Hows that Nvidia Frame limiter working out?

[SteveW 1,792]

Analysing the frame rate output from the sim shows the output frequency varies if limited or V-synced.

Instead, the internal Locked fps slider control in Prepar3D Display Settings works differently. Each next frame is calculated for the exact time ahead at that rate. Every movement is worked out to be where it shall be then. The system then attempts to maintain that rate with the help of a look-ahead buffer, that depletes to provide timely frames. Look at the trace output from those settings shows a flat line. That's what's needed for online time coincidence.

 

[SteveW 1,792]

Thanks Dave. This is a bit of a thankless task!

So although quite often the internal frame rate lock slider is referred to as a Limiter, it is not any kind of limiter at all. Because it need not be limited at all as it is an output at a definite frequency. When we use unlimited and attempt to limit externally either by a frame rate cap, or by a VSync hold, each next frame is offset by the cap. Sometimes this can become a sawtooth, particularly with setting unlimited in the P3D Display, and setting half refresh in the NCP.

[CFIJose 5,485]

SteveW,

How is it possible that with 442.19 win 10 driver I'm now able to run P3D v4.5HF2 with unlimited fps, vsync off and with the external limiter set at 31 fps without any stutters or tearing? 

Not only that but I always had blurry textures with a setting of unlimited. What has changed? Is it the latest P3D v4.5 of this driver?

Simply amazing!

Jose

[SteveW 1,792]

Well the worst you can be when limiting is no more than a frame either way, so as I mentioned this doesn't produce stutter only elongation of frame times. Stutter is more a symptom of fill-in when new scenery is coming in. So that may be something to do with the new driver if you've not made any other adjustments. I didn't see any changes other than the frame limit extended down to 20.

[SteveW 1,792]

You would likely improve on what you have limit at 30 or 29. No need to limit over what you want if that's 30. As Dave mentioned it's a bit less than 30 so 29 limit might be more effective and is 6% less demand.

[SteveW 1,792]

The external limiter seems mis-understood. It is a cap on frames arriving early, it is not some mechanism to induce more fps. So when I was testing 61 limit external I was averaging 58fps. When I was capping at 58 I was averaging 58.

[DaveCT2003 1,820]

1 hour ago, SteveW said:

You would likely improve on what you have limit at 30 or 29. No need to limit over what you want if that's 30. As Dave mentioned it's a bit less than 30 so 29 limit might be more effective and is 6% less demand.

Yeah, it's been a very long time and I don't recall exactly why it's 29.97, I believe it was mathematical in origin, but I might not be remembering that correctly.  Things change so fast and I'm getting so old... LOL!

 

 

[DaveCT2003 1,820]

I have no looked into the new NVIDIA frame rate limiter yet (it's why I'm monitoring this thread) and you guys are light years ahead of me when it comes to this.... but I've been using the MSI Afterburner/RivaTuner limiter for a while and it's been, well, incredible.  I've never had frame rate or smoothness issues in flight sim, but man even I can tell a difference with MSI/Rivatuner.

But I'm looking forward to learning more about what you guys are working on.

 

 

[bbuckley 133]

5 minutes ago, DaveCT2003 said:

I have no looked into the new NVIDIA frame rate limiter yet (it's why I'm monitoring this thread) and you guys are light years ahead of me when it comes to this.... but I've been using the MSI Afterburner/RivaTuner limiter for a while and it's been, well, incredible.  I've never had frame rate or smoothness issues in flight sim, but man even I can tell a difference with MSI/Rivatuner.

But I'm looking forward to learning more about what you guys are working on.

 

 

I've been using RTSS Scanline sync x/2 -1 for quite awhile and it's been awesome. I tried the NCP limiter but didn't seem to be as smooth. It was only a flight or two and I went back to RTSS also because I use MSI/RTSS for onscreen stats display when I want them, so if MSI/RTSS is going to be running anyway for the stats when I test then using it for scanline sync seems natural.

[DaveCT2003 1,820]

10 minutes ago, bbuckley said:

I've been using RTSS Scanline sync x/2 -1 for quite awhile and it's been awesome. I tried the NCP limiter but didn't seem to be as smooth. It was only a flight or two and I went back to RTSS also because I use MSI/RTSS for onscreen stats display when I want them, so if MSI/RTSS is going to be running anyway for the stats when I test then using it for scanline sync seems natural.

"Oh dat's right!".  LOL!  That's exactly what I do and for the same reasons.  Hard to beat!

 

[TurboKen 221]

14 minutes ago, bbuckley said:

I've been using RTSS Scanline sync x/2 -1 for quite awhile and it's been awesome. I tried the NCP limiter but didn't seem to be as smooth. It was only a flight or two and I went back to RTSS also because I use MSI/RTSS for onscreen stats display when I want them, so if MSI/RTSS is going to be running anyway for the stats when I test then using it for scanline sync seems natural.

I may have to try this scanline sync thing you are talking about. As I also use the program for real time stats.

[CFIJose 5,485]

2 hours ago, SteveW said:

You would likely improve on what you have limit at 30 or 29. No need to limit over what you want if that's 30. As Dave mentioned it's a bit less than 30 so 29 limit might be more effective and is 6% less demand.

I'm going to set it to 30.

RTSS was a good option at 1/2 refresh rate, but sometimes when I was trying to load a flight or flight plan in the menu, P3D would pause to load the planner. Sometimes I had to press again and that resulted in a double load of the planner. I noticed this right away. Tested RTSS extensively and it was the only small issue that affected my sim.

The Vsync on & no tearing is still perplexing when tearing was always an issue for me when I did NOT enable Vsync.

[Twenty6 693]

22 hours ago, Avidean said:

Hmmm, I doesn't seem to work at all on my system. I installed 442.37 and set the Max Frame Rate, under Manage 3D Setting to 29fps for Prepar3D.exe and in the game I am getting in the 50's with or without Vsync and TB

NCP > 3D Settings > Adjust image settings with preview. Make sure to:

Bullet the middle selection "Use the advanced 3D image settings" > Take me there

[CoolCaptain 14]

Using a non 4K 60Hz display, I set the internal frame rate limiter to 60. I tried 30 with Vsync at 1/2 refresh rate but the lower the value I set, the more FPS I lose. 

[bbuckley 133]

On 2/21/2020 at 9:18 AM, DaveCT2003 said:

If you're not familiar with how this plays out... 

Assume Aircraft A and Aircraft B are flying the same flight plan, the same speed and altitude, and the same aircraft (thus the same ground speed), and they both have the exact same weather engines and weather.  Aircraft B starts in trail of Aircraft A by 20 miles.  But for some inexplicable reason, Aircraft B is over taking Aircraft A, and it becomes problematic for everyone especially as the aircraft being to approach an arrival or approach.  The reason?  The difference in frame rates.

Anyway, 30fps (29.97) is the optimal frame rate for flying cooperatively, and since I used the term "cooperatively" I'd like to say that I wish we as a community still referred to flying online with others as "cooperative flight" and that we taught people what that means and the part they play in it.  Alas we are today mostly a collective of individuals who don't care much for anyone else (sad, but I'm not speculating when I say that), so maybe I'm just howling at the moon by sharing this, I don't know.

Flight simulation is not a first person shooter, and pumping your frame rates just as high as you can get them is not advantageous in any respect and does in fact serve to take life off of a lot of systems (relative to how hard you push things and the cooling, of course).  In the end, 30fps is the mark you most of us want to hit.

Hi my friend @DaveCT2003, I absolutely believe you on this but intuitively I don't understand the issue if your FPS exceeds the "real time FPS of 29.97". I can intuitively see how a slower FPS would cause sim calculations to need to slow down the physics to match the scene, but why doesn't a higher FPS just mean the sim essentially displays nearly the same scene in consecutive frames but the physics remains at real time? Is there a paper or something I can read to understand?