Ground textures gets more blurry the further I fly

[SteveW 1,792]

...in P3D it's worse because you have a desktop window and the timing is based on the desktop not the sim. Whereas FSX is using exclusive mode and the lower graph shows there's an improvement available although slight..

[Adamski_NZ 948]

Steve - what did you use to get the graphs? Looks handy!

Looks like setting Vsync via NVI is a slightly better bet than in P3D ... is that correct?

Adam.

[SteveW 1,792]

Hi Adam,

Unfortunately we can't use the NVI to such good effect as we did with FSX 30 - 1/2 vsync - stick to P3D internal settings. Graphs from my commercial tool Ideal Flight.Graphs from P3D show the same story. The monitor won't store excess frames and they are discarded.

[Murmur 3,941]

2 hours ago, SteveW said:

No.

 

Here's a graph of around 20fps so it shows up better:

 

As you can see pushing an extra frame or two simply induces stutter and takes up process time that is wasted.

In your graph, there's not the "19 fps limit" that you are suggesting is better. Do you have the graph for that? Also, do you have longer graphs? Those above cover only about 2 seconds, if I interpret them correctly.

[SteveW 1,792]

Why do you need more than  2 seconds? That's 40 frames - right? Also I'm suggesting, quite obviously, it is for you to try 59 or 60 (or 20 or 29 whatever you are doing) to see which is best on your rig. The point you clearly missed is that setting too high a frequency is lost processing. Does that help?

[SteveW 1,792]

...on top of that - there's nothing to stop anyone doing their own tests and making their own graphs. I even supply professional tools to do it so there's no excuse making guesses about this technology.

[SteveW 1,792]

23 minutes ago, Murmur said:

In your graph, there's not the "19 fps limit" that you are suggesting is better.

I see - I believe I suggested to try 29 rather than 31, but 30 might be OK who knows untl it's checked. I was showing 20 for a 21, in the graph so where's 19 come from? At 60Hz a lot more frames, you can see often 59 might work better than 60 there, maybe not 19 for 20 as that's slow for that case. Certainly not 21 for 20 or 31 for 30 or 61 for 60 - anyway - you get the idea hopefully.

[Virtpilot 151]

2 hours ago, SteveW said:

Hi Adam,

Unfortunately we can't use the NVI to such good effect as we did with FSX 30 - 1/2 vsync - stick to P3D internal settings. Graphs from my commercial tool Ideal Flight.Graphs from P3D show the same story. The monitor won't store excess frames and they are discarded.

Regarding vsync, does it have any effect to use D3Doverrider? I have pointed Prepar3D.exe to D3Doverrider using vsync but not triple buffering. Can´t see any harm at least, maybe some benefit instead. Or then I´m just hypochondriac about it...😄

[SteveW 1,792]

I don't think that can do any more than P3D already does. The Triple Buffer simply means there's a more consistent time between frames maintained and we can set that in P3D.

[Virtpilot 151]

Ok, thanks for your answer, Steve!

[Michael Moe 494]

2 hours ago, SteveW said:

I don't think that can do any more than P3D already does. The Triple Buffer simply means there's a more consistent time between frames maintained and we can set that in P3D.

Hi, is this a mostly CPU or GPU hog? 

If GPU maybe my SLI could work some more in daytime.

Thanks Michael Moe

[SteveW 1,792]

Hey Michael, you mean using the TB? Uses 1 frame more GPU memory than standard backbuffer rendering, and it's going to be rendering 'each next frame' with less delay on average so uses a little more CPU.

Edited August 12, 2018 by SteveW

[Adamski_NZ 948]

11 hours ago, SteveW said:

Graphs from my commercial tool Ideal Flight.Graphs from P3D show the same story.

For those interested in "Ideal Flight" for P3D, the link is here:
http://www.flight1.com/products.asp?product=idealflightpro4
or here: http://www.codelegend.com/idealflight/default.htm

Not cheap, but it does appear to have a bucketload of features (over and above the diagnostic graph-making).

Edited August 13, 2018 by Adamski_NZ

[gboz 157]

On 8/10/2018 at 2:41 PM, Rob Ainscough said:

 

 I tested on 3 PCs

7700K 1080FE SSDs (2160p @ 60Hz capped to 29.5 via NI)
8700K 1080Ti  SSDs (1440p @ 60Hz capped to 29.5 via NI)
7900X TitanXP  SSDs (2160p @ 30Hz) 

 

Why are we limiting in NI.?

Two years ago Beau was advising not to externally limit:

http://www.prepar3d.com/forum/viewtopic.php?f=6312&t=120133&p=139706#p139706

" Lets say you're actually getting 60fps but externally limiting down to 30. The extra 16ms is being wasted sleeping rather than doing terrain loading requests. The part of the system calculating the fiber time percentage thinks you're running at 60fps because it's counting time form the start of the frame rather than the total time of the previous frame."

gb.

[Michael Moe 494]

8 hours ago, SteveW said:

Hey Michael, you mean using the TB? Uses 1 frame more GPU memory than standard backbuffer rendering, and it's going to be rendering 'each next frame' with less delay on average so uses a little more CPU.

Thanks Steve 🙂

Michael Moe