The speed of moving clouds may cause stutters...

[Potroh 14]

I have a nice stutter-free setup. Works absolutely fine.
Today found a possible - but interesting - small cause for some 1 sec stutters.
It seems to be connected to the moving clouds + wind speed can cause small stutters and interestingly related to the speed by the clouds are moving.
The stutters are constant, happening every 10-15 seconds or so.

Easy to test: fire the "Cold Fronts" weather theme at about 25-30000 ft and press slew. Go to Advanced Weather and increase the wind-speed to 199 kts (the max allowed).
Clouds will start moving fast and the stutters appear. If you increase the sim-speed to 4x just temporarily, the appearance of stutters become more obvious.
Less wind-speed: less stutters.
Interesting...

Potroh
 

[vgbaron 3,627]

If clouds are moving rapidly then more rendering must be done - more rendering puts more load on the system - more load can induce stutters. Fortunately we don't often fly in 199kt winds

 

Been that way since forever.

 

Vic

[Potroh 14]

 

 

If clouds are moving rapidly then more rendering must be done - more rendering puts more load on the system - more load can induce stutters. Fortunately we don't often fly in 199kt winds
Been that way since forever.

 

Thanks. But it is not the case.
While in slew (no reliable other way to test) the stutters occur even if all scenery and other performance hungry features are OFF. That's why I dared to mention the phenomenon.

If you know some of the the internals of FS, you may know that the plane is actually "stationary" and it's everything else that move in relation to the plane being in the middle of the "cube".
Hence, when a 199kts wind blows the clouds, it is much slower than an airliner's own velocity in calm conditions. If all sceneries DO NOT cause the stutters moving faster, then the much slower clouds in themselves definitely shouldn't.

BTW, it's not been like forever, FSX and P3D 1.4 is free of this effect.

Potroh

[vgbaron 3,627]

Sorry, it is not free of this "effect" - it's just that it takes more to cause it. If you slew across the ground, you get blurries - regardless of addons. Remember, nothing is actually "moviong" - it is being redrawn in a different location. If you "move" faster - you need to redraw faster - if your system can't keep up - stutters.

 

Again - been that way since day one.

 

I fail to see your point. that's like saying if you bang your head against the wall, it hurts.

 

If the system gets overloaded by ANY means, it will eventually blur or stutter or flat out crash.

 

If you are slewing you are possibly increasing the disk reads which will also cause stutters. If you cross boundaries - same thing.

 

It's just another in the long list of things that can cause a stutter on certain systems.

[Guest]

 

 

While in slew (no reliable other way to test) the stutters occur even if all scenery and other performance hungry features are OFF.

 

Not sure I understand you, why would "slew" be the only reliable way to test?  Why can't one just turn off AI traffic, road/ship traffic and setup weather and test?

 

 

 

BTW, it's not been like forever, FSX and P3D 1.4 is free of this effect.

 

P3D V2.x onwards uses DX11 with different AA restrictions, unfortunately AA can't be turned OFF for cloud processing like it can in DX9/DX10.  Also, another factor in V2.2 onwards is cloud shadows (you would need to turn those off in testing) which are not present in FSX and P3D V1.4 ... time of day with cloud shadows on would also impact performance the shadows becoming increasing more complex as the light source (sun) moves.

 

You want to most accurately present what someone would normally fly in ... when one slews I'm guessing they don't really care about stutters, FPS, visuals, etc. ... unless they return to normal flight ... or am I missing something?

 

Cheers, Rob.

[Potroh 14]

Sorry, it is not free of this "effect" - it's just that it takes more to cause it. If you slew across the ground, you get blurries - regardless of addons. Remember, nothing is actually "moviong" - it is being redrawn in a different location. If you "move" faster - you need to redraw faster - if your system can't keep up - stutters.

 

Sorry buddy but you didn't read what I tried to depict...

My system CAN keep up with stutters and it is almost 100% stutter free.

Of course you don't need MOVE while being in slew-mode, you place the plane in slew to LET the CLOUDS move around you.

I merely mentioned "standing in air" when redraws are minimal or none. ONLY those clouds move. The situation is the same in the middle of the ocean... and can be easily tested almost without any scenery.

 

 

 

Again - been that way since day one.

 

Not at all. Most probably Rob is damn right and DX11 has to do with the changed and unwanted effect.

 

 

I fail to see your point. that's like saying if you bang your head against the wall, it hurts.

 

Yes because you didn't understand why to test in Slew.

 

 

If you are slewing you are possibly increasing the disk reads which will also cause stutters. If you cross boundaries - same thing.

 

Absolutely the opposite is true, when you stand in the air in Slew - after some time - there will be no disk-reads whatsoever.

 

Potroh

 

Hi Rob,

 

I think you are damn right and DX11 has something to do with this.

Nevertheless, with cloud shadows OFF you get more or less the same result.

Same with traffic, etc. because the stutter are there in the middle of the ocean too, with or without traffic.

Slew is a great mode to test certain features. This is a good example of that.

You would hardly guess the reason of an unexpected stutter problem if normally flying, but almost without any scenery.

For instance scenery designers call themselves "slew masters" because one never flies when designing sceneries. Slew is a great tool for measuring any static effect, without the flight-model and other factors being involved.

 

I simply mentioned the problem because it becomes evident when not flying, hence the cause could have been narrowed down to the high-speed movement of clouds.

 

Potroh

[ryanbatc 12,857]

Stuttering just means there's an increased load on your system and it can't cope. Identifying the culprit is the hard part

[vgbaron 3,627]

Potroh - I guess I'm not fully wrapping my head around what you are saying.....

 

I understood  you to mean you were  slewing but if I understand your comment, you are just sitting still in slew mode.

 

I don 't KNOW but I would sure suspect that even entering slew mode would create a totally different set of conditions for the sim. Just for example I would doubt that they would maintain threads or multiple core usage in slew mode. Typically slew mode is used to quickly move from point A to point B with no real regard for maintaining perfect visuals, etc.

 

I may be wrong about that but I would suspect that any opinions derived from actions in slew mode would be suspect.

 

You say your system is stutter free, as is mine - so you don't get any stutters unless the clouds are moving when you are in slew mode, or is your system not free and you are looking for a source of 1 second stutters that repeat every 10-15 seconds?

 

 

 

Vic

[Potroh 14]

Potroh - I guess I'm not fully wrapping my head around what you are saying.....

 

Vic,

 

I notice that, yes you seem to misunderstand the entire test scenario.

 

 

 

I understood you to mean you were slewing but if I understand your comment, you are just sitting still in slew mode.

 

I never said I WAS slewing, I talked about Slew-Mode, which make it possible to test certain sim elements apropriately. Yes, "sitting still" in the air is the way to test numerous things in any flight simulator.

 

 

 

I don 't KNOW but I would sure suspect that even entering slew mode would create a totally different set of conditions for the sim. Just for example I would doubt that they would maintain threads or multiple core usage in slew mode. Typically slew mode is used to quickly move from point A to point B with no real regard for maintaining perfect visuals, etc.

 

Yes, you do not seem to understand slew mode. It is not a bit different from flying, except that the plane is not moving according to the flight-model's rules, but you control (by different inputs) the pitch, bank, velocity, etc.

When in slew, everything works, animations, effects, traffic, everything - except the flight-model and certain instruments that are dependent on the model.

Slew mode is indeed typically used to  go from one point to another WITHOUT flying, but it is also a great possibility to test certain elements of the sim, without paying attention to the flight itself.

(When you are pushing back for instance, FS employs an internal slew-mode and nothing else changes).

 

Multi-core-usage and the sim's other core features are the same in Slew. Conditions - apart from the flight-model - remain the exact same.

Slew is a also a design-mode and at the same time a diagnostic-mode for the developers and testers.

 

 

 

You say your system is stutter free, as is mine - so you don't get any stutters unless the clouds are moving when you are in slew mode, or is your system not free and you are looking for a source of 1 second stutters that repeat every 10-15 seconds?

 

I was not talking about my own system, which is indeed working just fine, I simply mentioned an interesting possible source of stutters, which one may test and notice above version 2.0.

It was not a complaint but an observation.

When being around heavy or moderate clouds performance will drop for sure but I didn't know that the actual velocity of the moving clouds represents a factor that may lead to small stutters.

10-15 sec stutters appear when the wind is high and clouds are being on the run, but when flying an airliner at Flight level, it is very common to encounter very high winds, hence it is important to know and notice that the moving clouds in themselves can be the sourse for small performance drops.

 

As I've mentioned, Rob seems to be right and the DX11 change might have a significant role in the phenomenon.

 

Potroh

[vgbaron 3,627]

Potroh -

I've always maintained that Avsim is a great source for learning and I have learned something from this thread. I follow what you are saying now - my problem was a lack of understanding the operation of slew mode. It's one of those things that I rarely use except for a quick move from a-b.

You may very well be correct in your observations, I certainly do not have the technical expertise to dispute them.  :smile:
Yup, Rob has a habit of being pretty spot on.



Vic

[Guest]

Stutters can be caused by many things ... and I have noticed the 10-15 sec stutters when using water = ultra and high winds (actually anything 16Kts or higher) with NO Clouds but lots of water around on the coasts.  I haven't narrowed it down but my hunch is CUDA update and/or some code used to restrict wave heights ... I'm in a holding pattern right now in terms of reporting this as I'm waiting for something  But I need to really do A LOT more testing here before I can present anything that might be useful to LM.

 

As far as slew, I would NOT recommend using that for any stutter/performance evaluation.  My testing:

 

1.  Base PC with no add-ons and moderate hardware older 3960X and 970GTX ... use that as primary replication reference because if it can be replicated in this environment that will "almost" ensure LM can replicate.  If LM can't replicate, it will NOT get fixed.

 

2.  Loaded PC ... use this for testing how 3PDs work with new builds.  This is the much more difficult environment (that's why I have several fast SSDs to copy things in/out) ... also some 3PDs just can't be tested until they provide Beta builds (ObjectFlow, SODE, ASN, Virtuali, FSUIPC, etc. etc.) and some 3PDs don't provide builds until after release ... so for me as an "end user" I have my hands tied somewhat.

 

Slew is ok for moving to a specific location, but a much faster way is to use the MAP and just position the aircraft/avatar where you want it (either visually or enter coordinates) ... coordinates tend to work better when reporting issues.

 

Cheers, Rob.

[Potroh 14]

 

 

As far as slew, I would NOT recommend using that for any stutter/performance evaluation.

 

I'm fine with your personal choice, as long as the situation doesn't really need to be in slew mode.

Nevertheless, there are numerous scenarios when slew is unavoidable. Foir instance your own tests with the maxed out water and high winds don't really need slew, as you do not have to be steady in the air in order to observe the waves and their effect on performance.

But when you try out the high wind and moving clouds scenario, if you are flying besides or through those clouds, you mainly notice and concentrate on your own movement in the A/C. But when in Slew and being steady, only the clouds move, hence much easier to observe their behaviour and the stutters.

 

(Unless one flies an advanced plane that uses external dll stuff for its own displays, systems and internal calculations, like the NGX, etc. the advantage of slew-mode can be found in the fact that while being in slew, neither the flight-model nor the gauges are working in the otherwise "normal" runtime mode, thus there are a couple of related factors can easily be eliminated, in case one tries to find or measure a problem not related to those systems. The easiest way to find a problem is always run as few systems as possible, besides those being suspitious).

 

Regarding slew-mode, I guess only the old-fart designers (like myself) will truly appreciate the benefits offered by Slew, pointing out that scenery design for instance simply wouldn't be possible without it. (The lack of a genuine slew-mode was the cause for years why X-Plane sceneries for instance came very slowly and designing even the simplest airport for that sim used to be a nightmare...).

 

But to be on the positive side: I do agree with you and slew shouldn't be the primary tool for stutter or performance testing. For many other things, slew is simply unavoidable.

 

Potroh

[Guest]

Potroh,

 

I was doing some more testing tonight and I discovered a source(s) of long frames (aka stutters) at "some" 3rd party locations (Orbx KEGE for example) not all.

 

Disable the Following:

 

Water Reflections:

Clouds

User Vehicle

SimObjects

Terrain

Vegetation

Buildings

 

Dynamic Reflections: 

Off

 

NOTE: this is ONLY for "some" locations, other locations like KDEN are fine ... has me a little puzzled as in V3.0 I had no issues running those Water Reflection settings.

 

Let me know if it makes a difference.

 

Cheers, Rob.

[Potroh 14]

 

 

Let me know if it makes a difference.

 

Rob,

 

Will definitely have a look.

 

Potroh