I recorded a video using Shadow Play and found that the video showed stutters that simply were not present in sim in real time. I think the recorded "stutters" are a limitation of the recording process...but I could be wrong

Thanks for explaining that Jeroen.

 

But I really think what I am seeing are REAL micro stutters, because I'm not just seeing some specific objects passing by in a jerky manner.  Everything on the screen is being rendered in a jerky way, including my aircraft. The rendering in v.2.3 is noticeably less smooth than it was in v.2.2 . . . which is especially apparent during heavy cloud cover.

 

I'm a Nordic skier, and flying in v.2.3 is much like skiing in sticky snow.  I'm still moving along at a good clip, but I'm not moving smoothly.  I have both MS Flight and X-Plane 10 installed and I compared the "feeling" of flying in all 3 flightsims in cloudy weather (notice that I'm ONLY comparing the feeling that I get while flying similar aircraft at virtually the same speed in all 3 sims). Both Flight and X-Plane feel silky smooth, while Prepar3D v.2.3 now feels sluggish (even when my frame rates remain decent).

I recorded a video using Shadow Play and found that the video showed stutters that simply were not present in sim in real time. I think the recorded "stutters" are a limitation of the recording process...but I could be wrong

 

That's what I thought at first. Now I believe that Shadowplay is making the micro stutters that are indeed present in the sim, more noticeable. After viewing my own video, I went back and looked closer at how P3D was performing, while viewing from the spot-wing tip viewpoint and low and behold I could see stutters (I nearly always fly from VC view). And then I went to VC view and, while the stutters were harder to spot, my aircraft definitely felt sluggish,

 

That's when I decided to fire up both MS Flight and X-Plane 10 and do some flightsim comparisons, under cloud cover, with similar aircraft, from the same airport (default airport for all three). I do get that comparing P3D to Flight and X-Plane (or FSX) is like comparing apples to oranges, but the only thing that I was trying to compare was the feeling of flight . . . and the smoothness of the other two sims really made P3D look bad (and I was getting nearly the same FPS with X-plane as I had with P3D).

But I really think what I am seeing are REAL micro stutters, because I'm not just seeing some specific objects passing by in a jerky manner.  Everything on the screen is being rendered in a jerky way, including my aircraft. The rendering in v.2.3 is noticeably less smooth than it was in v.2.2 . . . which is especially apparent during heavy cloud cover.

 

+1 exactly the same issue here with tessellation ticked ON. Even with 50-60fps it is still jerky.

 

Are you running with tessellation ticked ON or OFF?

 

With mine ticked OFF I get smooth fps but then the sim crashes within minutes with the R6025 runtime error.

 

 

are you running with tessellation ticked ON or OFF?

 

It is on. Tessellation has always been on for me, since v.2.0.

 

My understanding is that one of the significant differences between FSX and P3D2 is that much of the graphic processing was moved to the GPU . . . and this is mainly accomplished through tessellation (and DX11). I'm pretty sure MS Flight also uses tessellation (for its dynamic water and trees, that blow in the wind). 

The trees in Flight have nothing to do with tessellation... it's a product called SpeedTree.

 

http://www.speedtree.com

The trees in Flight have nothing to do with tessellation... it's a product called SpeedTree.

 

http://www.speedtree.com

 

Thanks Ed, I actually knew that ... I just thought that perhaps SpeedTree also did this through tessellation.  Do you know anything about Flight's dynamic water?

 

 

@everyone: this was just posted on LM's forum:

 

Quote from Beau on August 28, 2014, 10:34

 

Hello all,

 

Just to clarify a few points:

1) We did add a volumizing visual effect to nearby clouds which does add a sample and a bit of math to the pixel shader.  As Zach noted, this was not a huge hit in any of our tests, but it is more work.  

2) The main thing to be aware of when comparing GPU performance is that the number of pixels rendered and rasterized in a frame directly effects your results.  One big cloud that fills your screen will do more work than 50 far-off clouds because there will be fewer samples, and raster operations. We were not saying that people are seeing differences.  50 clouds that fill your screen will drastically overstress the GPU.  As Zach noted, we are working on ways to cut down on that.

 

There may be other issues contributing to this.  If you want to test things with volumizing disabled you can turn off the volume-effect that uses the noise texture by commenting out one line in ShadersHLSL\Cloud.fx. You are looking for this line:

 

#define VOLUMIZE

 

remove it or comment it out like this:

 

//#define VOLUMIZE

 

Then you'll need to clear your shaders cache by deleting the shader folder here:

 WINDOWS_DRIVE:\Users\USERNAME\AppData\Local\Lockheed Martin\Prepar3D v2

 

If that solves all your problems, we can add a cfg option for this effect in the next release.  If it doesn't really help then there is something else going on.

 

Speedtree is nothing short of amazing code that was around before tessellation was in the rendering loop. As for the water, I have to admit I was never impressed with Flight's water... and really don't know who's code it is. I'm pretty certain it's not strictly Microsoft's as they haven't created something amazingly original in decades.

Speedtree is nothing short of amazing code that was around before tessellation was in the rendering loop. As for the water, I have to admit I was never impressed with Flight's water... and really don't know who's code it is. I'm pretty certain it's not strictly Microsoft's as they haven't created something amazingly original in decades.

I like Flight's water ok, but I'm also mostly happy with P3D's water in v.2.3. For me the biggest issue is that the same wave size is used for small lakes (and even small ponds and rivers) as what is used for oceans.

 

I would LOVE to see LM incorporate Speedtree into P3D!

 

 

Anyway (maybe he can chime in and explain it once again) there is a clear miscommunication in this regard: the people who claim to have no stutters and smooth flight usually DO have that jerky passing by of objects when they look sideways but they don't call it stutters because officially they aren't.

 

Did you ever look for and analyze the FrameTime? One can do it with FRAPS and FRAFS. read my post #48 here http://forum.avsim.net/topic/450222-many-stutters-while-running-a-steady-fps/page-4 Such analysis shows very well if one has too long frames even if it looks like that the FPS is very stable. Maybe you want read this too? http://www.vortez.net/articles_pages/frame_time_analysis,1.html

Spirit

Well, I tried Beau's suggested edit (and deleted my Shades folder). I then loaded a test saved flight and the FPS do seem to be back closer to what I had in v.2.2 under similar conditions. The micro stutters were also reduced quite a bit (they did not totally disappear, but they are no longer nearly as annoying as they had been).

 

This test was done with ASN running, using its Historic Weather option to load the weather doing my saved flight time.

Well, I am not happy with the rendering of water at anything above extremely low altitudes. I really do think that work needs to be done to improve the realism of water at higher altitudes (and by "higher", I mean anything above 1000 feet), and I am off right now to post that on the P3D forum.

Well, I tried Beau's suggested edit (and deleted my Shades folder). I then loaded a test saved flight and the FPS do seem to be back closer to what I had in v.2.2 under similar conditions. The micro stutters were also reduced quite a bit (they did not totally disappear, but they are no longer nearly as annoying as they had been).

 

This test was done with ASN running, using its Historic Weather option to load the weather doing my saved flight time.

I concur..this is very welcome...kinda skeptical at first..but like you said ..feels like2.2j

and Im back in the ballgame..thanks Arwen

                                                    Rick

You're most welcome Rick.

 

Good to know that someone else had same results after making this edit.

I found no increase at least in FPS (initial paused fps) by making this cloud.fx change with my sample saved flight (before/after were the same)..   The only line i commented out was the one single line.. this is correct right?  There may be a smoothness increase, ill have to examine that later..

 

(This saved flight is an external view with multiple cloud layers.. a hard hitting test.. one that in fsx rings in around 30 fps and in p3d v2.3 about 21 fps, but p3d has a lot more going on in improvements )

 

 

For me the biggest issue is that the same wave size is used for small lakes (and even small ponds and rivers) as what is used for oceans.

 

 

 

Well, I am not happy with the rendering of water at anything above extremely low altitudes. I really do think that work needs to be done to improve the realism of water at higher altitudes (and by "higher", I mean anything above 1000 feet),

 

Amen!!! I have disabled water effects completely because of this: I can't stand those huge waves in little ponds when I fly at 3500 ft... Yikes.

 

 

 

Well, I tried Beau's suggested edit (and deleted my Shades folder). I then loaded a test saved flight and the FPS do seem to be back closer to what I had in v.2.2 under similar conditions. The micro stutters were also reduced quite a bit (they did not totally disappear, but they are no longer nearly as annoying as they had been).

 

Sounds good! The performance on my PC is great as it is but if this helps even a little, I will also use this trick. I doubt if anyone could notice the difference this new option brought us...?