Framerates!!

[misson]

I've been trying to relativize problem of framerates, and studying.

What affects a plane to the framerates are mainly the amount of polygons and colors to be projected on the screen.

For this, it is key in the case of airplanes multi-lod, what lod how far CFS2 load.

In the event that the aircraft is not multi-lod, the number of frame rates will be the same at any distance. And the advantage of a plane multi-lod good framerates to be greater in the distance, especially in formations. But at medium or short, these multi-lod aircraft formations may produce worse framerates than a formation of planes not multi-lod.

This is the explanation of why a short distance in quick combat, I have better framerates fighting against a formation of AF Scrub spits than a formation of the same number of BSK Spits. Because BSK Spit has a much greater number of polygons in the first and second lod than AF Spit. But is clear BSK Spits are much more detailed than others, but his advantage (in framerates) is only in the very long distance formations.

This is my thought!

 

[alpha_1]

Great Research, Mission

Thanks for doing the lifting on this. I've often wondered exactly who some and not others. You're research has given me a clue. Thanks.

Cheers.:ernae:

 

[Rami]

Reply...

Misson,

Your analysis is somewhat correct, but take into account a couple of points.

1) Even in formation, the multi-lod planes further away from yours are drawn in less detail...especially if you're using a large formation of six or eight.

2) Bruno took that into account when he designed the model, to an extent. The furthest-away model is very basic, and that's the model you see in combat for the most part unless you are in formation, or are within close quarters in a dogfight with one of these Spitfires.

Once you are in combat, especially a large-scale furball, these planes tend to balance in favor of multi-lod.

Twin bombers like the Blenheim and the Dornier Do-17 can also have framerate problems with slower rigs.

Here's one question I have for your research, however...

When comparing these models, did you take into account the influence of exhaust effects? Exhaust effects are notorious for having compounding effects over time in terms of negatively impacting framerates during missions and quick combat. While they are great eye candy, they can also be terribly destructive.

 

[misson]

Misson,

Your analysis is somewhat correct, but take into account a couple of points.

1) Even in formation, the multi-lod planes further away from yours are drawn in less detail...especially if you're using a large formation of six or eight.

2) Bruno took that into account when he designed the model, to an extent. The furthest-away model is very basic, and that's the model you see in combat for the most part unless you are in formation, or are within close quarters in a dogfight with one of these Spitfires.

Once you are in combat, especially a large-scale furball, these planes tend to balance in favor of multi-lod.

Twin bombers like the Blenheim and the Dornier Do-17 can also have framerate problems with slower rigs.

Here's one question I have for your research, however...

When comparing these models, did you take into account the influence of exhaust effects? Exhaust effects are notorious for having compounding effects over time in terms of negatively impacting framerates during missions and quick combat. While they are great eye candy, they can also be terribly destructive.

U right Andrew in somewhat too, I do not think the same about your first point, and about to be in combat the ballance in favor of multilod. AND YES , the A.F.Spit has exaust efects too.

BUT , I´m not trying to initiate here a discussion, I just want that ALL WE , talk and share valid evidence and arguments about this, sorry but some times i feel limited in my english! to express here what i mean exacly.

 

[Ettico]

My 2 cents:

I use the trial-and-error method to test framerates. I download a plane. I create a simple mission with a formation of 8. Then I look at the formation using external view from short and long range, noting the frame rates. If the plane passes my little test, I keep it. I require 25+ fps at long range with the entire formation in view.

I've found that all multi-lod planes do not necessarily perform well in formations, frame-rate-wise. I don't know why. I simply know from experience that some planes produce good frame rates alone, but a formation of the same plane might turn a mission into a slide show.

I've had to toss a lot of great planes because of that problem. I've decided that my next computer will be a gaming rig armed with a 4 GH dual processor, state-of-the-art graphics, and tons of RAM. Maybe that will solve my problem.

 

[bobhegf]

I have had the same problem and decided to get a good rig built. I decided to use the large monitor that came with the junk rig I use on the internet and switch it out so that I can use it on the gamming rig., this one runs at 75Hz. I decided to use my old monitor That came with my 1st rig back when win98se was the thing. I found that using my old 60Hz monitor set at 1024x768x32 looks good and runs at 80fps or more to about 35fps on the low end with clouds and 68 AI AC going at it on my junk rig.

 

[kelticheart]

Absolutely true!

...When comparing these models, did you take into account the influence of exhaust effects? Exhaust effects are notorious for having compounding effects over time in terms of negatively impacting framerates during missions and quick combat. While they are great eye candy, they can also be terribly destructive.

This is particularly true with permanent flame effects, more specifically when added to V-12 engines with individual exhaust stubs.

Each aircraft has then 12 flame effects multiplied by the total number of AC's in a formation. In a simple 4 against 4 Quick Combat situation with 4 Bf109s against 4 Spitfire MkIX, for example, we would have 96 flame effects displayed at any given time.

Enough to slow down any computer, including today's computing monsters!

I resolved such problem with Green Ghost's exhaust gauge. First of all, the flame effects are displayed on the player's aircraft only and then only above 75% power, a percentage that can be increased at will.
As permanent effects, I left only single thin smoke trailers per each exhaust stub on radial engines and only two, one per each engine side, on V-12 engines.

As a result, 95% of my installed aircrafts will display only a few thin smoke trailers each, up to a top of 8 on 4-engine bombers with dual exh stubs like, for example, a B-29.

After running several fps tests, I found that the negative framerate impact is extremely minimal, about 1-2% at the most. I ran all of my tests over open sea, in order to keep ground mesh and GSL objects from affecting the results.

Cheers!
KH :ernae:

 

[misson]

This is particularly true with permanent flame effects, more specifically when added to V-12 engines with individual exhaust stubs.

Each aircraft has then 12 flame effects multiplied by the total number of AC's in a formation. In a simple 4 against 4 Quick Combat situation with 4 Bf109s against 4 Spitfire MkIX, for example, we would have 96 flame effects displayed at any given time.

Enough to slow down any computer, including today's computing monsters!

I resolved such problem with Green Ghost's exhaust gauge. First of all, the flame effects are displayed on the player's aircraft only and then only above 75% power, a percentage that can be increased at will.
As permanent effects, I left only single thin smoke trailers per each exhaust stub on radial engines and only two, one per each engine side, on V-12 engines.

As a result, 95% of my installed aircrafts will display only a few thin smoke trailers each, up to a top of 8 on 4-engine bombers with dual exh stubs like, for example, a B-29.

After running several fps tests, I found that the negative framerate impact is extremely minimal, about 1-2% at the most. I ran all of my tests over open sea, in order to keep ground mesh and GSL objects from affecting the results.

Cheers!
KH :ernae:

​I agree!! about the efects

but seems we need to be more carefull (perhaps) with the efects than if it is multilod or not!

 

[misson]

Some time ago, FdeBressy bring my atention to a Hurricanes series from "Touch the Sky simulations" , for me these hurrys are THE KING OF THE KINGS about detail , but is a pity flying them alone in the CFS2 sky, i have poor frames rates, in this pics u can see how many poligons they have.

But if I convert many of these type of planes (many of them working in FSX) is because I just collect them or I use them in Multiplayer combat against a few planes, etc. And I do not think that I´m the only one!.

That is why CFS2 has not only campaign or missions options to play, it has also freeflight alone, quickcombat, and multiplayer options. Every one of us knows why we like this game.

So I do think there is not reason to wast any plane, or any object. The key is use each plane or object for such wanted performance, and for that perhaps we need several cfs2 installs, after all.. how many type of games we play?

 

[rhumbaflappy]

It's not just the number of polys ( or vertices ) that will eat up framerates. All polys need to be textured... not colored. and the textres for all the aircraft should be on the same texture sheet, to avoid extra drawcalls and load times on a single aircraft.

Dick

 

[misson]

It's not just the number of polys ( or vertices ) that will eat up framerates. All polys need to be textured... not colored. and the textres for all the aircraft should be on the same texture sheet, to avoid extra drawcalls and load times on a single aircraft.

Dick

please Dick,! explane that! Does mean is better a plane with less texture files than other with much more files in his texture folder?

 

[misson]

It's not just the number of polys ( or vertices ) that will eat up framerates. All polys need to be textured... not colored. and the textres for all the aircraft should be on the same texture sheet, to avoid extra drawcalls and load times on a single aircraft.

Dick

So , if u have a multilod plane , cfs2 needs to load each texture by lod (because could be not the same textures) if the planes are in not same ranges! ?

 

[Desert Rat]

post #11 = Yes.

post #12= no.

On the second point, for the most part LOD's use the same texture file as the main model, this is already loaded into memory, whether it be mipped or not. Frames will be increased more-so if it is mipped as distance uses a lower detailed mip of the bmp. The ofset is that this can cause blurred textures when the sime uses lower mips evenon close quarter views of planes.

Jamie

 

[misson]

post #11 = Yes.

post #12= no.

On the second point, for the most part LOD's use the same texture file as the main model, this is already loaded into memory, whether it be mipped or not. Frames will be increased more-so if it is mipped as distance uses a lower detailed mip of the bmp. The ofset is that this can cause blurred textures when the sime uses lower mips evenon close quarter views of planes.

Jamie

thank u Jamie!

 

[rhumbaflappy]

It's better to have a huge texture sheet as this is one drawcall per texture file. Also, if you use colors rather than textures for parts of the object, each color will have a drawcall... if those 'colors' are on the same texturesheet, there will be only one drawcall for all the parts.

This has been discussed in depth at FSDeveloper.com by Arno Gerretson.

Dick

 

[misson]

It's better to have a huge texture sheet as this is one drawcall per texture file. Also, if you use colors rather than textures for parts of the object, each color will have a drawcall... if those 'colors' are on the same texturesheet, there will be only one drawcall for all the parts.

This has been discussed in depth at FSDeveloper.com by Arno Gerretson.

Dick

Ok Dick , thank u for reply.

So i can see to this point , that we have a some variables. a) polys amount and his texture drawcalls
b) efects
But if I understand correctly , both variables does not indicates better performance of a multilod or not multilod plane, exept if the multilod plane is in enough distance to cfs2 load a lod with lower polis and drawcalls than other plane with only one lod.
And at the same time , a multilod object or plane in his first lods to be displayed could have worse framerates (if it has superior amount of polys&drawcalls or efects)?

 

[mav]

Ok Dick , thank u for reply.

So i can see to this point , that we have a some variables. a) polys amount and his texture drawcalls
b) efects
But if I understand correctly , both variables does not indicates better performance of a multilod or not multilod plane, exept if the multilod plane is in enough distance to cfs2 load a lod with lower polis and drawcalls than other plane with only one lod.
And at the same time , a multilod object or plane in his first lods to be displayed could have worse framerates (if it has superior amount of polys&drawcalls or efects)?

Mission,

Bottom line... in a cfs2 environment, for say QC and missions the more LoDs the better especially when it's a very detailed model and when you consider there's other moving and non moving objects like scenery etc ... really helps the F/R ,so you don't get a slide show effect.... feel most fs9 conversions lite to medium 2 or 3 LOD while something like Robert Sanderson's Hurricane 4 or 5 would to really benefit.


mav

 

[kelticheart]

It's better to have a huge texture sheet as this is one drawcall per texture file. Also, if you use colors rather than textures for parts of the object, each color will have a drawcall... if those 'colors' are on the same texturesheet, there will be only one drawcall for all the parts.

This has been discussed in depth at FSDeveloper.com by Arno Gerretson.

Dick

Hence, this is why stock CFS2 aircrafts had only one mipped file for the regular textures plus one mipped for damage textures, and the aircrafts liveries were mirrored.

But that was necessary with 1999 computers. Even with my 8-year old pc, equipped with an ATI Radeon of the same age, mipped textures aren't necessary anymore.

Cheers!
KH :ernae:

 

[misson]

Mission,

Bottom line... in a cfs2 environment, for say QC and missions the more LoDs the better especially when it's a very detailed model and when you consider there's other moving and non moving objects like scenery etc ... really helps the F/R ,so you don't get a slide show effect.... feel most fs9 conversions lite to medium 2 or 3 LOD while something like Robert Sanderson's Hurricane 4 or 5 would to really benefit.


mav

Ok Mav ! I CAN ACEPT IT! but that is somebody says, but which is the esplanation for that, and why sometimes does not happen?

 

[Desert Rat]

Another, yes another, thing to take into consideration is that some models (especially FS9) render real gauges on the panels in spot view, this adds another load to the processor, as these gauges function as they do in the VC.

This doesn't really affect use as AI as panels are not loaded for AI, otherwise we could use greenghost gauges on AI too.

Jamie