VR Performance Question (HTC Vive)

Bluelight,

decided to reply to your inquiries in this thread as it's more appropriate :).

 

 

The Tbilisi area, I'm good about 90%+ of the time, as you can see from the video's I've posted "Tbilisi River Run". This is the area I've done most of my flying to date.

 

But, Batumi, is more so in the "reprojection" territory and even more so when my field of view completely entails only the urban/city developed areas of Batumi. While the reprojection dose keep the fps locked at no lower than 45fps 90%+ of the time in this area, which is playable, it does causes a "ghosting effect". Actually I didn't notice the ghosting effect until mentioned by others on the forums.

Why Tbilisi is mostly 90fps and Batumi is less :dunno:, we'll have to ask Wags or Chizh I guess...

 

At times, on the Tbilisi airfields I do get the reprojection drop also, but I've not been able to recreate this consistently in the same manner and place yet, but at Batumi it's consistent. So I do not know what the root cause is. But I'm sure the Eagle Dynamics team is knowledgeable of the issue and it's on someone's honey do list.

 

And as stated, shadows in VR are a fps killer, plane and simple..:cry:

 

No, haven't messed with MSI afterburner yet for stats, just not enough time, but it on my list of things I might get too. :juggle:

I'm using 1.5.3 release version, exclusively, although again, haven't got to messing with the others yet, but on my list.

And haven't tried any multiplayer in VR yet...

To be honest, I enjoy doing "pleasure flights" so much in the World of DCS in virtual reality, it's taking me forever to get to exploring all the other DCS World things and experiences with my limited DCS World time!!

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

 

Ok, just tried the "renderTargetMultiplier : x.x" edit of the "steamvr.vrsettings" file, and wow, does it work!!:thumbsup:

Man I can just about make out the bikini babes on the beach!!

 

But, it is a huge performance hit. With it activated my fps in the Tbilisi area in now a very solid reprojection locked 45fps. But it is still very doable, enough that I'll likely keep it activated, as the cockpits and DCS World scenery are really beautiful with it on.

In the Batumi area, it'll drop below the 45fps on occasions, but mostly 45fps, so it's still usable.

 

I've tried 1.5, 2.0 and 2.5 variables.

2.5 is a no-go,

2.0 is OK for performance, but the cockpit and scenery are just beautiful.

1.5 is better performance, visuals are still very much better than 1.0.

I'll likely settle somewhere between 1.5 & 2.0 after playing around with it.

Awesome to hear and thanks for the response! Are you running high medium or low type of settings? Don't want to bother you but I would love to know what settings you are using to I can sort of see if thats what I have already.

One thing I was curious about were these Starway textures I keep hearing about. Are they helpful or hurtful to the fps or do you know?

I figured they would be bad cause they are higher resolution but maybe they are better optimized.. someone suggested using them with VR but I dont think he has VR so Im not sure of how it would affect the fps. :)

Hey guys just a heads up... I bought NTTR this weekend and I tried 2.0 in VR finally. I have tried the vegas tour and what I am finding is that no matter how low my settings are, I am getting 30fps just sitting in the huey before even taking off. The big city is slamming my fps bad in 2.0. I have messed with so many settings but nothing seems to really help it. It just seems those big cities are what is doing it for me and considering I just paid $49.99 for Vegas... I have to say i am pretty down in the dumps about it.

I wish there was some way to get atleast a solid 45fps in reprojections with this.

As it stands now, it seems 1.5 is working better than 2.0.

Ok so I just found out something interesting! Before when running in 2d, I tested DCS 1.5 using MSI afterburner and did notice that one of my cores of my CPU was running high like 95% which is what everyone was saying and was the reason the gpu wasn't running full.

What I had never tested was MSI afterburner while in VR. So I did so.. both on 1.5 and 2.0.

WEIRD thing is that I loaded up problem areas on both versions (2.0 was vegas tour in the Huey and 1.5 was the Harbor tour in the Huey)

Guess what.. my CPU never got about around 65% while in VR and my GPU usage was anywhere from 40% to 60%.

This makes NO SENSE to me.. BTW these problem areas are 30 fps or LESS in VR. The Vegas Tour pretty much is 30 all the time and the Harbor tour is 18 fps when looking at the main city that is next to the harbor.. everywhere else is 45 or 90.

All this time I was thinking the fact that it only uses one main core of the CPU maxed out was why this stuff was happening and why the GPU wasn't running full speed but now I have no idea why I am only running half my GPU power if it isn't bottlenecking.

In both of these tests I had the Nvidia Control Panel set to "let application decide" so DCS should be in full control. I'm baffled and irritated that I can't get atleast 45 fps in these towns, even on low settings.

Wow wonder if everybody else using the Vive has same problem hope others with it comment. What Gpu and processor do you have?

Wow wonder if everybody else using the Vive has same problem hope others wiIh it comment. What Gpu and processor do you have?

I7 950 oc to 4ghz and gtx 970 FTW. Neither running at more than about 50%. it made more sense when I thought the one cpu core was running at 95-99% which it was in 2d, but this makes no sense to me.

When it comes to being stuck at 45 fps in VR, I know I'm not alone but when it comes to being under 30 fps in cities with this lack of gpu and cpu usage, I feel sort of alone right now.

People putting up performance details without showing at least what CPU and GPU they have is a total waste of time.

Get them added to your sig please.

970 is probably at the low end of acceptable HW for VR. But you should give the beta a shot. I just read from another user that performance for Vive got a boost.

970 is probably at the low end of acceptable HW for VR. But you should give the beta a shot. I just read from another user that performance for Vive got a boost.

2.0 is a beta and does the same thing but I guess u mean 1.5.4 beta. It should be releasing soon so I was waiting for that. And it's true that the 970 is on the lower end for VR.. But what difference does that make when neither the cpu or gpu is being used past 50-60%?

People putting up performance details without showing at least what CPU and GPU they have is a total waste of time.

Get them added to your sig please.

Responding to a thread with a comment like that when you clearly didn't read the first post is a waste of time. If you had of read the OP, you would have seen my entire rig details.

That behavior is typical of DCS since always. When the FPS sink it's because the CPU and the GPU are idling and cold waiting for something to happen instead of them being topped out. Without more information, I've always thought that it's dued to information flow between the CPU, GPU and the RAM. They sit idle waiting for data transactions to finish most of the time and then consume it in a very short moment, and then back to waiting.

That would explain why overcloking has such a positive impact on performance in DCS. More than switching to a better GPU alone.

That behavior is typical of DCS since always. When the FPS sink it's because the CPU and the GPU are idling and cold waiting for something to happen instead of them being topped out. Without more information, I've always thought that it's dued to information flow between the CPU, GPU and the RAM. They sit idle waiting for data transactions to finish most of the time and then consume it in a very short moment, and then back to waiting.

That would explain why overcloking has such a positive impact on performance in DCS. More than switching to a better GPU alone.

But why would the CPU/GPU activity not spike when you look at a populated city like Vegas vs looking away from it at a desert. In that case, the FPS changes drastically and the CPU/GPU never actually kick in at all it seems. The more I see how this sim works the more I can't understand why they haven't optimized or fixed these things already.

I understand them resisting changing the engine to work with more than 2 cores.. fine. But when the sim isn't even utilizing all of one core and half of the gpu.

But why would the CPU/GPU activity not spike when you look at a populated city like Vegas vs looking away from it at a desert. In that case, the FPS changes drastically and the CPU/GPU never actually kick in at all it seems. The more I see how this sim works the more I can't understand why they haven't optimized or fixed these things already.

 

I understand them resisting changing the engine to work with more than 2 cores.. fine. But when the sim isn't even utilizing all of one core and half of the gpu.

When you look at a city the flow of information rises and the CPU and GPU begin the waiting game instead of being overused.

Trying to understand things too.

Parallel programming is complex, requires a very different approach, and has a similar problem: when two or more threads need to interchange data they get halted until that data is generated decreasing the theoretical performance.

I don't know how things will evolve. We're already in a situation where big enough software is almost an unmanageable problem. Adding parallelism makes things even worst.

I posted a new thread with videos of what i was talking about here. I decided to post it in a new thread as I just discovered this after starting this one and it applies not only to VR (although more importantly to it.)

http://forums.eagle.ru/showthread.php?t=169738

Responding to a thread with a comment like that when you clearly didn't read the first post is a waste of time. If you had of read the OP, you would have seen my entire rig details.

If you look through this thread and many others, you will find that a lot of people comment on their performance and make no mention of their Spec making their comments virtually useless to others trying to gauge relative performance.

It takes two minutes to make a change to ones profile signature to show at least CPU and GPU.

That is the point I was making.

If you look through this thread and many others, you will find that a lot of people comment on their performance and make no mention of their Spec making their comments virtually useless to others trying to gauge relative performance.

It takes two minutes to make a change to ones profile signature to show at least CPU and GPU.

That is the point I was making.

I did add my details to my signature after you said that.. but you could have been less rude in the way you conveyed the suggestion IMO... maybe u didn't mean it to sound sarcastic thats just how it read to me.