PDA

View Full Version : tech question: basic simulation loop duration



XyZspineZyX
08-17-2003, 10:22 PM
Hi group,
I hope I don't ask for classified information with this one,
and if so, it only doubles my curiosity http://ubbxforums.ubi.com/infopop/emoticons/icon_wink.gif
What is the step size (duration) of the physics simulation loop for the flight model in IL2, and how long that for the damage calculation? Are graphics and physics decoupled or not? Is there a page that answers more of these questions?

thanks,
slarti

XyZspineZyX
08-17-2003, 10:22 PM
Hi group,
I hope I don't ask for classified information with this one,
and if so, it only doubles my curiosity http://ubbxforums.ubi.com/infopop/emoticons/icon_wink.gif
What is the step size (duration) of the physics simulation loop for the flight model in IL2, and how long that for the damage calculation? Are graphics and physics decoupled or not? Is there a page that answers more of these questions?

thanks,
slarti

XyZspineZyX
08-18-2003, 11:22 PM
Nice questions.

Good Luck finding answers. /i/smilies/16x16_smiley-wink.gif

You know how everybody whines for Oleg to releace his flight performance data he uses? Well if you use real math approximations rather than look up tables, then the data is a mere target for the math which does not always reach its intended target. It is possible, however unlikely, that Oleg's data may very well be acceptable to all but the math doesn't know this. Aerodynamics is some horribly complicated stuff for a PC.

SkyShrimp says P~47 rolls too slow, Fw190fan says Fw190 rolls too fast. If these fellas are correct, of course Oleg probably knows this (like DUH!) and knows the difference between the FB math results and the hysterical data. If for example in the math, roll is connected to the shinbone which is connected to the elevator response, then pushing down a lump in the FB rug by correcting elevator may make a lump pop up elsewhere, say in the Fw190 roll, cos the shinbone moved when the elevator was corrected. This may be what is happening, and explain the wild mood swings in FM that we see.

I heard the FB math was originally designed for a single low level ground attack IL~2 plane. Now its doing 80 flyables?

Heresy:: I'd be Happy if Oleg would do a Microsoft~esque look up table and simplified damage model so we could calculate 200+ aircraft in the air at one time, or put more CPU resources into much more intricate AI programming (but keep his performance table closely guarded Secret).

XyZspineZyX
08-19-2003, 03:34 AM
LEXX_Luthor wrote:
- Nice questions.
-
-
[snip]

I love that "lump in the rug" metaphor. It really does help explain mathematical cause-and-effect in non-technical terms.

-
[snip]
-
- Heresy:: I'd be Happy if Oleg would do a
- Microsoft~esque look up table and simplified damage

Not me! I would like to see Maddox put scale models of some of the planes (built to blueprint specifications) complete with scale ordnance (in some cases) and loaded with sensors: into the wind tunnel. Then, use that resulting data to tweak the flight model for particular planes (and eventually develop even more sophisticated flight models).

Alright, that is probably not going to happen; but those 1C:Maddox guys, being a bunch of engineers, would probably like doing that research if they had the money and/or the time. /i/smilies/16x16_smiley-happy.gif

<center>http://www.bloggerheads.com/mash_quiz/images/mash_potter_anim.gif (http://www.bloggerheads.com/mash_quiz/)</center>

XyZspineZyX
08-19-2003, 05:42 AM
Right on, Luthor. I think some people fail to recognize that we're flying on simplified math routines. Some 80 flyables on a FM designed for ONE plane really shows how good the FM was from the very beginning.

But hey, I wouldn't be happy if IL2 goes the look-up table way. That'll give you awful (poorly modelled) stalls, spins.... AI planes maybe, but not player planes.

XyZspineZyX
08-19-2003, 05:48 AM
tintifax32 wrote:
- Hi group,
- I hope I don't ask for classified information with
- this one,
- and if so, it only doubles my curiosity /i/smilies/16x16_smiley-wink.gif
- What is the step size (duration) of the physics
- simulation loop for the flight model in IL2, and how
- long that for the damage calculation? Are graphics
- and physics decoupled or not? Is there a page that
- answers more of these questions?

My GUESS would be yes, decoupled... mater of fact, dont see how they would have ever been coupled.. But even if they were the damage model would be running on the pc processor.. and most of the graphics calculations are offloaded to the video card.. where it would recive.. ques from the processor.. ie EXPLODE...



TAGERT
-------------------------------------------------------------------------------------------------------
If WAR was not the ANSWER.. Than what the H was your QUESTION?

http://dictionary.reference.com/search?q=forum
http://dictionary.reference.com/search?q=discussion

XyZspineZyX
08-19-2003, 05:52 AM
rbstr44 wrote:
- Not me! I would like to see Maddox put scale models
- of some of the planes (built to blueprint
- specifications) complete with scale ordnance (in
- some cases) and loaded with sensors: into the wind
- tunnel. Then, use that resulting data to tweak the
- flight model for particular planes (and eventually
- develop even more sophisticated flight models).

Scale models and wind tunnels are used to predict how things will work; they do not give more detailed data. You can make it very close to the real thing but it's not the real thing. If Oleg has all the necessary data then he'll benefit little from more, not-the-real-thing, data. If he has some missing parts, well yes, maybe fitting in some wind tunnel data would help.

And there's really no such thing as "flight model for particular planes". That's for table look-up sims, not IL2. Like there is no FW 190 AI or I-16 AI; they are the same.

XyZspineZyX
08-19-2003, 05:59 AM
LEXX_Luthor wrote:
- Aerodynamics is some horribly complicated
- stuff for a PC.

Actually it aint.. relitive to OTHER stuff in the sim.. Im no expert (But I play one on TV) but with regrads to..

- I heard the FB math was originally designed for a
- single low level ground attack IL~2 plane. Now its
- doing 80 flyables?

The MATH is basically the same.. the MATH of the flight model that is.. It is the PARAMETERS that drive/limit the MATH that make it specific to an individual aircraft.. And dont confuse those PARAMETERS with the static type of data we tend to get in books.. ie weight, top speed, fuel load, etc.. Not that they are not part of it.. but not as big as a part as the PARAMETERS (Coeficints) use by the MATH..

Now dont quote me on this.. it has been years sense I even looked at a fligh model.. And at that time it was writen in fortan.. I do seem to recal there was a POINT where some MATH was beter than others.. That point was the speed of sound.. If I recall correctly.. That is to say some MATH does a better job at modling an aircraft at less that the speed of sound (props stuff) and the other does a better job for the faster than (jest stuff)

In a nut shell, the DATA Oleg has that you refer to is used to CHECK the math.. And I *think* it can be used to generate the PARAMETERS.. not sure about that.. mater of fact not sure about any of this.. Just my understanding of it.. which if you ask my ex wife.. is allways wrong! /i/smilies/16x16_smiley-wink.gif



TAGERT
-------------------------------------------------------------------------------------------------------
If WAR was not the ANSWER.. Than what the H was your QUESTION?

http://dictionary.reference.com/search?q=forum
http://dictionary.reference.com/search?q=discussion

XyZspineZyX
08-19-2003, 06:03 AM
My educated guess would be over 30 loops/sec. Otherwise you might not see smooth motions. But the X-plane simulator ( http://www.x-plane.com ) says it does FM at 15 loops/sec....

XyZspineZyX
08-19-2003, 06:08 AM
HomeboyWu wrote:
- My educated guess would be over 30 loops/sec.
- Otherwise you might not see smooth motions.

My guess is it could be longer, in that the video card could just extrapole the last vector.. but.. that is a guess.

- But the X-plane simulator

Does X-plane calculate AI, tanks moving, Flack, etc?



TAGERT
-------------------------------------------------------------------------------------------------------
If WAR was not the ANSWER.. Than what the H was your QUESTION?

http://dictionary.reference.com/search?q=forum
http://dictionary.reference.com/search?q=discussion

XyZspineZyX
08-19-2003, 06:24 AM
tagert wrote:
-- But the X-plane simulator
-
- Does X-plane

No.

XyZspineZyX
08-19-2003, 06:51 AM
HomeboyWu wrote:
-
- tagert wrote:
--- But the X-plane simulator
--
-- Does X-plane
-
- No.

There is your other 15 than! /i/smilies/16x16_smiley-wink.gif

TAGERT
-------------------------------------------------------------------------------------------------------
If WAR was not the ANSWER.. Than what the H was your QUESTION?

http://dictionary.reference.com/search?q=forum
http://dictionary.reference.com/search?q=discussion

XyZspineZyX
08-19-2003, 07:59 AM
tagert wrote:
- There is your other 15 than!
Than what? What do you really want to say, tagert?

XyZspineZyX
08-19-2003, 02:41 PM
Visually, you're only able to detect about 25 fps at the most. Anything above that will only give you a nicer motion blur. However, tactile feedback takes close to 60 Hz to fool the senses and thus, the big sims tend to use 60Hz.

As for the above FM and math discussion, it is clear that some people should keep their feeding holes shut when they have no idea. Tagert excluded, he seems to know what he is talking about.

And no, I can't be bothered to fill in the holes. Not today and not in this context.

Cheers,
Fred

No sig as of now, as people apparently can't handle reality without creating too much trouble for the poor mods.

XyZspineZyX
08-19-2003, 03:42 PM
HomeboyWu wrote:
-
- rbstr44 wrote:
-- Not me! I would like to see Maddox put scale models
-- of some of the planes (built to blueprint
-- specifications) complete with scale ordnance (in
-- some cases) and loaded with sensors: into the wind
-- tunnel. Then, use that resulting data to tweak the
-- flight model for particular planes (and eventually
-- develop even more sophisticated flight models).
-
- Scale models and wind tunnels are used to predict
- how things will work; they do not give more detailed
- data. [ellided]

Certainly, wind tunnel testing of scale models predicts how the full-size, real plane reacts in a similar airflow. AND, the tests certainly can provide more data! Don't you think if NASA re-examined the Wright Flyer wing model in its wind tunnels that they would collect immensely more, and different kinds of data than Orville and Wilbur ever dreamed of collecting from their own primitive wind tunnel? Is electronic sensor technology more advanced, and computer data collection more thorough than what engineers were using in 1939-1945? No doubt they are.

[continued]
- And there's really no such thing as "flight model
- for particular planes". That's for table look-up
- sims, not IL2. Like there is no FW 190 AI or I-16
- AI; they are the same.
-
-

Right, what I said was poorly worded. What I meant was that the developers can adjust the simulated flying performance of the individual planes as handled by the big FB FLIGHT MODEL (FM). The adjustment might affect the inputs or just the outputs. To be sure, the core aerodynamics equations should remain mostly the same; I am not suggesting it would be prudent to change those at all unless a large inconsistency was found that affected most of the planes in IL2/FB.

So, without touching the core FM equations, there are two ways to affect the simulation OUTPUTS:
1) INPUTS: each plane has particular characteristics (PARAMETERS), which are described by a set of data inputs; use of more precise data can change the simulated flying performance (OUTPUTS), hopefully, improving the simulation (better realism);
2) CUSTOM CODE: no simulation is perfect. Custom code is needed when the FM produces unexpected or unrealistic results. That is, some planes may have particular flight handling characteristics which the INPUT data does not adequately describe, or the FM does not address; that is, there may be design nuances which affect flight handling for particular real aircraft which were not anticipated by either the collected input/parameter data or the FM. This code "customizes" the FM output results for such a unique condition, or aircraft type to bring it closer to known flight test results or manufacturer specifications. Still, purists may refer to custom code as "a hack" (particularly, when it is added later as a fix to a particular plane).

Having started with a superb FM, IL2/FB probably uses a minimal number of lines of custom coding. As patches, and new aircraft are added, however, I would expect that custom solutions tend to increase so that the big FM can be left alone.

<center>http://www.bloggerheads.com/mash_quiz/images/mash_potter_anim.gif (http://www.bloggerheads.com/mash_quiz/)</center>

XyZspineZyX
08-19-2003, 04:05 PM
tagert wrote:
- Does X-plane calculate AI, tanks moving, Flack, etc?

You can get cars on roads, and there are other planes,
but otherwise no. The weather is very nice, though,
and you can design your own aircraft, assign wings and
airfoils, etc.

The other main force based flight sim that I am aware
of Fighter Squadron, Screaming Demons over Europe, which
is rather better with the recent patches, although I
haven't tried calculating anything for the P-47 with the
latest base pack, etc.

XyZspineZyX
08-19-2003, 04:20 PM
HomeboyWu wrote:
- Than what?

Than nothing.

- What do you really want to say, tagert?

Thought that last reply was pretty cut and dry... What part were you having trouble with?





TAGERT
-------------------------------------------------------------------------------------------------------
If WAR was not the ANSWER.. Than what the H was your QUESTION?

http://dictionary.reference.com/search?q=forum
http://dictionary.reference.com/search?q=discussion

XyZspineZyX
08-19-2003, 04:33 PM
rbstr44 wrote:
- Right, what I said was poorly worded. What I meant
- was that the developers can adjust the simulated
- flying performance of the individual planes as
- handled by the big FB FLIGHT MODEL (FM). The
- adjustment might affect the inputs or just the
- outputs. To be sure, the core aerodynamics
- equations should remain mostly the same; I am not
- suggesting it would be prudent to change those at
- all unless a large inconsistency was found that
- affected most of the planes in IL2/FB.

Agreed.

- So, without touching the core FM equations, there
- are two ways to affect the simulation OUTPUTS:
- 1) INPUTS: each plane has particular characteristics
- (PARAMETERS), which are described by a set of data
- inputs; use of more precise data can change the
- simulated flying performance (OUTPUTS), hopefully,
- improving the simulation (better realism);

Agreed.

- 2) CUSTOM CODE: no simulation is perfect.

No simulation ever will be.

- Custom code is needed when the FM produces unexpected
- or unrealistic results.

Hmmmmm.

- That is, some planes may have
- particular flight handling characteristics which the
- INPUT data does not adequately describe, or the FM
- does not address; that is, there may be design
- nuances which affect flight handling for particular
- real aircraft which were not anticipated by either
- the collected input/parameter data or the FM. This
- code "customizes" the FM output results for such a
- unique condition, or aircraft type to bring it
- closer to known flight test results or manufacturer
- specifications. Still, purists may refer to custom
- code as "a hack" (particularly, when it is added
- later as a fix to a particular plane).

The only custom code Im aware of is when the plane exits the flying mode.. ie crosses over to not flying.. I think they call/called it departer code.. And the only time it kicks in is during a stall/spin situation.. But.. as I pointed out, it has been years.. So please give me an example of this HACK your refering too.. It sounds interesting.

- Having started with a superb FM, IL2/FB probably
- uses a minimal number of lines of custom coding. As
- patches, and new aircraft are added, however, I
- would expect that custom solutions tend to increase
- so that the big FM can be left alone.

Maybe.. Another area that is gray is when the FM nears that speed of sound point.. where the other FM would do a better job.. But only the Me262 and Me163 and P47 would even come close to that.. So probally beter to just let it act a bit wacky in that no real data exists in that area anyways..



TAGERT
-------------------------------------------------------------------------------------------------------
If WAR was not the ANSWER.. Than what the H was your QUESTION?

http://dictionary.reference.com/search?q=forum
http://dictionary.reference.com/search?q=discussion

XyZspineZyX
08-19-2003, 04:39 PM
rbstr44 wrote:
- Is electronic sensor technology more
- advanced, and computer data collection more thorough
- than what engineers were using in 1939-1945? No
- doubt they are.

No doubt.. But.. if you ever sad down and read some of them old NACA reports from the 20s and 30s and 40s I think you would be amazed at how brilient those men were back then in the methods they came up with to collect data... Amazing stuff really!




TAGERT
-------------------------------------------------------------------------------------------------------
If WAR was not the ANSWER.. Than what the H was your QUESTION?

http://dictionary.reference.com/search?q=forum
http://dictionary.reference.com/search?q=discussion

XyZspineZyX
08-20-2003, 03:02 AM
tagert wrote:
- rbstr44 wrote:

I should have included a big "IMHO" before my last statements. Of course, only the developers know exactly how the software is organized, but I think an arrangement such as I have stated is plausible given the developer's statements about using physics-based flight models, and assuming they do not get re-written very often!

[ellided]
-
- The only custom code Im aware of is when the plane
- exits the flying mode.. ie crosses over to not
- flying.. I think they call/called it departer code..
- And the only time it kicks in is during a stall/spin
- situation.. But.. as I pointed out, it has been
- years.. So please give me an example of this HACK
- your refering too.. It sounds interesting.
-

When I was at Old Dominion University, we had a software engineering class term project that was created for us by NASA Langley: student programming teams were required to write high-level computer code for an inter-planetary lander simulation. NASA already had a flight model and flight control model that was written into a NASA requirements/specification document. After writing the computer program by following the requirements/specifications document, we found some unanticipated behavior in the simulation created by the documented flight control model which NASA engineers had not anticipated. So, despite our nice, neat pseudo-code charts, and extensive program documentation, we had to quickly write a few lines of additional logic code to keep the simulated lander vehicle from spinning out of control during the simulated descent to Mars. So, in a way, we violated the engineers' documented flight control model with this undocumented "custom code" (or "hack"/"kludge" if you will) at the last minute before submitting our project at the end of term--it was then too late to re-write our own program documentation. After my team's pointing out the lander model problem to our professor (and NASA), NASA fixed their requirements/specifications for the next class who had that same project the following school year.

After the recent crashes by NASA spacecraft on Mars, I wonder if NASA made a mistake and used our student project computer code despite the disclaimer we wrote into the program block comments. /i/smilies/16x16_smiley-very-happy.gif

[ellided]
-
- Maybe.. Another area that is gray is when the FM
- nears that speed of sound point.. where the other FM
- would do a better job.. But only the Me262 and Me163
- and P47 would even come close to that.. So probally
- beter to just let it act a bit wacky in that no real
- data exists in that area anyways..
-
Yes. There might possibly be custom code written for just the planes which can fly near Mach speed to handle those unique flight performance situations, or perhaps there is a separate branch of the FM which only gets executed when velocity starts reaching 80-90% of Mach. I am thinking that twin boom constructed aircraft (i.e., P-38 Lightning), and multi-engined aircraft might require some custom coding since the original flight model centered around mono-wing, mono-fuselaged, single power-plant planes. Of course, Maddox & Co. may have changed the FM since the original IL2 to anticipate all these new aircraft types.

<center>http://www.bloggerheads.com/mash_quiz/images/mash_potter_anim.gif (http://www.bloggerheads.com/mash_quiz/)</center>

XyZspineZyX
08-20-2003, 04:15 AM
rbstr44 wrote:
- tagert wrote:
-- rbstr44 wrote:

I did? /i/smilies/16x16_smiley-wink.gif

- I should have included a big "IMHO" before my last
- statements.

IMHO you didnt need to, it was pretty clear it was an (your) opinion.

- Of course, only the developers know
- exactly how the software is organized,

Never said they didnt.. But there are limits to what that organiztaion is.. Unless they are doing something that no one has done before.. Which I doublt BIG TIME!

- but I think an arrangement such as I have stated
- is plausible given the developer's statements
- about using physics-based flight models, and
- assuming they do not get re-written very often!

Ok.. dont think I said you your statements werent..?

- When I was at Old Dominion University, we had a
- software engineering class term project that was
- created for us by NASA Langley: student programming
- teams were required to write high-level computer
- code for an inter-planetary lander simulation. NASA
- already had a flight model and flight control model
- that was written into a NASA
- requirements/specification document. After writing
- the computer program by following the
- requirements/specifications document, we found some
- unanticipated behavior in the simulation created by
- the documented flight control model which NASA
- engineers had not anticipated. So, despite our
- nice, neat pseudo-code charts, and extensive program
- documentation, we had to quickly write a few lines
- of additional logic code to keep the simulated
- lander vehicle from spinning out of control during
- the simulated descent to Mars. So, in a way, we
- violated the engineers' documented flight control
- model with this undocumented "custom code" (or
- "hack"/"kludge" if you will) at the last minute
- before submitting our project at the end of term--it
- was then too late to re-write our own program
- documentation. After my team's pointing out the
- lander model problem to our professor (and NASA),
- NASA fixed their requirements/specifications for the
- next class who had that same project the following
- school year.

Sounds like a case of two wrongs make a right!

- After the recent crashes by NASA spacecraft on Mars,
- I wonder if NASA made a mistake and used our student
- project computer code despite the disclaimer we
- wrote into the program block comments.

Well.. I work for a big aerospace company.. And it is scarry how some code get reused without a review.. In the SPACE of aerospace a *thing* has alot of value if it has FLOWN before.. To the point that it becomes golden.

- Yes. There might possibly be custom code written
- for just the planes which can fly near Mach speed to
- handle those unique flight performance situations,
- or perhaps there is a separate branch of the FM
- which only gets executed when velocity starts
- reaching 80-90% of Mach. I am thinking that twin
- boom constructed aircraft (i.e., P-38 Lightning),
- and multi-engined aircraft might require some custom
- coding since the original flight model centered
- around mono-wing, mono-fuselaged, single power-plant
- planes. Of course, Maddox & Co. may have changed
- the FM since the original IL2 to anticipate all
- these new aircraft types.

If I remember correctly.. the flight model.. ie the dynamic physics type of flight model can handle mutli eng stuff no problem.. Just some different/extra prameters a single eng wouldnt have.

Still interested in hearing about a HACK that is necarry for code that does not have any mistakes in it.. The only I can still come up with is the departure code and high speed stuff I alluded to in last post.



TAGERT
-------------------------------------------------------------------------------------------------------
If WAR was not the ANSWER.. Than what the H was your QUESTION?

http://dictionary.reference.com/search?q=forum
http://dictionary.reference.com/search?q=discussion

XyZspineZyX
08-24-2003, 11:32 PM
The main reason for my question was if it actually can happen that the physics loop gets too slow and therefore gives wrong results, therefore influencing the "realtime" FM in a way. If IL2 is using the physics calculation the way I suspect (and I am quite sure, little chance to do it differently though still performant) then missing the deadline for a physics frame will produce a wrong flight path and wrong properties for an aircraft. The key question is: is my cpu fast enough to deliver a correct FM or will it put me at a disadvantage in online play, because I fly a different (most likely slower in all respects) aircraft?

slarti

XyZspineZyX
08-25-2003, 06:15 AM
tintifax32 wrote:
- The main reason for my question was if it actually
- can happen that the physics loop gets too slow and
- therefore gives wrong results, therefore influencing
- the "realtime" FM in a way.

This can, and indeed, happen. Evident when you play back a track with lots of units, the track goes completely wrong, even when record & playback using the same version (hence same physics routine).

That's because your CPU is overloaded with FM calculations at the time of record OR playback. Some (random, you know CPU load depends on graphics, too) of the loops get dropped, and it either records a wrong track or plays wrong a good track.

I suppose you understand the track system in IL2/FB?

- If IL2 is using the
- physics calculation the way I suspect (and I am
- quite sure, little chance to do it differently
- though still performant) then missing the deadline
- for a physics frame will produce a wrong flight path
- and wrong properties for an aircraft. The key
- question is: is my cpu fast enough to deliver a
- correct FM or will it put me at a disadvantage in
- online play, because I fly a different (most likely
- slower in all respects) aircraft?
-
- slarti

Hey, it might be possible. But try to record a track then playback and see if it plays correctly. If so, then there shouldn't be any missing physics frames. At least that's what I think. /i/smilies/16x16_smiley-tongue.gif

XyZspineZyX
08-25-2003, 10:12 AM
Yah just push the graphics and load up objects in a mission till the thing is in or near slideshow mode then see how it flies! Maybe the flight itself isn't so bad but input timing from the stick is (see how it does in a straight path or glide) and that creates an impression of a vastly changed FM. The FM may take the loop time into itself, may adjust to load, we don't know.

I agree with Effte that Tagert has the basics right.

The parameters are a form of table in themselves but the FM's of the past few years model planes better and better. EAW has parameters but they apply to the plane as a whole unit, you don't get aerodynamic effects from the parts being applied to lengths and inertias nor one effect balancing against another -- no simultaneous equations! Does IL2/FB do this? I think to some degree some of it, yes because wing damage doesn't seem to produce "canned" results nor do stalls and spins. If you have modelling down to the parts then you have less "departure code" because your model still applies under more conditions. But then, I may be fooled by some slick coding too.


Neal