P3D V2.3 Beta 2 Testing

[abranpuko 369]

Hello...

maybe is easy: some people need we spend money. All is invented, and Prepar3d can really go very well...But, if having any problem, forum wouldnt be there and people wouldnt be spending for better computer.

Show bussiness...

[carlito777 712]

Help! Where did all these trolls talking nonsense actually come from? And as mentioned above: Give Rob a break. He spends so much time on all the beta testing, video making and keeping us in the loop that he shouldn't spend the rest of his time answering all those nonsense posts. And the initial thread has really gone off topic on the last couple of pages...

[dihelson 1]

Help! Where did all these trolls talking nonsense actually come from? And as mentioned above: Give Rob a break. He spends so much time on all the beta testing, video making and keeping us in the loop that he shouldn't spend the rest of his time answering all those nonsense posts. And the initial thread has really gone off topic on the last couple of pages...

 

Could you please, list the names of the nonsense people ? I think all deserve the right to talk and express their point of view without to be punished or ridiculed for that.

[Guest]

Rob, many many people, me included are asking the LOD RADIUS control back.

 

I did inquiry and the answer was not going to happen any time soon ... maybe in the future but the task to change this is not trivial at all.  Because of how tessellation is implemented there is a hardware max of 2048 for the length of a texture array.  In order to increase the LOD RADIUS there would need to be multiple texture arrays ... the tessellation shader does the array lookups so that that adjacent "tiles" stitch together correctly.  In order to change/expand this limit there would be significant work needed to implement multiple arrays and those arrays would have some duplicate textures (not efficient and serious performance implications).

 

So tessellation brings some good and some bad and unfortunately one can't re-work the design to accommodate tessellation On/Off -- otherwise LM have to code and manage two very different code paths for rendering which would make maintenance very time consuming.

 

What LM have done is increase the LOD level out further in the LOD RADIUS ... so LOD level 10 (more detail textures) will now go out further with a lower LOD RADIUS value ... so you are getting more detail texutres out further but never beyond 6.5 as that's limited by tessellation (hardware limit).

 

Cheers, Rob.

 

EDIT: this also ties into the key issue of "distance" ... that's what separates a Flight Sim from "the rest".

[Richard Sennett 2,499]

I did inquiry and the answer was not going to happen any time soon ... maybe in the future but the task to change this is not trivial at all.

 

Holy crap awesome driving brother - thanks for sharing that video - very cool   

Edited August 5, 2014 by n4gix
Please! Do not quote the entire post you are replying to!

[Guest]

 

 

I run P3D with 3 monitors on a th2go (5760x1080x32). It hurts to look on the left monitor (vc mainly on center and right monitor with only less scenery).

 

Have you tried running P3D using 3 connected computers with each computer driving a display?  I haven't done this myself, but I have seen it done.  To achieve your goals I think you are going to need to hope nVidia provide driver support and SLi support.

 

2.3 has improved the time delay between frames consistency issue considerably ... however, certain add-ons can and will increase loads such you that could be sent back to having "long frames" ... it will really depend on your hardware and graphics settings and Add-ons and how you configure those add-ons.

 

The GPU and CPU are being utilized, I've provided other videos showing the utilization ... if you expectations are 100% cpu utilization on all cores, that's not going to happen (unrealistic to think that would ever happen) ... best case will be around 50-60%.

 

Cheers, Rob.

[niyoko 7]

Rob, you've done an amazing job informing and enlightening everyone with your posts and replies. You've also done an amazing job sticking with this thread, trying to answer and respond back to the questions and comments. 

I've got Prepar3D v2.2 running fairly well, but still trying to find my sweet spot with the settings. I run locked at 30 fps in my settings now, but have been locked at 20fps for ages. I had no problem with 20fps and actually prefer it slightly more than 30fps in Prepar3D or FSX. For me 30 or higher is just a number or bragging right. I don't see any significant difference between 20fps and going higher than 30fps. 15fps and below is where I see input lag and things get bad, but not unplayable.

[Guest]

15fps and below is where I see input lag and things get bad, but not unplayable.

 

That's my experience also ... on extreme dense areas with some very complex airports and scenery I try to aim at 20 fps or higher.  Input lag below 20 fps starts to be a problem for me.

 

I consider my graphics settings and resolution on the higher end so it's relatively painless to drop an item or two down when I hit those problem locations/add-ons.  2.3 has certainly improved the situation with Beta 2 (which is NOT final).

 

My hunch (and that's all it is) is those last few "long frames" I might see here and there will most likely be solved with an nVidia profile -- I keep hoping we see something from nVidia because just about every other title on the market including "goat simulator" has a profile specific to their software.  (sorry I have no idea if AMD is involved also)

 

Cheers, Rob.

[niyoko 7]

That's my experience also ... on extreme dense areas with some very complex airports and scenery I try to aim at 20 fps or higher.  Input lag below 20 fps starts to be a problem for me.

 

I consider my graphics settings and resolution on the higher end so it's relatively painless to drop an item or two down when I hit those problem locations/add-ons.  2.3 has certainly improved the situation with Beta 2 (which is NOT final).

 

My hunch (and that's all it is) is those last few "long frames" I might see here and there will most likely be solved with an nVidia profile -- I keep hoping we see something from nVidia because just about every other title on the market including "goat simulator" has a profile specific to their software.  (sorry I have no idea if AMD is involved also)

 

Cheers, Rob.

 

I wonder if AMD drivers are having the same problems as nVidia's. I came from using a nVidia GTX460 with FSX a year ago to using an AMD R9-270X. I forgot the version number, but there was version that nVidia release last year, perhaps, that had a noticeable positive impact on FSX's performance for me on my machine. I do think it's to far off to think the driver is at fault to some extent. I haven't done any tweaking like which is done with nVidia Inspector for my AMD card.

 

At the moment I don't hear much negativity from AMD users about driver problems with Prepar3D. I do recall that in the early days of DX11 AMD had a tessellation performance deficit to similar nVidia cards, but I believe the problem has been fixed. The performance issues I'm getting are more CPU related as I'm also running an AMD CPU at 4.2GHz, so my single threaded throughput is not as good as it could be. I do seems to get better performance more consistently in Prepar3D verses FSX.

 

All I can say is that it seems fine to me so far on my AMD system. The only problems I'm having are the graphical issues related to the recent v.1.010 release of MajesticSoftware's Dash-8-Q400. 

[carlito777 712]

2.3 has improved the time delay between frames consistency issue considerably ... however, certain add-ons can and will increase loads such you that could be sent back to having "long frames" ... it will really depend on your hardware and graphics settings and Add-ons and how you configure those add-ons.

 

 

That sounds good. Have you done that frame time analysis you did for the previous beta? The one where you showed the graph with the frame times for one second of game play. Would be really interested to see some of those results if you have something readily available that you can post. Of course, no need to put extra work into it if you haven't done it already. Don't want to overstress your time as others do.

[ChaoticBeauty 6,692]

 

 

I do recall that in the early days of DX11 AMD had a tessellation performance deficit to similar nVidia cards, but I believe the problem has been fixed.

 

AMD GPUs are still somewhat weaker than NVIDIA GPUs when it comes to tessellation (even the latest cards), this is hardware-dependant.

[vr1 0]

Hi Rob,

 

unfortunately the quoteoption is out of order on my computer at the moment just as every past and copy.

 

 

Post #306: multiple Monitors ...

 

 

I could only test with 2 nearly similar computers. Left view driven by one and other views (center and right) driven by the second one. No change in the effect that you called "long frames".

The only benefit is a less distorted sideview.

 

But as I said, the effect is also significant seeable on every single display setup. It is possible, that addons achieve the effect. But the effect is also significant on a vanilla installation with all settings at the lowest levels with default beech, mooney or other GA.

 

It's not the question, that P3D is playable even on less fps. But we are talking about a simulationplattform explicit apart from a game. And that's disappointing because you only get unrealistic views and the sideview stutters causes errors in perception and distract attention away from flying especially on flight training and FTC's.  

 

 

 

 

Post #306 cpu/gpu utilization

 

I'm not sure if you understood me right. My expectation is higher workload over urban areas especially on lower altitudes, thus on take off or approach, causes higher cpu/gpu utilization and if the limit is raeched or exceeded the overload will necessarily reduce the FPS while the cpu/gpu workload / utilization remains on high (highest) level.

 

But against all expectations the contrary is the case.

 

Inflight an on the beginning of the descent the cpu/gpu utilization is around 60-80% depending on the weather. Until FL above 10000 ft AGL everything is fine. Below approx. 10000 ft AGL the utilization and FPS drops abruptly to two-thirds and below approx. 5000 ft AGL drops again to the half of inflight-utilization.

The result: in the most important part of the flight you get the biggest input lag with only 16-18 fps on 30% cpu/gpu usage.

 

This makes no sense. That couldn't be the goal for a *simulation plattform*.

 

And there are many threads about this "less fps AND cpu/gpu-utilization" without statement from LM.

 

As I said before, I noticed this since the update to version 2.2. I tried the update 2.1>2.2 at first and also later a vanilla installation without difference.

 

 

With other words:

I expect and accept that every workOVERload will drop fps or/and smoothness because the workload raeches the cpu/gpu-limits.

But this is not the case. It seems to me that LM willfully reduces the workload by intern limits to fix the oom/vas-problem. Vice versa there is no other sense for me in the current workUNDERload during approach with "high end" computers.

 

Your Rampage IV 3960X @ 4.8 with Titan 6 Gb win 7 64x is quite equal with my system. Marginal differences are only the motherboard and cpu (Rampage black, 4930k @ 4.5).

 

And your observation are also 14/15 fps approaching urban areas. Have you compare your cpu/gpu utilization on these parts of flight against inflight on higher FL? I bet you could observe the same divergence.  

 

 

 

 

 

 

 

 

 

And by the way, sorry that everybody is beating your ears. Unfortunately it got quite around LM's supportstatements.

That's the reason everybody calls on your services  

 

Me too. Looking forward to your influence on development. B)

[Juicehead 16]

Help! Where did all these trolls talking nonsense actually come from? And as mentioned above: Give Rob a break. He spends so much time on all the beta testing, video making and keeping us in the loop that he shouldn't spend the rest of his time answering all those nonsense posts. And the initial thread has really gone off topic on the last couple of pages...

 

No one here is trolling, they are just asking valid questions and making valid comments about the current state of P3D.  Framerate has as much to do with V2.3 as any other optimizations/fixes, so I disagree with you when you say the last couple pages have been off topic.  Whether or not Rob decides to respond to questions or comments are his business, not yours.

 

I am fairly sure most people here are thankful the info/videos Rob provides, not just you.  

 

 

So lighten up a bit...

[Christopher Low 5,353]

I am so glad that I have stuck with a 19" monitor and 1280x1024 resolution. In addition to the power benefits (ie. I don't need SKYNET to run the graphics), I also don't have to worry about increasing the LOD radius to some silly value. Based on what I have seen, I honestly can't see why anyone would need a LOD radius higher than 6.5.

[AviatorMan 253]

The problem with Aerosoft airport runway lights --- is this addressed in the v2.3? or is this an Aerosoft problem?