Question 7900XT/XTX for PC VR gaming. Thoughts?

[moonbogg]

Early reviews for these cards with VR were bad, but things have changed and they seem to do better now. The cards have raster grunt with enough Vram. Any experience with these cards and VR? I'm not waiting around for Nvidia to offer a real GPU, so they're out. It's either I keep what I got or floor it to the nearest Microcenter and grab one of these AMD cards and smile all the way home. So, do these things work for VR or not?

 

[CP5670]

I know you hate nvidia prices, but the 4090 is far and away the best VR card, with a bigger lead than in anything flatscreen. It was really a night and day difference from a 3090 for me, which often struggled in sims. The AMD cards do look much more attractive at current prices, but no idea about their current VR performance.

 

[gdansk]

No consideration of the 4080 Super?

 

[moonbogg]

No consideration of the 4080 Super?

For $800 yeah.

 

[Aapje]

I would personally wait a few days for the 4070 Ti Super reviews and decide then.

 

[Dannar26]

I can't personally speak to VR, but I remember a buddy of mine who brought in his rig with an HTC vive back in 2016. Seemed more than adequate for me, and it was running a 1080 (non TI).

Don't know if VR has gotten a lot harder to run since then with things like Alyx...but it's hard to beat getting something like a 6950XT for less than $600. Not sure how much grunt you need or what you're upgrading from.

I too refuse to pay for nvidia prices. There's a valid competitor, and I'll happily buy team green again when their products aren't on average $100-200 for a similar experience.

 

[Aapje]

Alyx is not actually that hard to run. It's optimized very well and was perfectly playable even on a 1660 Super. It's really VR simming that can be very demanding.

 

[CP5670]

The sims are all poorly optimized and run much worse than flatscreen, and VR at native runs at a much higher resolution too. Alyx dynamically adjusts the resolution in different parts of the eye and runs well on any hardware.

 

[dr1337]

My 6900xt and before that RX580 drive VR just fine. Granted I only play VR native games and haven't bothered with something like Assetto (which I hear runs really bad). Your headset is really gonna determine the performance load for most games. My Oculus Rift runs fine enough on a 3050 laptop to play VTOL and Pavlov, but I don't bother trying it with my Index because I'm sure I'd have to turn SS all the way down.

I highly doubt there's any issue at all with the 7000 series running VR. I have been reading people say 'AMD is bad for VR' for 7 years now and it's literally never been true to my experience.

 

[Aapje]

The sims are all poorly optimized and run much worse than flatscreen, and VR at native runs at a much higher resolution too. Alyx dynamically adjusts the resolution in different parts of the eye and runs well on any hardware.

It's a bit more complicated than that. Alyx has smartly chosen to have almost exclusively very enclosed spaces with short draw distances, which means that they don't have to render as much. You can't really get away with that in flying or driving sims.

My Oculus Rift runs fine enough on a 3050 laptop to play VTOL and Pavlov, but I don't bother trying it with my Index because I'm sure I'd have to turn SS all the way down.

I highly doubt there's any issue at all with the 7000 series running VR. I have been reading people say 'AMD is bad for VR' for 7 years now and it's literally never been true to my experience.

Both Pavlov and VTOL are very undemanding games (and they look it). It's really another world if you want to run realistic and filled out sims or even a high quality game like Lone Echo or Lone Echo II (highly recommended, btw, despite a bit of jank).

And both with the 6000 and 7000 series cards from AMD we've seen that the initial performance in VR wasn't that good, with them taking their sweet time to fix it. Once the drivers are fixed the cards are perfectly fine. This is not just anecdotal, but Babeltechreviews showed it in their early driver reviews (when they still did VR reviews) with frametime graphs and Maraksot78 also had his issues. But if you bought those cards later on when the drivers are fixed, you are not going to understand what people were complaining about.

 

[dr1337]

It's really another world if you want to run realistic and filled out sims or even a high quality game like Lone Echo

No, its not, I played A LOT of echo arena and beat lone echo once, all on that RX 580 and rift CV1. Ran great and I had a good time.

Idk why some ppl are so bent on perpetuating this myth, one person had a problem at launch and thusly that's enough to write off every card?

Here's a video of someone playing Alyx at crazy resolution on a 7900xtx, shortly after launch, and getting an average of 119fps... I would say the problems with 'AMD and VR' are greatly over exaggerated.

 

[Aapje]

No, its not, I played A LOT of echo arena and beat lone echo once, all on that RX 580 and rift CV1. Ran great and I had a good time.

The Rift S has a substantially lower resolution than modern headsets. My experience with a 1660 Super driving a Quest 2 was that I was just able to run Lone Echo at medium settings, which was the minimum acceptable quality for me, with borderline frame rates.

To be honest, I don't really understand what you are trying to prove with this, since moonbogg wants to upgrade from a 1080 Ti, which is twice as fast as that 580. He thus wants a solid level above 1080 Ti performance, so he obviously has higher standards than you, which I am addressing. It's totally fine for you to have different standards, but it makes your comments rather irrelevant to his question.

Idk why some ppl are so bent on perpetuating this myth, one person had a problem at launch and thusly that's enough to write off every card?

I absolutely never said that. But it is a fact that VR issues are more common with AMD cards.

Here's a video of someone playing Alyx at crazy resolution on a 7900xtx, shortly after launch, and getting an average of 119fps... I would say the problems with 'AMD and VR' are greatly over exaggerated.

Yes, a single game tested without comparing it to the competitor cards is truly definitive. Especially a game that is not the most taxing. /sarcasm

Or you could read a review by an actual reviewer who tested a bunch of games and found that the 7900 XTX underperformed compared to its flat screen performance at launch and had glitches in some games: https://babeltechreviews.com/hellhound-rx-7900-xtx-vs-rtx-4080-50-games-vr/

I'm not saying to avoid those cards at all, but with AMD cards you seem to have to wait at least 6 months for VR to work pretty much as expected, and even then it is a little worse than Nvidia (in part to worse support by third parties, like SteamVR messing up AV1 support for AMD and sim games prioritizing DLSS over FSR). And just because VR tends to require more tinkering, and there are more resources to help with Nvidia cards when doing VR.

So I would think twice about purchasing an AMD card for VR shortly after the launch of a new line and later on, would personally still pay an (extra) premium to get an Nvidia card, everything else being equal. Of course, AMD cards are cheaper, so it is perfectly valid to go for them, especially if you do your due diligence and check the forums of the VR games you want to play, like the DCS and MSFS forums, if you are into flight simming.

With the new 4070 Ti Super, depending on your local prices, you may find it very competitive to a 7900 XT/XTX. Reviews will be out in 2 days or so and the actual card is out in 3. So I would suggest just waiting for those reviews and seeing how the card actually performs, before deciding.

 

[CP5670]

How did you get Lone Echo working? I have it on Revive but it crashes constantly and is basically unplayable. It seems to only work properly on the oculus headsets.

 

[blckgrffn]

No consideration of the 4080 Super?

The new 4080 Super should be an extremely solid card. $1k is $1k but at least its beast (4080 is a beast in my book and the Super will be faster & cheaper... at least its supposed to be cheaper). Yeah, you are missing out on an extremely large framebuffer but I guess the call is whether that's worth the crazy premium.

The 4080 in any flavor is so much faster than the 1080Ti it should provide quite the upgraded experience.

 

[dr1337]

To be honest, I don't really understand what you are trying to prove with this, since moonbogg wants to upgrade from a 1080 Ti, which is twice as fast as that 580.

I mean you're the one bringing up launch day issues and pointing to really old reviews when they're shopping for a card that is over a year old. If you really believe that AMD cards have problems for 6 months (they don't thats a myth), why even bother bringing it up now eh? Unless perhaps you're just trying to scare people away from a particular vendor.

A tiny minority of clickbait journalistic outlets stand to gain a lot of money by weaving a narrative. On youtube there are two sets of videos, ones where 7900xtx users are playing VR no problem, and the other where 'journalists' are playing the algorithm.

If anyone just searches "7900xtx VR" all they get are a million thumbnails and titles saying "DID I MESS UP? IS AMD BAD? AMD IS GOOD NOW? :O", But finding gameplay videos with peoples actual performance requires dedicated searching... Its extremely easy in this instance for people to be accidentally misinformed and take in all the clickbait all at once.

You'd think that if it were so horrible for VR before 6 months ago, it would be USERs complaining and not tech 'journalist' outlets covering this. When the 5700xt launched, it had issues, real issues. People talked about them being bad all the time in the pavlov discord.. Yet when I search 7900xtx, its people recommending the card. Nobody in any VR discord I'm in has complained about it either... it's only these people that test a card for one day that seem to have ever had an issue.

 

[Aapje]

I mean you're the one bringing up launch day issues and pointing to really old reviews when they're shopping for a card that is over a year old.

Yes, because you were going offtopic and denying the facts.

A tiny minority of clickbait journalistic outlets stand to gain a lot of money by weaving a narrative.

Except that I saw a ton of complaints by regular people as well, on places like Reddit and on game forums. And I've already provided proof from perhaps the only serious VR reviewer...which you choose to ignore because it doesn't fit your narrative.

You've been posting kneejerk redacted comments all over this thread.

 

[igor_kavinski]

7900 XTX would give you better performance than a 4070 Ti Super.

7900 XT might give you similar or slightly lower performance compared to the 4070 Ti Super.

The Hellhound RX 7900 XTX Takes on the RTX 4080 with 50 VR & PC Games

The Hellhound RX 7900 XTX takes on the RTX 4080 in more than 50 VR & PC Games , GPGPU & SPEC Workstation Benchmarks The ... Read more

babeltechreviews.com

 

[dr1337]

game forums.

Interesting site, never heard of that one.

places like Reddit

Ah yes the website where brands like samsung directly astroturf and votes are completely anonymous. I see a lot of people talking about 7900xtx on reddit too, but they are just repeating what others are saying. And even then if you go into search and try to find actual owners of the card they tend to only have good things to say.

Yes, because you were going offtopic and denying the facts.

My personal experience is off topic? Sharing the fact that what I've seen from the press and on pop-internet is off topic? The real fact is you have been mislead by clickbait spammers that only stand to gain a profit. But so are many people every day, the algorithms are easy enough to game and "IS AMD BAD" gets more clicks than "here is my experience".

If anything you're proving my point, people will crawl out of the woodwork who seemingly don't even own a 7900xt or AMD card telling the OP why its a risky choice, because apparently it was bad 6 months ago (even tho thats evidently debatable with clips of youtube gameplay). Edited: Like I am not even trying to ignore your source, I'm just saying it was debatable back then and not even relevant now. You're literally the one that brought it up, in response to my personal experience... but sure I'm off topic... You directly ignore my experience and sources because they don't fit the 'forprofit tech reviewer' narrative after all.

But go ahead and keep moving those goal posts.

 

[Aapje]

Interesting site, never heard of that one.

Now you are obviously trolling. You are not worth my time.

 

[GodisanAtheist]

7900 XTX would give you better performance than a 4070 Ti Super.

7900 XT might give you similar or slightly lower performance compared to the 4070 Ti Super.

The Hellhound RX 7900 XTX Takes on the RTX 4080 with 50 VR & PC Games

The Hellhound RX 7900 XTX takes on the RTX 4080 in more than 50 VR & PC Games , GPGPU & SPEC Workstation Benchmarks The ... Read more

babeltechreviews.com

- That review is shortly after the launch of the cards in 2022.

Presumably there have been some kind of driver level VR improvements since then?

What is fascinating is how difficult it is to find a proper VR performance review. Even the bigger sites don't spend much time on it. Sort of speaks to the VR market's overall popularity me thinks.

 

[igor_kavinski]

Sort of speaks to the VR market's overall popularity me thinks.

Which is a real shame coz VR rocks! Nothing like actually losing yourself in the immersiveness of a game or even just a 2D movie playing on a large virtual screen just inches from your eyes. Been using the Vive Flow to watch movies daily now. Despite the lower resolution, it's more engaging and some movie scenes look spectacular. Not sure why but maybe coz they are meant to be viewed on a really large screen.

 

[Aapje]

Presumably there have been some kind of driver level VR improvements since then?

AMD had VR improvements in the release notes for 23.7.1. I've seen a lot of people who are happy with it now, but also a bunch of people who still have issues.

The difficulty is that it's hard to know people's perspective. Some people are more sensitive to microstutter and some people are happy with better performance than their much weaker card, even if the new card underperforms for the price. Then again, you can have people that have poor settings or other issues and falsely blame it on the GPU. That's why proper reviews are so helpful.

And of course it's also a matter of use case. If you are into Beat Saber, then there should be zero issues since a 7900 XT/XTX is just total overkill for that game. With DCS and MSFS you can never have enough power and maxing out a video card tends to more often expose issues.

What is fascinating is how difficult it is to find a proper VR performance review. Even the bigger sites don't spend much time on it. Sort of speaks to the VR market's overall popularity me thinks.

Mainly the lack of popularity of PCVR outside of simming, which is a niche market. VR now mainly seems to be a standalone market. More used like a console than together with a PC.

But I think that a major factor is also that there is a big gap in costs. The standalone games are mostly optimized and designed for the phone chip in the Quest 2, but if you get into PCVR, then suddenly you are dealing with games that are very demanding in themselves like DCS and MSFS, which are even more demanding in VR. And for those games you have a substantial group of people who are willing to spend a lot. So the standard for VR in DCS and MSFS is more cards like a 4090/4080. A 4070 is considered a budget card in that community, even though it is still way more expensive than the price that most people are willing to pay for a GPU.

There is also not really a great choice in VR headsets for PCVR. The Quest 2/3 are more optimized for standalone. Headsets like the Pimax Crystal have a lot of issues and are way more expensive.

So there is a substantial barrier for more casual VR users to get into PCVR. With all the tinkering that is often needed, the GPU prices, the lack of a cheaper headset that is also very good in PCVR, etc; I understand fully why people tend to just stick to standalone games.

 

[CP5670]

The G2 is cheap now and still one of the best headsets for sims and seated games, which I think is all we are going to get on PC VR from big AAA developers going forward. There are some great indie titles and Alyx mods though. Otherwise everything is moving to standalone VR.

I'm not a huge simmer, but like racing and space games in VR. I would say the 4090 is the first card with truly fluid VR performance in these games, and nothing else I've previously had met the bar.

 

[Aapje]

The G2 requires WMR, which will be end of life in two years though.

The Quest 3 has pancake lenses, a bigger sweet spot, the option to run wireless, a 120 Hz screen, far better controllers, no WMR and the option to run standalone (which you may not rate highly but my own plan was to use my Quest 2 for PCVR, but it turned out that I love certain standalone games a lot). It can also use a high-quality USB cable instead of the G2 cable which is specific to that headset (which seems to no longer be produced, so if it breaks...).

Downside is no displayport and it being pretty much required to replace the headstrap.

I have a hard time seeing the G2 as the better buy over the Quest 3 for pretty much everyone, given that the Quest 3 seems equal or better on almost all fronts.

 

[igor_kavinski]

There is also not really a great choice in VR headsets for PCVR.

I think the Valve Index is a good bet for PCVR. Unfortunately, price is too high.