Interestingly... we've had the VR Optimization since May

[Quekel]

Nothing here, VR performance is horrible on a 4790K 4.4Ghz and a 1080AMP Extreme.

[dburne]

I find this all so confusing - VR performance optimization did not include the Normandy map yet per Wags, yet I have seen a nice improvement in performance on that map.

[DeltaMike]

Hey folks,

As an outsider getting very interested in VR and a return to flight sims there seems to be a lot of dissatisfaction with the current state of VR and DCS or am I not reading it right? Also, why is there not clear information coming from DCS about VR updates etc? So much reading between the lines and conjecture. Seems ridiculous considering how much people are looking forward to improvements.

 

Well, there are two things.

 

The first is the limitation of VR technology right now. The devices are heavy and hot, prone to fogging up. Field of view is limited. Screen and lens technology are in their infancy. Remember how your mother said, "Don't sit so close to the TV, you'll go blind!" We have these little teeny computer screens right on our eyeballs, we can see every pixel, and we are going blind.

 

That said, the hardware is still way ahead of the software. It takes a lot of horsepower -- a lot -- to pipe VR through two high-res monitors, tiny though they may be. We aren't sure what it will take to drive a high-res, high-field-of-view device like say a Pimax. GPU's aren't really optimized to brute-force software originally designed to run on a single VGA monitor (remember, DCS has been around for like 20 years).

 

I get the feeling this is an active area of research on the software side, how to realize efficiencies on the software level. AMD for example is working with Vulkan on VR optimization. One potential solution that AMD is working on is, dual GPU systems, where each eye gets its own GPU. Which could be kinda slick but the software has to accommodate that.

 

Given the circumstances, I find it amazing that DCS can accommodate VR at all, hats off to the engineers. And they are doing pretty durn well. I was flying around in the Oculus F18 app the other day. It's a native VR app and it looks like crap, DCS is way better.

 

The Rift S is exciting because it inexpensively solves some of our problems. They swapped out the old screen for a new one, which makes instruments easier to read, without flogging your GPU too hard. Basically. So, you can get in the game with realistic hardware, in fact you can have fun in DCS with a 1080 and a modern CPU that can burst to 3.8GHz or so. That's still a lot of computer but at least you don't have to bring a GPU server to the game, which it wasn't that long ago you had to.

 

VR isn't "there" yet. Hence the debate about whether or not it's worth it, hence all the people still flying around on big-screens and Track-IR. For some of us, it's close enough. Personally I find the experience amazing -- I'll put up with the pixels and the low graphics settings -- but you know how it is. YMMV right.

 

I've been doing flight sims since the 80's, kinda lost interest until VR came out. In spite of all the problems, DCS is the killer ap far as I'm concerned. It's both astonishing and frustrating. Guess it depends on how you deal with frustration, and astonishment. (Note, having been to Burning Man several times, I don't use the term "astonishing" lightly.)

[Flummi]

Well Said DeltaMike! I Hope ED will find performance solutions and bugfixes soon, so that the break-through in VR for new VR-Simmers will come.

[Harlikwin]

I find this all so confusing - VR performance optimization did not include the Normandy map yet per Wags, yet I have seen a nice improvement in performance on that map.

 

I think there were also some general improvements (memory leak) and non-vr guys noted improved frames too. So there are at least 2 different things going on (probably more).

 

And the VR fix seems to work for some very lucky guys, and its decentley big, from 45 to 70-80fps in a few cases. But it doesn't seem to be working for 80-90% of people.

Edited June 19, 2019 by Harlikwin

[Johnny Dioxin]

On the whole though I'm happy cause I really wasn't playing DCS recently due to the bad performance.

For me its more like something got broke and fixed, rather than some revolutionary new update but that may be a little unfair.

 

You and me, both ;)

 

Went from an addicted VR Pilot getting as much stick time as possible in 1.5.8 to leaving it for months at a time when 2.5 dropped (though I was still buying campaigns and the Hornet & F-16 cos it was always just a separation and never a divorce).

 

With this optimisation, my Index on the horizon and my new VPC HOTAS I can see it going back to a full-time hobby. Vows will be renewed :)

[Quekel]

I wonder why they wont just make VR shaders to improve performance like Kegetys did, I would easy give away some nice shaders for better performance with higher resolutions and no shimmering

[imacken]

I wonder why they wont just make VR shaders to improve performance like Kegetys did, I would easy give away some nice shaders for better performance with higher resolutions and no shimmering

 

Kegetys mod does nothing for a lot of people, myself included. I see a very marginal improvement and the visual artefacts are too much to bear!

I believe it tends to help people with lower spec systems.

I guess it is for the reduction in visual quality that ED wouldn’t go down that route.

[Wolf8312]

You and me, both ;)

 

Went from an addicted VR Pilot getting as much stick time as possible in 1.5.8 to leaving it for months at a time when 2.5 dropped (though I was still buying campaigns and the Hornet & F-16 cos it was always just a separation and never a divorce).

 

With this optimisation, my Index on the horizon and my new VPC HOTAS I can see it going back to a full-time hobby. Vows will be renewed :)

 

Well I just flew and am hard pressed to find anything wrong with performance. It really is very smooth for me and this is WW2 dogfighting too which was always problematic in VR fluidity wise.

 

Not only performance but I can't get over how good it looks these days! Granted I always avoid playing at midday or at times when everything looks overly exposed in VR, but still its come on by leaps and bounds.

 

Not only cosmetic changes such as sparks and smoke when firing into enemy planes (which means alot in VR), but the lack of shimmering, and the lighting/color and relfections upon the distant houses, as well as the overall smoothness of the terrain animations, all seem to be vastly improved.

 

I trust you don't need any convincing as to how the modules themselves feel in VR, but take it from me if you havent already get a jetseat!

 

It really is something I couldn't do with out! Just to be able to feel the weapons when firing (which is very cool!) would be worth it alone, but being able to feel the engines/stalls, landing gear, touchdown etc, is just the perfect compliment to the VR experience.

Edited June 20, 2019 by Wolf8312

[hansangb]

I think there were also some general improvements (memory leak) and non-vr guys noted improved frames too. So there are at least 2 different things going on (probably more).

 

And the VR fix seems to work for some very lucky guys, and its decentley big, from 45 to 70-80fps in a few cases. But it doesn't seem to be working for 80-90% of people.

 

 

I just updated my stable release. Because, I tell you what, I'm seeing really smooth performance with Reverb. I'll report in the thread that Wags opened.

[Harlikwin]

I just updated my stable release. Because, I tell you what, I'm seeing really smooth performance with Reverb. I'll report in the thread that Wags opened.

 

We should compare notes, because my system is pretty close to what you have and I'm getting not so great performance with my CV1. I mean its not bad, but it seems like you are doing something that I'm not to get good performance with the reverb.

[hansangb]

We should compare notes, because my system is pretty close to what you have and I'm getting not so great performance with my CV1. I mean its not bad, but it seems like you are doing something that I'm not to get good performance with the reverb.

 

 

I'll collect the data for Wags in the other thread and we can compare.

 

Out of curiosity, are you killing Oculus drivers/services?

 

Have you excluded SAVED GAMES and ProgramFIles\DCS from AV scanning? Anyway, I'll upload the detailed settings and spec to Wag's VR performance help thread.

[Icebeat]

 

AMD for example is working with Vulkan on VR optimization. One potential solution that AMD is working on is, dual GPU systems, where each eye gets its own GPU.

 

This was one old naive idea, fortunately, it was discarded, Nvidia has support for Multi-View Rendering in Turing and Single Pass Stereo in Pascal, why ED doesn't want to implement this is an incognita. The official answer was that ED wants to be platform agnostic, I call it mediocrity. Unity is implementing a similar technique and not limited to Nvidia.

Unreal Engine and Unity are using Hybrid Monoscopic Rendering to accelerate Oculus GO and Quest with very good results.

VR is already there, we only need developers to move to the next level.

We should observe that the game is heavy CPU bounded, maybe they should start fixing that. The whole experience will improve and not only for VR players.

Edited June 21, 2019 by Icebeat

[hansangb]

This was one old naive idea, fortunately, it was discarded, Nvidia has support for Multi-View Rendering in Turing and Single Pass Stereo in Pascal, why ED doesn't want to implement this is an incognita. The official answer was that ED wants to be platform agnostic, I call it mediocrity. Unity is implementing a similar technique and not limited to Nvidia.

Unreal Engine and Unity are using Hybrid Monoscopic Rendering to accelerate Oculus GO and Quest with very good results.

VR is already there, we only need developers to move to the next level.

We should observe that the game is heavy CPU bounded, maybe they should start fixing that. The whole experience will improve and not only for VR players.

 

 

Have you ever had to support four different branches of code? It's not as simple as "just four" Because the permutation of the difference start to get taxing. I would love if DCS supported NVidia. But then what about AMD? But then what about something Vive comes up with, and why not Pimax? Or Oculus specific tweaks? And WMR specific ones? And as programmers are working on all the different branches, who's going to work on the modules?

 

So I can't fault DCS for going with the vanilla middle of the road. You call it mediocrity, but i call it realistic given the constant strain or product managers' wishes vs what engineering can deliver. It always starts with "this should be easy....." and it never is.

[Pinello]

....call it realistic given the constant strain or product managers' wishes vs what engineering can deliver. It always starts with "this should be easy....." and it never is.

 

What the man said.

[remi]

Have you ever had to support four different branches of code? It's not as simple as "just four" Because the permutation of the difference start to get taxing. I would love if DCS supported NVidia. But then what about AMD? But then what about something Vive comes up with, and why not Pimax? Or Oculus specific tweaks? And WMR specific ones? And as programmers are working on all the different branches, who's going to work on the modules?

 

So I can't fault DCS for going with the vanilla middle of the road. You call it mediocrity, but i call it realistic given the constant strain or product managers' wishes vs what engineering can deliver. It always starts with "this should be easy....." and it never is.

 

It's mediocrity, since additional programmer resources used would mean much better experience for a portion of users.

 

I'd rather have a program that gives better performance for certain users than bad performance for everyone.

[Hippo]

It's mediocrity, since...

 

I mostly agree with you although I might try to be a bit more diplomatic about it.

 

OTOH,

 

"I would love if DCS supported NVidia. But then what about AMD?"

"So I can't fault DCS for going with the vanilla middle of the road"

Surely it's a case of balance - how many users benefit / benefit obtained / work required. Just saying don't bother because it's a bit hard just sounds like a cop out.

 

However, I suspect that if the previous poster had cut out most of his post and just left the following:

 

"I can't fault DCS"

it would probably also be a fair reflection of his views (and of the majority of the posters on these forums).
Edited June 22, 2019 by Hippo

[nrosko]

I decided to do the TF-51D benchmark, just to compare my performance to others. I'm getting 70-90 fps.

 

1.5PD in debug 0 in game

Textures High

Terrain textures High

Civ traffic Medium

Water High

Visib Range Medium

Heat Blur Low

Shadows High (3072) normally run 8192 but this was over written after update.

Resolution 3840x2160

Ratio 1.77

res cockpit 1024

MSAA Off

DOF Off

Lense Off

Motion blur Off

SSAA OFF

Clutter 270

Trees 60%

Preload 150000

smoke 5

Gamma 1.5

AF 8x

Terrain object shadow flat

Cockpit global on

[hansangb]

I decided to do the TF-51D benchmark, just to compare my performance to others. I'm getting 70-90 fps.

 

--snippage--

 

Cockpit global on

 

 

The benchmark is to see if there is a difference between the stable and the open beta. Do you get different FPS between the two?

[hansangb]

I mostly agree with you although I might try to be a bit more diplomatic about it.

 

OTOH,

 

Surely it's a case of balance - how many users benefit / benefit obtained / work required. Just saying don't bother because it's a bit hard just sounds like a cop out.

 

However, I suspect that if the previous poster had cut out most of his post and just left the following:

 

it would probably also be a fair reflection of his views (and of the majority of the posters on these forums).

 

I can't see what remi wrote since he's on my blocked list. But as to your comment, remember, there was a time when ATi ruled the GPU market.

 

I think most folks interested in performance will buy whatever card gives them the best experience. I know I am. So if you spend the time and effort to modify your game to support a bespoke solution, what happens if the GPU company changes direction or another company takes a commanding market lead?

 

Efforts like vulkan make sense since it's a multi vendor consortium.

 

I'm all for open and common standards. It allows for larger market penetration that single vendor solutions. For example, Digital, Intel, and Xerox never - on purpose- trademarked Ethernet. They also only charged $1,000 for a block of Ethernet MAC addresses. And the only reason they charged the $1K was to prevent every curios grad student from requesting blocks and blocks of addresses. During the same time, Token-Ring was being pitched by IBM. A few other vendors like Madge licensed it, but it was an IBM tech. When IBM celebrated shipping 1,000,000 TR cards, Ethernet was around 100,000,000 if memory serves. Maybe even 300K cards.

 

So again, I don't begrudge ED for not supporting Nvidia only features.

[Madmatt_BFC]

I don't want to poopoo on this parade but am I the only one that is a little annoyed by how poorly ED handled this whole VR Enhancement Patch with regards to its announcement?

 

I mean first we were told it was coming but needed time, then were told it was going to be in 2.5.5. then we were told it wasn't going to be in due to some last minute incompatibilities and then several days later we get a rather buried post by Wags that yes it was in but surprise, it was actually in months ago. There is no mention of it in any patch notes that I could find and no actual announcement post to let everyone know it was really included. I'm totally fine with the fact that it took a while to get it out, that's not my issue at all, it just seems that a lot of excitement was sort of lost and now I am still not entirely sure what the current situation is.

 

I'm thrilled that it appears to be in our current open beta codebase (at least I think it is?) and I don't want to begrudge ED anything, but this whole matter was handled a bit sloppily. Even now I haven't really seen much else posted about it.

 

In the grand scheme of things, this is just a slight misstep but I do wish comms on this would resume so we know where things stand.

[nrosko]

Seems it was a bit of a Schrödinger's patch.

[volk_50]

I don't want to poopoo on this parade but am I the only one that is a little annoyed by how poorly ED handled this whole VR Enhancement Patch with regards to its announcement?

 

I mean first we were told it was coming but needed time, then were told it was going to be in 2.5.5. then we were told it wasn't going to be in due to some last minute incompatibilities and then several days later we get a rather buried post by Wags that yes it was in but surprise, it was actually in months ago. There is no mention of it in any patch notes that I could find and no actual announcement post to let everyone know it was really included. I'm totally fine with the fact that it took a while to get it out, that's not my issue at all, it just seems that a lot of excitement was sort of lost and now I am still not entirely sure what the current situation is.

 

I'm thrilled that it appears to be in our current open beta codebase (at least I think it is?) and I don't want to begrudge ED anything, but this whole matter was handled a bit sloppily. Even now I haven't really seen much else posted about it.

 

In the grand scheme of things, this is just a slight misstep but I do wish comms on this would resume so we know where things stand.

 

I bet (1) some ED developer tried couple things and got 50% FPS boost -- showed off for a quick second this was picked up further up the chain in excited frenzy.

 

(2) "50%" announcement was made.

(3) they started looking at it in more detail and realized it was more "nuanced".

(4) "dang it" moment for any more public statements/mention of roll-out in beta.

(5) damage control.

(6) we're here now.

 

p.s. great mystery is why GPU is running at around 70-75% and refuses to break higher for more FPS. If ED can figure out how to kick that render pipeline in the rear, that'd be a great start (70% to, say, 95%)

[mdee]

I bet (1) some ED developer tried couple things and got 50% FPS boost -- showed off for a quick second this was picked up further up the chain in excited frenzy.

 

(2) "50%" announcement was made.

(3) they started looking at it in more detail and realized it was more "nuanced".

(4) "dang it" moment for any more public statements/mention of roll-out in beta.

(5) damage control.

(6) we're here now.

 

p.s. great mystery is why GPU is running at around 70-75% and refuses to break higher for more FPS. If ED can figure out how to kick that render pipeline in the rear, that'd be a great start (70% to, say, 95%)

 

You are very likely right. This is very common story, esp. with companies who have weak/inadequate internal testing processes.

 

GPU may not be fully utilized because of CPU being saturated - there are 1000s of other possible reasons though, so can't really tell.

[Hippo]

I don't want to poopoo on this parade but am I the only one that is a little annoyed by how poorly ED handled this whole VR Enhancement Patch with regards to its announcement?

 

Probably not, but then there is a huge pro-ED imbalance in these forums, with many only capable of praise (then there's the moderation...). Personally, I just found the whole thing comical. It is an open beta, so it's unfair to demand too much, and best to take things with good humour.

 

I'm still not sure whether the VR performance improvement that has been promised for months has actually been delivered or not. I wish this would be stated unequivocally (apologies in advance if it already has been).