SMP is a pascal only feature and has to be implemented per game (or engine). I think nvidia's funhouse and maybe raw data use it. I would expect Fury X to be closer to the 980ti than it is
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
AMD & NVIDIA GPU VR Performance: Valve's Robot Repair @ [H]
SMP is a pascal only feature and has to be implemented per game (or engine). I think nvidia's funhouse and maybe raw data use it. I would expect Fury X to be closer to the 980ti than it is
MavericK
Zero Cool
- Joined
- Sep 2, 2004
- Messages
- 31,898
AFAIK Raw Data does not use SMP...performance would probably be a lot better if it did.
Yes, I am aware; I mentioned SMP in passing to emphasize the significance of the double geometry processing for VR applications. As nVidia products have much more capable geometry processing than AMD's products, this could be a significant source of performance degradation for AMD, but I'm just speculating.
- Joined
- May 18, 1997
- Messages
- 55,626
Kidding aside, one thing is seriously burning in my mind when I showed these VR reviews to my colleagues.
If AMD is doing this badly, how badly would the PS VR run?
If Sony makes PS VR work at Nvidia levels, AMD is going to need some serious inflection.
If AMD is doing this badly, how badly would the PS VR run?
If Sony makes PS VR work at Nvidia levels, AMD is going to need some serious inflection.
- Joined
- May 18, 1997
- Messages
- 55,626
I don't think the PS4 VR HMD is going to be as high resolution at Vive/Rift. Not sure on that though...
Poorer graphics quality - like, say, last-gen-console-grade graphics?
PSVR is 1920*1080, 120hz or 90hz. Most games are 60fps (reprojected at 120). 1920*1080*60 is almost 50% less pixels per second than rift/vive (2160*1200*90). That's ignoring supersampling on PC which makes it even more demanding. I'm not sure if PSVR does SS and if so how much.
I tried PSVR at best buy, and while it is inferior to the vive in IQ it's not terrible. I have a pretty good PC and I still may get a PSVR for the exclusive games.
I tried PSVR at best buy, and while it is inferior to the vive in IQ it's not terrible. I have a pretty good PC and I still may get a PSVR for the exclusive games.
And that's also ignoring the performance gains when coding close to the metal. Uncharted 4 running on PS4 hardware (which is ancient by today's PC standards) looks absolutely stunning and maintains stable 1080p30 with superior antialiasing.
Nvidia's answer to AMD's "premium VR" is... "enthusiast VR"
Nvidia's answer to AMD's "premium VR" is... "enthusiast VR"
EchoWars
Limp Gawd
- Joined
- Jan 7, 2010
- Messages
- 497
Wound, meet salt.
Not sure why that paragraph is necessary. AMD's lack of chops in the VR world is pretty apparent already.
Considering how much of a 'premium VR experience' you're supposed to get from the Polaris launch, per AMD/RTG, the paragraph is very necessary and the reason why I read HARDocp and not SOFTocp from the other sites.
EchoWars
Limp Gawd
- Joined
- Jan 7, 2010
- Messages
- 497
Salient points were made long before that paragraph. Pissing in their Cheerio''s doesn't add much to the article, except to give the Nvidia fanboyz a woodie.
And to trigger you.
- Joined
- May 18, 1997
- Messages
- 55,626
He is missing his safe space. I think he can find it over at the AMD marketing department under Chris Hook's desk.
- Joined
- May 18, 1997
- Messages
- 55,626
Have now talked to Valve and NVIDIA about our testing and both have confirmed that Adaptive Quality was broken with our driver sets.
- Joined
- May 18, 1997
- Messages
- 55,626
EchoWars
Limp Gawd
- Joined
- Jan 7, 2010
- Messages
- 497
Touché.
In my defense, I was poked with a sharp stick beforehand.
In my defense, I was poked with a sharp stick beforehand.
For a VR user who does neither read benchmarks like HardOCP's nor has ever owned any Nvidia cards, what is the experience like for that VR user?
Will he still think VR is great? Or will he think VR is not ready for the masses? Or worse, VR is sick inducing crap?
Will he still think VR is great? Or will he think VR is not ready for the masses? Or worse, VR is sick inducing crap?
So the quality/loading of each GPU could have been different? Uneven? How much time is spent subjectively and objectively for looking at rendering quality - number of objects rendered - errors? To ensure the outcome or numbers accurately reflect similar if not same GPU loading.
My first thought is does AMD do well in any VR title? So far AMD solutions or options for VR looks to be the weakest by far of the bunch. FuryX consistently performed poorly from a VR standpoint. Then I wonder why nothing DX12 has surfaced at all for VR since that API seems to be a better API to use for VR?
Keep up the excellent reviews, while some are waiting for Vega - if Vega performs like this or below Nvidia current generation then Volta is probably the real deal performer for the 14nm generation and that looks like late next year. Now VR is not even using some of Nvidia VR potential with Multi-Projection (saving geometry pass for both eyes, performance enhancing, Nvidia calls it Single Pass Stereo), Multires Shading (performance enhancing by rendering less looked at areas like the sides at a lower resolution), then of course Nvidia VR SLI. Nvidia provides the hardware but the tools/libraries/software/support to back up that hardware as well. Nvidia is looking really good so far in VR - AMD needs to get to work in a more serious way. If Vega continues the poorer performance in VR I can see me going with Volta next year keeping the 1070 for VR once I get a headset.
My first thought is does AMD do well in any VR title? So far AMD solutions or options for VR looks to be the weakest by far of the bunch. FuryX consistently performed poorly from a VR standpoint. Then I wonder why nothing DX12 has surfaced at all for VR since that API seems to be a better API to use for VR?
Keep up the excellent reviews, while some are waiting for Vega - if Vega performs like this or below Nvidia current generation then Volta is probably the real deal performer for the 14nm generation and that looks like late next year. Now VR is not even using some of Nvidia VR potential with Multi-Projection (saving geometry pass for both eyes, performance enhancing, Nvidia calls it Single Pass Stereo), Multires Shading (performance enhancing by rendering less looked at areas like the sides at a lower resolution), then of course Nvidia VR SLI. Nvidia provides the hardware but the tools/libraries/software/support to back up that hardware as well. Nvidia is looking really good so far in VR - AMD needs to get to work in a more serious way. If Vega continues the poorer performance in VR I can see me going with Volta next year keeping the 1070 for VR once I get a headset.
- Joined
- May 18, 1997
- Messages
- 55,626
Both.
- Joined
- May 18, 1997
- Messages
- 55,626
No. I spent a lot of time ingame making sure we were apples to apples. No differences that I could tell at all.
- Joined
- May 18, 1997
- Messages
- 55,626
Well, our testing is what pointed out to NVIDIA and Valve that AQ was broken. Funny though, AMD never contacted HardOCP about it. Given that AQ is now fixed, it does not make for much of an objective test. If AQ is working on both, about all I could tell you is that one looked X and the other looked Y. That said, I know how to turn AQ off, but then we would just get the same results.