RTX 3080 vs Radeon 6800 XT - VR testing

reaper12

2[H]4U
Joined
Oct 21, 2006
Messages
2,581

Dayaks

[H]F Junkie
Joined
Feb 22, 2012
Messages
8,288
Same as the last few generations so. For VR stick with Nvidia. Maybe AMD will improve performance through driver updates.

Yeah. Some of those charts are pretty nasty. Dropped frames are a huge problem in VR for me...
 

reaper12

2[H]4U
Joined
Oct 21, 2006
Messages
2,581
Yeah. Some of those charts are pretty nasty. Dropped frames are a huge problem in VR for me...

And that doesn't even include Nvidia features, like VRSS.

At least AMD are in the same ballpark now.
 

MavericK

Zero Cool
Joined
Sep 2, 2004
Messages
30,792
Yeah. Some of those charts are pretty nasty. Dropped frames are a huge problem in VR for me...

Some of these games are not built for VR from the ground up, though, and clearly optimization is very poor.
 

Bees

[H]ard|Gawd
Joined
Jul 24, 2011
Messages
1,208
Gonna be great to test out performance once Nvidia gets the stuttering and lag issues fixed. It's a mess right now, completely unplayable experience in VR.
 

Dayaks

[H]F Junkie
Joined
Feb 22, 2012
Messages
8,288
Gonna be great to test out performance once Nvidia gets the stuttering and lag issues fixed. It's a mess right now, completely unplayable experience in VR.
You mean AMD? They had way more frame drops....
 

MaZa

2[H]4U
Joined
Sep 21, 2008
Messages
3,246
You mean AMD? They had way more frame drops....

Nvidia currently has some driver issues with VR and they are aware of it. Not sure if everyone is affected by it but for those who are need to install an old 446 driver version. It does not work with RTX3000 series though.
 

Aireoth

Supreme [H]ardness
Joined
Oct 12, 2005
Messages
5,050
reading that it looks like AMD has VR driver issues. Not entirely surprising as it is niche and AMD historically has driver hiccups.
 

reaper12

2[H]4U
Joined
Oct 21, 2006
Messages
2,581
RDNA2 has Variable Rate Shading. Ball is in dev's court now, to support it.

VRSS isn't Variable Rate Shading, it's Variable Rate Super Sampling. It's based on Variable Rate Shading but not the same.

Just been pedantic I know!! :p
 

funkydmunky

2[H]4U
Joined
Aug 28, 2008
Messages
2,663
I would have hoped the article had used a modern headset. And maybe more then one.
Some great VR comparison articles on the [H] in the day. Nvidia trumped AMD. But that was on older GCN.
 

funkydmunky

2[H]4U
Joined
Aug 28, 2008
Messages
2,663
I found the results quite close inspite of the conclusion. AMD has obviously come a long way here. The infinity cache does seem to be hitting its performance limits at VR resolutions (SS and such pushing it) but over all it is very close and game dependent. For me I looked at PC2 which I have and play with an investment. Very close as most titles were.
 

bobzdar

[H]ard|Gawd
Joined
Jun 6, 2003
Messages
1,834
Unity and UE4 both have a lot of Nvidia optimizations (Unity has single pass stereo in main branch, UE4 in an Nvidia branch) and around 2/3 of VR games leverage one of those two engines. That's the main reason I went Nvidia this gen (the other being CUDA), and I don't see that advantage going away anytime soon. If AMD manage to take 1/3 of the market this gen, it could cause the main engines to incorporate LiquidVR and close that ~10-15% perf gap, but you won't see that in older games, only newer ones going forward.
 

bobzdar

[H]ard|Gawd
Joined
Jun 6, 2003
Messages
1,834
Nvidia currently has some driver issues with VR and they are aware of it. Not sure if everyone is affected by it but for those who are need to install an old 446 driver version. It does not work with RTX3000 series though.
No issues in VR with my 3090 so far, knock on wood.
 

cybereality

Supreme [H]ardness
Joined
Mar 22, 2008
Messages
7,044
Honestly didn't look that bad for AMD.

I mean, they are behind, but it looks like a viable alternative and still somewhat trading blows (or getting close).

So I would have no issue running AMD on my main rig, even if there was a small performance hit, it seems like an option now at least.
 

Andrew_Carr

[H]ard|Gawd
Joined
Feb 26, 2005
Messages
1,993
Anyone know of a review with a ryzen 5xxx cpu using smart access memory? From what I understand pairing a ryzen cpu with a 6xxx series GPU can result in some decent performance gains, but it seems like a lot of people are still using 10900k's for reviews.
 

bobzdar

[H]ard|Gawd
Joined
Jun 6, 2003
Messages
1,834
I saw this review, doesn't look great for AMD but at least they are trading blows in a couple titles.

https://babeltechreviews.com/vr-war...rtx-3080-15-vr-games-performance-benchmarked/

They tested SAM in VR (well Superposition) and there were some minor gains.

https://hothardware.com/reviews/amd-radeon-rx-6800-big-navi-gpu-review?page=2
Problem is SAM is super engine dependent, at least so far, so unless they test in actual VR titles it's pretty useless data. RDNA2 roughly the same as RDNA, trading blows in 20% of titles and losing by 10-15% in the rest vs Nvidia counterparts. I honestly think it's on the game engine side and not anything to do with AMD's drivers or hardware, just lack of optimization for AMD.
 

funkydmunky

2[H]4U
Joined
Aug 28, 2008
Messages
2,663
Why is every review so different? I see dirastically different numbers and can't make heads or tails out of it all. Add SAM to the mix and it adds another level of confussion. It is entirely unimpressive or very exciting with a lot of promise depending on the review. Too many variables it seems.
 
Last edited:

atarumoroboshi18

Limp Gawd
Joined
Nov 9, 2013
Messages
181
Any new reviews showing up for the 3080 Founders Edition? I'm about to get one next week and would like to have some benchmarks for them.
 

Aegir

Gawd
Joined
Apr 15, 2020
Messages
565
The reviews that are up should still apply...
I'd disagree with that, to be honest.

Driver optimizations and games that have thrown in new patches could result in significantly different performance.
I'd still consider it a rehash review, but it's always nice to see where a card sits after it's gotten optimized via GeForce Now style mass-user bug reports and data collection.
Surely there's something that Nvidia has done to increase performance. =p

Well, maybe. It'll take testing to find out. A system and benchmark suite must be tested in 2020, and that exact same system, except with updated drivers, again in 2021, Then if the results are different, it might lead to new conclusions.
 

MavericK

Zero Cool
Joined
Sep 2, 2004
Messages
30,792
I'd disagree with that, to be honest.

Driver optimizations and games that have thrown in new patches could result in significantly different performance.
I'd still consider it a rehash review, but it's always nice to see where a card sits after it's gotten optimized via GeForce Now style mass-user bug reports and data collection.
Surely there's something that Nvidia has done to increase performance. =p

Well, maybe. It'll take testing to find out. A system and benchmark suite must be tested in 2020, and that exact same system, except with updated drivers, again in 2021, Then if the results are different, it might lead to new conclusions.
I mean...maybe. I don't think they've fixed that stuttering issue with SteamVR yet, but I guess it's possible newer drivers have improved performance. Though, TBH, I doubt it's significant.
 

Aegir

Gawd
Joined
Apr 15, 2020
Messages
565
I mean...maybe. I don't think they've fixed that stuttering issue with SteamVR yet, but I guess it's possible newer drivers have improved performance. Though, TBH, I doubt it's significant.

It's certainly not significant, but if hardware is your hobby, and not just gaming, then it's certainly nice to know the final state of hardware.
Like, I hear AMD played some "aged like fine wine" thing with their drivers, where on launch, it was absolute madness, but by the end of its lifetime, it was more... bearable.

So a person buying a used GPU might benefit from seeing the current state of performance, rather than the initial reviews, if it means buying one used model instead of another.
 

Andrew_Carr

[H]ard|Gawd
Joined
Feb 26, 2005
Messages
1,993
It's certainly not significant, but if hardware is your hobby, and not just gaming, then it's certainly nice to know the final state of hardware.
Like, I hear AMD played some "aged like fine wine" thing with their drivers, where on launch, it was absolute madness, but by the end of its lifetime, it was more... bearable.

So a person buying a used GPU might benefit from seeing the current state of performance, rather than the initial reviews, if it means buying one used model instead of another.
There was a meta review done on this last year I believe where they found significant gains on AMD hardware from initial drivers to current drivers a year or so later. It fits with my experiences of rough launches followed by a scramble to cleanup things.
 

reaper12

2[H]4U
Joined
Oct 21, 2006
Messages
2,581
AMD's RDNA and RDNA2 cards are fine for VR. Way better than the previous generations!!

It's just for VR, Nvidia are better. The gap is closing though.
 
Top