Is it just me, or is the car traffic in P3Dv4 a lot smoother when you limit to 29 fps instead of 30?  With 30, the cars and trucks are kind of jerking around, while it's smooth with 29. Not that this is so big of a deal, but I am wondering if others see the same.

1 hour ago, Ray Proudfoot said:

This discussion really should be taken to their forum.

I'm not discussing anything here, I just provided their reference.

2 minutes ago, Afterburner said:

Is it just me, or is the car traffic in P3Dv4 a lot smoother when you limit to 29 fps instead of 30?  With 30, the cars and trucks are kind of jerking around, while it's smooth with 29. Not that this is so big of a deal, but I am wondering if others see the same.

As SteveW explains this is because 1 frame is lost by your monitor? Goes quite in line with what he has just said above.

Edited June 28, 20214 yr by Dirk98

8 minutes ago, SteveW said:

If your monitor is 30Hz refresh with vsync the monitor is showing each frame at intervals of 1/30second. Then every frame sent ahead of that interval will be ignored since it is still showing the last frame. Setting to 29 can reduce the number of frames ignored and gives you 3% bandwidth on average. However, the actual frame to frame time is not consistent, you may see 34 or 26 on the display from time to time, you get an averaged result slightly better overall.

Will those 26 to 34 and down again don’t cause stutters ?

4 minutes ago, GSalden said:

Will those 26 to 34 and down again don’t cause stutters ?

See, stutters are also caused by your monitor skipping some frames. It jitters around some set frequency and the task here is to find your optimal number just a frame or two below what you are aiming at (30?) if I get it right though..

Edited June 28, 20214 yr by Dirk98

2 minutes ago, Dirk98 said:

As SteveW explains this is because 1 frame is lost by your monitor? Goes quite in line with what he has just said above.

I am not referring to the overall smoothness, but only to the smoothness of the motion of car traffic even if my monitor's refresh rate is set to a multiple of fps. With FPS set at 30 and monitor refresh rate at 60, the overall smoothness is fine, but the cars stutter. With FPS set at 29 and monitor refresh rate at 58, both car traffic and the rest is smooth.

3 minutes ago, Afterburner said:

I am not referring to the overall smoothness, but only to the smoothness of the motion of car traffic even if my monitor's refresh rate is set to a multiple of fps. With FPS set at 30 and monitor refresh rate at 60, the overall smoothness is fine, but the cars stutter. With FPS set at 29 and monitor refresh rate at 58, both car traffic and the rest is smooth.

The motion of airport vehicles is directly related to FPS if I remember correctly. When they jerk it means some frames are missed.

Edited June 28, 20214 yr by Dirk98

Not airport vehicles, road vehicles.

2 minutes ago, Afterburner said:

Not airport vehicles, road vehicles.

might be the same mechanism involved.

Edited June 28, 20214 yr by Dirk98

11 minutes ago, SteveW said:

If your monitor is 30Hz refresh with vsync the monitor is showing each frame at intervals of 1/30second. Then every frame sent ahead of that interval will be ignored since it is still showing the last frame. Setting to 29 can reduce the number of frames ignored and gives you 3% bandwidth on average. However, the actual frame to frame time is not consistent, you may see 34 or 26 on the display from time to time, you get an averaged result slightly better overall.

Hi Steve,

Soooo...., how does this logic extrapolate for monitors refreshing, as in my case, at 120Hz or, indeed, 144Hz?

Currently, I am using the NCP refresh rate limiter to limit at 30Hz with unlimited set in the sim. Prior to that I was limiting internally at 30Hz on the basis that this was achieving coincidence every 4th refresh (120/4). Providing I can sustain actual frame rates > or = 30 then these measures seem to be effective for most of the time.

Is this a reasonable approach or is the reality less straightforward?

Also, following this logic:

1. Does it make any sense bringing VSync, as is currently applied to Prepar3D, into the mix?

2. Should G-Sync be enabled? I have noted others enabling G-Sync which delivers positive activity within the frequency range of 30 up to the maximum refresh rate of the monitor. However, if FPS are being pegged at 30 by a limiter (internal or external) I assume G-Sync will still be active (if enabled), but will it actually be doing anything in terms of provision of additional frame rate stability?

Regards,

Mike

 

Hello,

thanks for the reply’s to my earlier question, reading on, I had always limited my Fps to 30 in P3d, after reading this thread I now have limited it to 30 in Nvidea control panel (never even considered this before now) with p3d being set at unlimited, I have not included any affinity mask,  core 0 of my cpu now is running at around 80 to 90% with no stutters,rather than 100% it was, I still think something has changed in P3d V5.2 but this has improved

Thanks

37 minutes ago, GSalden said:

Will those 26 to 34 and down again don’t cause stutters ?

Some will show micro stutters. When the frame output rate is slow the current frame is shown for the next period. When the output is fast the frames are lost. Real stutters are caused by big changes in the consistency of output.

5 minutes ago, Cruachan said:

...how does this logic extrapolate for monitors refreshing, as in my case, at 120Hz or, indeed, 144Hz?

A faster refresh cycle leaves less time before any new uploaded frame is shown. With 120Hz monitor refresh, that's a quarter of the time to wait compared to 30Hz, plus the time left for the current frame display period. If the output rate is at 30fps then each frame is on display four times in 120Hz refresh, but once or even twice at 30Hz. So it should be slightly better at 60 or 120 and so on.

The slight variance of Unlimited + VSYNC is not so great as to cause proper stutters. When there is some variance in frames output interval then there will be some variance in the time each frame is displayed.

Setting a higher fps at 60 vs 30 ultimately means half time between frames. However, the higher output fps can produce bigger swings when CPU throughput becomes max which might account for more real stutter. That's why many claim output in the 30fps range is good for flight sims because they might have to display so much per frame that inevitably some frames will be very time consuming.

52 minutes ago, Cruachan said:

Currently, I am using the NCP refresh rate limiter to limit at 30Hz with unlimited set in the sim. Prior to that I was limiting internally at 30Hz on the basis that this was achieving coincidence every 4th refresh (120/4). Providing I can sustain actual frame rates > or = 30 then these measures seem to be effective for most of the time.

Is this a reasonable approach or is the reality less straightforward?

Either approach should work fine externally limiting at 30fps in NCP or internally setting the frame rate target, nothing wrong with that on a 120Hz monitor should be quite good. Externally limiting can show quite a large variance in output as each frame takes a different time to render but the 120Hz refresh ensures very little waiting after any frame buffer is uploaded. The internal frame rate target is not a limiter in the same sense, since it outputs each frame after a set interval so it must be set low enough to allow consistent output otherwise will not work well.

I prefer to use the slightly slower output of 28fps limit in NCP for my 60Hz refresh which gets very smooth results simply because it requires on average 7% less CPU throughput for very nearly as fast as 30 fps, not much in it visually. That is the main reason I am setting that way since there's never a really firm coincidence of frame rate to refresh rate unless the the internal target rate slider is used set low enough and display settings sparse enough to maintain a look-ahead frame buffer.

2 hours ago, Cruachan said:

Also, following this logic:

1. Does it make any sense bringing VSync, as is currently applied to Prepar3D, into the mix?

2. Should G-Sync be enabled? I have noted others enabling G-Sync which delivers positive activity within the frequency range of 30 up to the maximum refresh rate of the monitor. However, if FPS are being pegged at 30 by a limiter (internal or external) I assume G-Sync will still be active (if enabled), but will it actually be doing anything in terms of provision of additional frame rate stability?

When using the NCP or other external limiter, it is best to use VSYNC=OFF in P3D Display Settings. To check that I set up Ideal Flight Professional to record fps and to start recording after a 30second delay so that the sim was settled. The flight was built with Paused=Off and to Pause after 1 minute. The sim was set up with Display Settings so that the Unlimited fps could reach well over 30 fps (50-80) and the NCP limit was set down at 30 FPS. In the 30 second graphs the fps was more stable with VSYNC=OFF.

Usually with G-SYNC it would work better at higher fps values due to less latency or time between frames. For example the sim could be set to allow for 40fps limit and that allows the G-SYNC to follow when the fps output falls below 40 to save from screen tearing.

Edited June 28, 20214 yr by SteveW

7 hours ago, Ray Proudfoot said:

That's a typo by me. I have a 6 core, 12 VP CPU. Just loaded P3Dv5.2 at EGLC and 0 remains at 100% with 2, 6 and 10 heavily used for loading scenery then drop to near idle once it's finished. That's at 60Hz.

Same load at 30Hz and 0 averaging in the 90s. Same readings for the other cores.

Once airborne (in slew mode) 2, 6 an 10 are used for scenery loading as you suggest. They're up and down. Soon as I pause the sim they drop to zero.

 

I also have a 6-core AMD Ryzen CPU. Here is my question: Is limiting the number of virtual processors for scenery loading not counterproductive?  For example, on my system, VPs 3-12 work on scenery loading during flight, and I see very crisp textures and a lot of autogen even from an altitude of >10000 feet. If I limit the number of VPs via affinity mask, the CPU will be limited in the number of VPs to render the scenery, which will make it harder for the CPU to do this work and lead to potential delayed autogen loading.