the multi-GPU option I have 127% scaling with two 7970ghz cards and 7.9 score. GCN has some longevity.

Crossfire
http://s7.postimg.org/xpqfx5t1n/7970ghz_Xfire.jpg

Single card
http://s8.postimg.org/ou8ttw6n9/Untitled.jpg

February 23, 2016 | 02:20 AM - Posted by Ryan Shrout

That seems....unpossible.

February 23, 2016 | 08:23 AM - Posted by Anonymous (not verified)

I scored a 7.9 with my single GTX 970 and 3570K... The arrow was in the upper middle of the green bar.

February 23, 2016 | 08:49 AM - Posted by Topinio

Not unpossible if the scoring is an assigned quality metric (i.e. just a numerical representation of the effects that the engine has turned on or off to try and hit the frame rate target, so somewhat arbitrary) rather than a pure(r) numerical representation of work done.

In such a case, if the scoring used is ill-calibrated this type of scaling will happen -- though it's maybe that the use of Affinity Multi-GPU here (rendering one eye per GPU) removes a bottleneck that exists when this 1-GPU system tries to render both eyes.

The benchmark seems a bit simplistic in that the results from R9 290X, 390, 390X are very similar indeed, and seem also to be largely CPU-independent.

I'd love to see a fuller article test/show this by running on all the decent GPUs from the last 2-3 generations and superimposing the plots, and then with moving a decentish but not top-end GPU (e.g. R9 290X) around a few CPUs (e.g. Q9650, i5-2500, i7-6700).

February 23, 2016 | 05:49 PM - Posted by Anonymous (not verified)

The R9 290X and R9 390X are essentially the same GPU, the 390X just has 8 GB instead of 4 GB and a slightly higher clock. The 290 just has slightly reduced core count and clock (in some cases). On a scale from 0 (possible to score 0?) to 11, I would expect these cards to be very close, if not exactly the same score due to limited precision. This benchmark probably runs fine in 4 GB, so the 8 GB on the 390 cards probably doesn't help. Any 2 GB card is probably insufficient.

February 24, 2016 | 06:23 PM - Posted by Anonymous (not verified)

So would incredible double precision performance of the 7970ghz explain my score? Its higher than a lot of newer cards.

https://en.wikipedia.org/wiki/List_of_AMD_graphics_processing_units

February 22, 2016 | 07:00 PM - Posted by Anonymous (not verified)

Still labouring on an X58 platform, 12gb of ddr3 @1333 and a xeon hexacore CPU overclocked.

It seems this platform will run VR with one exception, still on a gtx680 GPU and the performance is too low. I have a pair, but no Sli scaling - will it squeeze by when Sli is sorted out, or am I looking at at least an expensive GPU upgrade?

February 22, 2016 | 07:01 PM - Posted by remc86007

Dynamic fidelity like tech was used in Halo 5 very successfully

February 22, 2016 | 07:07 PM - Posted by Tralalak (not verified)

Benchmark is smooth but result not.

2.0GHz QuadCore processor with 4GB Nvidia GeForce GTX 960 (PCIe 2.0 x16 @ x4)

link: http://i669.photobucket.com/albums/vv53/Artthurko/SteamVR%20Performance%...

February 22, 2016 | 07:16 PM - Posted by djotter

Where did you get a VIA C4650?

February 23, 2016 | 03:08 PM - Posted by Tralalak (not verified)

This is VIA EPIA M920-20Q Embedded Mini-ITX board. I bought it normally six months ago as other hardware.

link: http://i669.photobucket.com/albums/vv53/Artthurko/VIA%20QuadCore%20C4650...

February 23, 2016 | 12:56 AM - Posted by arbiter

yea i guess it should go without saying that your cpu is junk, "(PCIe 2.0 x16 @ x4)" likely doesn't help the card much either.

February 23, 2016 | 03:17 PM - Posted by Tralalak (not verified)

I agree. For me it is important to playable / smooth 30 fps vs 1080p. I'm not complaining.

Star Wars Battlefront
link: https://www.youtube.com/watch?v=JotRC775YWg

Call of Duty Black Ops 3 Multiplayer
link: https://www.youtube.com/watch?v=HXttAE2v2mM

METAL GEAR SOLID V: THE PHANTOM PAIN
link: https://www.youtube.com/watch?v=N90Q7fkdzi4

etc.

February 22, 2016 | 07:35 PM - Posted by Anonymous (not verified)

Struggling to see why a GTX680 isn't capable, when a 970 is, and they're not really more powerful..

is this a 2GB Vram vs 4GB vram debate when it comes to generations:

http://gpuboss.com/gpus/GeForce-GTX-970-vs-GeForce-GTX-680

February 22, 2016 | 08:46 PM - Posted by Shambles (not verified)

There is a significant difference between a 680 and a 970 and i would imagine that is the case but only in frame rate but frame times which will be much more important with VR.

February 22, 2016 | 08:47 PM - Posted by Shambles (not verified)

But only should read not only

February 23, 2016 | 12:58 AM - Posted by arbiter

gtx680 is around power of a gtx770, gtx970 is little bit over or about a gtx780. Yes likely 2gb vs 4gb would hinder the case as well since gpu has to render scene for each side with a slightly different angel on it.

February 23, 2016 | 06:02 PM - Posted by Anonymous (not verified)

It is the same geometry, the same textures, etc; it just offsets the view slightly. I don't see why this would take significantly more memory unless it is implemented badly. You are gong to have a larger frame buffer, but this is actually a small amount of memory in a modern system. It would be interesting to see how much memory the test case is actually using. I suspect 2 GB would be limiting texture resolution and other settings for this test case, even if was not rendering it twice for every frame. It would be interesting to compare if the test case allows rendering only one view at a time also. This would allow comparisons of resources between VR and non-VR rendering.

February 22, 2016 | 10:13 PM - Posted by -- (not verified)

gpuboss is subject to error, a lot.

stop using it unless you know what you are looking at.

same goes for cpuboss

February 22, 2016 | 08:51 PM - Posted by Anonymous (not verified)

Dont know where you are seeing any scaling at all on AMD. The test runs in windowed mode so only 1 GPU is used. I can confirm on my system the second GPU was still in ULPS. Fans were off and no lights.

February 23, 2016 | 02:20 AM - Posted by Ryan Shrout

VR doesn't use CrossFire like you are used to it - it enumerates each GPU individually. The game engine is accessing each GPU indepdent of one another.

February 22, 2016 | 09:02 PM - Posted by zMeul (not verified)

the test is flawed: https://i.imgur.com/r8A1N45.png
a Q9550 with a GTX970 is VR ready ... in what parallel universe?!

February 22, 2016 | 10:17 PM - Posted by -- (not verified)

yea the fish smell is really strong now.

3dmark demo will even let you know if you are VR ready...while holding the side of tuna.

February 22, 2016 | 10:34 PM - Posted by Anonymous (not verified)

A Q9550 is still a fast CPU. CPUs haven't scaled that much compared to GPUs, at least since 2007 or so. If the test is well multithreaded, then even one of these old quad cores may do quite well.

February 22, 2016 | 11:31 PM - Posted by zMeul (not verified)

that doesn't excuse that fact people might actually buy a VIVE based on this test; and when they start to play games .. oh boy

February 23, 2016 | 05:41 PM - Posted by Anonymous (not verified)

I have seen a lot of test indicating that most games, even without the added VR overhead for the GPU, are already GPU bound. The Q9550 was a very high-end CPU. It was released in 2008 at $530. I wouldn't be surprised if it can keep up fine in a GPU bound situation.

February 23, 2016 | 10:44 AM - Posted by -- (not verified)

well the oculus required at least a mid range i5 with the dev kit

February 23, 2016 | 05:31 PM - Posted by Anonymous (not verified)

VR is rendering almost the same scene twice, so I would expect it to be very GPU bound in most cases. It should be using the same data for both renders; the view is just shifted slightly. Most of the set-up work for the CPU can be shared if it is implemented properly. For a given scene, the load on the CPU should be almost the same as rendering a single view, while the GPU load is doubled.

Also, a Q9550 is 2.83 GHz with 12 MB of L2 cache (2x6 MB). A Core i5 only has 6 MB of L3 cache. It would be interesting to run some benchmarks to compare, but I wouldn't be surprised if a few applications run faster on the older chip due to the larger amount of on-die cache. At a minimum, it would help make up for the much slower memory system. The Q9550 was just about the most powerful C2D made. There were a few with slightly higher clock, up to 3.2 GHz and 1600 MT fsb rather than 1333. This is the problem with only giving currently selling hardware as a minimum. That isn't a good replacement for actually testing the hardware.

February 26, 2016 | 02:20 PM - Posted by -- (not verified)

that old proc is also tied down to DDR2 memory as well.

It might have been very fast for its day, but its coughing dust.

February 22, 2016 | 10:01 PM - Posted by Anonymous (not verified)

"Radeon parts tend to offer better performance per dollar when it comes to benchmarks and many games." What?! in Europe one R9 nano cost around 600€. Good Job AMD guys!!!!:p

February 22, 2016 | 10:47 PM - Posted by Anonymous (not verified)

Nano is a niche product with a high cost. Obviously you picked this one thing out of the crowd because you are trolling. No way any human could be that dumb.

February 22, 2016 | 11:33 PM - Posted by Anonymous (not verified)

If you are talking about AMD being a good value, it is, I hope obviously, about the 390 cards. These currently give you a powerful GPU with 8 GB of memory for a reasonable price. If I was buying now, in that price range, I would go with a 390 of some kind over a 970. The 3.5 GB of memory on a 970 could be a big limitation going forward.

February 23, 2016 | 02:39 AM - Posted by Dood (not verified)

Ran the test a few times on my system. 4670 and GTX 980 with a reference cooler that gets temperature limited. The score topped out at 7.5.

February 23, 2016 | 02:41 AM - Posted by Jann5s

Would be nice if Valve will share some global statistics about this test

February 23, 2016 | 04:12 AM - Posted by JohnGR

Unlocked (5 cores) and overclocked Athlon II 640 passes the test, but HD 7850 2GB fails. Nothing strange here.

February 23, 2016 | 08:49 AM - Posted by Anonymous (not verified)

"Nano is a niche product with a high cost. Obviously you picked this one thing out of the crowd because you are trolling. No way any human could be that dumb."

"human what?!" Ok, I guess you win :(

February 23, 2016 | 09:03 AM - Posted by holyneo

It said my rig is bad ass!!!

Got the best score possible on my G1 980ti.

So excited!

:)

February 23, 2016 | 09:49 AM - Posted by fvbounty

My rig is ready....11 with a 980Ti and a 4790K...

February 23, 2016 | 11:12 AM - Posted by Anonymous (not verified)

Well. My slightly OC'd GTX980 scores a 9.1/9.2 everytime so I'd say that's rather good

February 23, 2016 | 11:12 AM - Posted by Anonymous (not verified)

Well. My slightly OC'd GTX980 scores a 9.1/9.2 everytime so I'd say that's rather good

February 23, 2016 | 12:29 PM - Posted by Rick Savoie (not verified)

AMD 8350 at 4.2GHz, 2ea SSD for system and gaming drives. Zoltac GTX 970 4GB. Scord 6.3 Should be OK for us poor folk!

February 23, 2016 | 02:35 PM - Posted by godrilla (not verified)

My 6 year old i7 980 xe @ 4.3 ghz and amp extreme 980 ti factory clocked got a score of 12.
Ive seen a guy get a score of 13 on hybrid 980ti oc.
This is very gpu bound.

February 24, 2016 | 05:55 PM - Posted by Anonymous (not verified)

You got a 12, someone else got a 13.

That's incredibly impressive when the scale tops out at 11.

February 23, 2016 | 05:51 PM - Posted by Anonymous (not verified)

GP100 amd GP104 should be great for VR with their HBM2 and possible GDDR5X on lower end models.

February 23, 2016 | 05:57 PM - Posted by siriq

I got 4.9 score with FX 8350 and GTX 570@OC .
This test is really funny because most of the time i am just under 90 fps by 12-3. There is a heavy gpu test and i am getting 5x fps, still.

February 23, 2016 | 06:31 PM - Posted by Anonymous (not verified)

Here is my result ^^ http://s15.postimg.org/ne07irjij/win.jpg

February 23, 2016 | 06:42 PM - Posted by siriq

i just did a one more round after all. I got less sore now. 4.5 and it says , VR capable :D
http://i64.tinypic.com/210jdrl.jpg

February 23, 2016 | 06:44 PM - Posted by siriq

So no panic guys. If i got VR capable status with GTX 570, most of you with new GPU should be fine!
Also , i had score of 4.9 before. This test is not really reliable!

February 24, 2016 | 03:49 AM - Posted by Anonymous (not verified)

My GTX970 gets 6.9 with a i5-3570K @4.2OC).

So no need for me to get a 6700K by the looks.

February 24, 2016 | 06:53 AM - Posted by siriq

GTX 570 with some extra OC and scores 5.5 !!

http://i68.tinypic.com/r90sxg.jpg

February 24, 2016 | 09:19 AM - Posted by Anonymous (not verified)

i5-2320, 16GB DDR3 1333, 780 GHz edition score 5 so im almost in the green(could probably update drivers and get better).

February 24, 2016 | 11:09 AM - Posted by siriq

I managed to get 5.9 with GTX 570.

February 24, 2016 | 07:52 PM - Posted by Anonymous (not verified)

Something tells me AMD didn't OC the 980 or 970 so they can save the slan.? I got an 11 with an i5 and 980Ti so here's to 4 cores.

February 24, 2016 | 07:58 PM - Posted by Anonymous (not verified)

7.3 with 3770K@4GHz + 290X 1040/1300

Will see how 3930K@5GHz + 290X 1200/1500 tri-CF performs

February 24, 2016 | 10:25 PM - Posted by Anonymous (not verified)

Got 1000000 with my msi gtxx 970 and i7-3770k!
Really though, it was a 7.6 score with my clocks: 970 @ 1500mhz/3770k @ 4.4mhz.

February 25, 2016 | 02:14 AM - Posted by Thordrune (not verified)

1.8 (low-mid capable) with a GTX 680 @ 1176 MHz, an i7-980X at 4.2 GHz, and a fresh install of Windows 10. 9,398 frames rendered, 0 below 90 fps, 0 CPU bound. GPU pegged at 99%, 1.6-1.8 GB VRAM usage. Strange scoring... Oh well, I'm not planning on going VR anytime soon :).

February 26, 2016 | 10:16 AM - Posted by bhappy (not verified)

Does this test support multi res shading tech?

February 27, 2016 | 05:35 PM - Posted by Anonymous (not verified)

I have two gtx 980ti and a i7 5820k overclocked to 4.6GHz and i can onlu get a score of 10.8?

February 28, 2016 | 04:07 AM - Posted by Anonymous (not verified)

3570k @ 4.4Ghz + MSI GTX 970 @ 1475Mhz

http://i.imgur.com/ep0LVCp.png?1

February 28, 2016 | 04:32 AM - Posted by razor512

Phenom II x6 1075t at 3.9GHz, and GTX 970 at 1.45GHz
7.9
http://i.imgur.com/JU8H5Et.jpg

February 29, 2016 | 03:15 PM - Posted by d7wish (not verified)

i5 4690k + gtx 980 hit 8.8 you have a serious flaw with your test

March 18, 2016 | 09:20 AM - Posted by CoffeeBlack (not verified)

Eh any idea how to get it to run properly on a laptop with a secondary gpu?

dell prec 3510
-i5 6300hq
-intel hd 530
-amd firepro w5130m
-16gb ddr4

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <blockquote><p><br>
  • Web page addresses and e-mail addresses turn into links automatically.

More information about formatting options

CAPTCHA
This question is for testing whether you are a human visitor and to prevent automated spam submissions.