SOS: Save our Smoke!

[PatrickR]

Noob question alert!

 

Is it intended in a future upgrade / release to move DCS to take advantage of multiple cores?

 

It runs so well ATM that the future possibilities of running on 4 / 6 cores are mind boggling!

[Kameni]

By transfering solely on x64 OS's limitations that occure as a result of stuff made to support x86 systems will be gone. That is a posible YES for you friend :)

Maybe not right away, but in time.

[TZeer]

@TZEER-I really am not understanding how you are getting the TiTan to run @ 100%. The highest I can get it is about 81% and I am running a i7 980X overclocked to 4.2Ghz. So you are saying that the CPU is still holding the system back? What CPU are you currently running and at what clock speed?

 

Regards,

 

Cobra

 

I'm running i7 3820 Sandy Bridge-E @ 4,6GHz

[ENO]

@ENO - overclocking from 3.9Ghz to 4.2Ghz did provide a gain in testing of 16% so Overclocking any CPU is a benefit. My question is how is he managing to get 100% load on the TITAN.

 

Regards,

 

Cobra

 

 

Not really my point, my friend but I can see from my post where it looks like that's what I said. My point was that an OC'd i3 is as beneficial as an OC'd 980X... and in some cases moreso. Glad you could net a 16% increase in performance from an 8% bump in proc speed... but that just emphasizes my point.

 

 

Like I said (and you confirm) speed is more important than the number of cores. If you had a 2 core processor cranked up to 5.3ghz it would beat your 980 to death (in this sim)- You were making it sound like you were surprised at the suggestion your CPU was a bottleneck (as was I, since we have the same one). I was simply explaining that without the speed of some of the newer processors, it may STILL be a bottleneck given the current inefficiency of the platform. Or not-

Anyway, I digress.

 

My question to you is what are you doing that gets it to 80%? The same thing he claims is raising it to 100? Sounds like he can line up smoke columns and crush his Titan... have you done the same in the offered track or mission?

Edited April 3, 2013 by ENO

[_Dredd]

If I get time I'll try this tonight.

However I have one question after going back through 5 to 6 pages. how is everyone comparing results if common testing specs/settings aren't listed first?

 

Are people running this test with HDR On or Off ?

[TZeer]

If I get time I'll try this tonight.

However I have one question after going back through 5 to 6 pages. how is everyone comparing results if common testing specs/settings aren't listed first?

 

Are people running this test with HDR On or Off ?

 

What settings you have is not that important. What is important is that your GPU is working at 100% through all the tests.

 

Then you do a test with the smoke. You now have a reference. Then you start removing effects like smoke etc. Run the test again, and watch for any difference in performance.

 

Do this multiple times so you know you can reproduce the effect if you notice a difference.

[ENO]

I suppose once someone can figure out what is cutting frames by about 80%, then we can start talking about side by side testing... I'm sure glad these guys know more about trying to eliminate layers than I do... my solution was to not look at smoke and start this thread in hopes someone could nail it.

[_Dredd]

Srry point noticed. I went through and re-read all posts again.

Heres a theory and have a read its quite interesting, and you'll find similar articles on google on particle effects.

As its was pointed out with smoke and fire for crashed planes etc, perhaps indeed what there is, is a particle effects code bug with 1.2.4? Memeory issues and polygon count extremities. But its not acknowledged as yet.

Check it out

http://wiki.yoyogames.com/index.php/Beginners_Guide_To_Particles

http://www.gamasutra.com/view/feature/131565/building_an_advanced_particle_.php?print=1

 

Well, before setting up a system it´s important to know a few things... First off is that most of a particle systems code is only ever called ONCE in the game, usually in some type of controller object in the first room or level (this is not always the case, but for the sake of simplicity, we´ll keep it like this!). This is because a particle system, once created, stays in memory and is ready to be used at any time, so if you create it more than once it can quickly up the memory usage and cause serious lag, before eventually crashing the game. In this tutorial we will make a global particle system that can be used by any object at any time.

 

 

Advanced particle systems can result in pretty large amounts of code, so it's important to design your data structures well. Also be aware of the fact that particle systems can decrease the frame rate significantly if not constructed properly, and most performance hits are due to memory management problems caused by the particle system.

Edited April 3, 2013 by _Dredd

[Rider1]

Topic:

I have found the files that control the smoke. Here is what I have so far.

 

First, I am testing my findings on one vehicle. A static fuel truck. The one I am using is at the harbor, just North of Batumi, sitting 50 yards from the cruise ships docked there. I use this one because it is one of the objects that when burning, drops my frame rates to single digits.

 

I am pretty sure that it is burning fuel that causes the drop.

 

Go here: C:\Program Files\Eagle Dynamics\DCS World\Config\Smoke

In here are two files:

JSmoke.cfg

JSmoke2.cfg.

 

If you look in these files there are settings for burn time, age, particles.. example:

 

[11]=

{

["value"]=1000,

["name"]="maxParticles",

["range"]=

{

[1]=0,

[2]=1,

},

 

If you look down at all these values, you will see that they are pretty low but you will also see that a few have settings of 1000. Very odd because most of the other similar parameters are not really over 5.

 

So I changed them to 1 and saw some improvement, the smoke is not as heavy, but still lagging my system. So I started changing all of the age and duration settings to 1. So now I have less smoke but the FPS drop only lasts 10 seconds or so.

 

I am probably going to change everything in there that looks like it has to do with smoke life, and particle values to 0 right across the board. and see if that takes care of this problem completely.

 

1) I do not know if one or both of these files need to be edited. In just did them both as they looked like they had a similar function and my patience with this is running out so i just said screw it and just took all of the parameters down in both files.

2) I do not know what it does to other burning vehicle smoke in the game, as I have edited the shit out of everything pertaining to smoke in the last few days.

 

Could someone back me up here and look at these two files, do some edits to see if you can see some improvement as well? Don't forget to back these files up first. I overlooked that. :)

 

EDIT: By changing every parameter to 0 in JSmoke.cfg, I was able to eliminate the FPS drop completely. :) However... It crashes my system every time I exit the mission now. :(

 

Could someone please post up JSmoke.cfg JSmoke2.cfg for me? Like I mentioned, I didn't back these up and I am going to have to edit and test line by line to see what parameter is causing the crash.. Thanks!

Edited April 3, 2013 by Rider1

[Milli]

msalama has had a look at these JSmoke files in his SmallFX mod which can be found here:

http://forums.eagle.ru/showthread.php?t=101083

 

There may be something of interest for you there.

 

Regards,

Milli

[TZeer]

Topic:

I have found the files that control the smoke. Here is what I have so far.

 

First, I am testing my findings on one vehicle. A static fuel truck. The one I am using is at the harbor, just North of Batumi, sitting 50 yards from the cruise ships docked there. I use this one because it is one of the objects that when burning, drops my frame rates to single digits.

 

I am pretty sure that it is burning fuel that causes the drop.

 

Go here: C:\Program Files\Eagle Dynamics\DCS World\Config\Smoke

In here are two files:

JSmoke.cfg

JSmoke2.cfg.

 

If you look in these files there are settings for burn time, age, particles.. example:

 

[11]=

{

["value"]=1000,

["name"]="maxParticles",

["range"]=

{

[1]=0,

[2]=1,

},

 

If you look down at all these values, you will see that they are pretty low but you will also see that a few have settings of 1000. Very odd because most of the other similar parameters are not really over 5.

 

So I changed them to 1 and saw some improvement, the smoke is not as heavy, but still lagging my system. So I started changing all of the age and duration settings to 1. So now I have less smoke but the FPS drop only lasts 10 seconds or so.

 

I am probably going to change everything in there that looks like it has to do with smoke life, and particle values to 0 right across the board. and see if that takes care of this problem completely.

 

1) I do not know if one or both of these files need to be edited. In just did them both as they looked like they had a similar function and my patience with this is running out so i just said screw it and just took all of the parameters down in both files.

2) I do not know what it does to other burning vehicle smoke in the game, as I have edited the shit out of everything pertaining to smoke in the last few days.

 

Could someone back me up here and look at these two files, do some edits to see if you can see some improvement as well? Don't forget to back these files up first. I overlooked that. :)

 

EDIT: By changing every parameter to 0 in JSmoke.cfg, I was able to eliminate the FPS drop completely. :) However... It crashes my system every time I exit the mission now. :(

 

Could someone please post up JSmoke.cfg JSmoke2.cfg for me? Like I mentioned, I didn't back these up and I am going to have to edit and test line by line to see what parameter is causing the crash.. Thanks!

 

Do you have the possibility to make a mission and place the unit in a remotely location for us to test? Just so we can be sure that the result is not affected by anything else?

[Rider1]

msalama has had a look at these JSmoke files in his SmallFX mod which can be found here:

http://forums.eagle.ru/showthread.php?t=101083

 

There may be something of interest for you there.

 

Regards,

Milli

 

Thanks. Those were what I needed:thumbup:

[Rider1]

OK I think I got it.

 

Could someone be so kind as to replace your JSmoke.cfg and JSmoke2.cfg with the ones I have attached?

 

They go in here: C:\Program Files\Eagle Dynamics\DCS World\Config\Smoke

 

Don't forget to back up your old ones first, right? :)

 

Then please run the attached mission. Shoot the Russian tanker and the Sue. Have a look at the burning wrecks and see what your frame rates are doing and report back here with your findings.

 

I get no FPS drop on my rig, well a maybe little because of the other effects going on, but the slide show is pretty much history.. Along with the smoke.

 

Fingers crossed...

 

Caveat: I have no idea if this affects any other smoke in the game, like marker rockets or burning ships...

JSmoke.cfg

JSmoke2.cfg

shoot the tanker.miz

Edited April 4, 2013 by Rider1

[SirBroud]

Thank you Rider1, works fine for me.

No significant fps drop noticeable.

There is still smoke, but it disappears very fast.

 

Another "smoke issue" for me is firing the guns on the p-51d, fps drop down to 15.

Anyone else having this problem? Deleting the gunsmoke.pfx obviously crashes the game, though deleting smokecloud.pfx doesn't.

Edited April 4, 2013 by SirBroud

[ENO]

Wow! Can't wait to try it out!

[SkateZilla]

Will give a run tonight after hockey.

 

so this kinda points to Particle Lifetime?, Emitted Particles last too long, and with New Particles being Emitted at a high Rate, eventually there's just too many particles and the game starts to take a dive (and this seems to happen within a few seconds).

 

So:

-Decreasing Lifetime of the particle seems to help.

 

how about decreasing the emission rate of the particles.

[TZeer]

Have good news and bad news.

 

Good news first.

 

This is before the modified files. 38 FPS

 

 

Here is after, with modified smoke files. 139 FPS

 

 

 

The bad news is that I saw no change in ENO's test mission with the fuel trucks.

 

 

Just a small tip to Rider1 or anyone else making test missions. Make them as far away from any other ingame "stuff". When I mean "stuff", I mean airfields, trees, roads, water, buildings you name it. You want the area to be as sterile as possible. This so you are 99% that nothing else is affecting your results.

 

I'm posting your mission again Rider1, where I have moved the units to a remote location :)

 

Hope you don't mind.

shoot the tanker2.miz

[wild turkey]

I'll take a crack at it aswell tonight after work. I would like to see in the test results, Hardware, Monitor Res and Graphics settings. I am running a 46" 1080p tv along with a 27" 144hz monitor at 1080p and a 1680x1050 20.5". Dual 680's in sli.

 

If these test are being run at min settings at 1024x768 or what not, it skews everything.

Edited April 4, 2013 by wild turkey

[Rider1]

Jeez TZeer Your before shots of the 38 FPS.. i am lucky to get 40 FPS consistently just flying a typical multilplayer mission under normal conditions :(

 

My specs: Win7 64 | ATI 6850 w Cat 13 driverset | AMD 2.6 Phenom Quad Core | 8GB ram

[ENO]

Well okay Tzeer- but let's look at it another way here. We've (well... by we I mean you guys) isolated two bugs... one is the smoke overall that, though it doesn't crush frames to single digits, still needed to be addressed.

 

The other is these ridiculous fuel trucks. If mission builders can go into their missions and eliminate these from the convoys then that should help resolve a good percentage of the problem.

 

And if it's NOT the fuel trucks- I'm wondering if it may be the structures...

 

test mission is edited slightly- removed the tankers on test 2 and also moved the overall out in the open a bit more to get the structures out of the equation.

Test.miz

Test2.miz

Edited April 4, 2013 by ENO

[Cobra99]

@Rider1 - I tested your jsmoke files in dragonshadow's Blindsided3 mission and what an improvement on the FPS during his intro bombardment of Batumi. As performance hits seem to be cumulative any instance addressed is well worth it especially to the CPU bottleneck ( in my particular case). I thank you for your commitment.

 

Regards,

 

Cobra

[TZeer]

I'm currently out of town, and will not be able to do any tests until Monday.

[Blooze]

@Rider1 - I tested your jsmoke files in dragonshadow's Blindsided3 mission and what an improvement on the FPS during his intro bombardment of Batumi. As performance hits seem to be cumulative any instance addressed is well worth it especially to the CPU bottleneck ( in my particular case). I thank you for your commitment.

 

Regards,

 

Cobra

I too was having major frame rate drops with the intro of the Blindsided mission but with your files installed I barely had any drop at all and the smoke appeared to look the same. That is an accurate test to me because all other factors were equal. By the way Cobra.....Dragonshadow is a she not a he.:megalol:

[Rider1]

I'm happy it is working out for you guys. Thank you. I spent the last few weeks rage editing the crap out of any file that had the word smoke in the file name in a selfish effort to make my game fun again. Glad it worked out for the better. Here's hoping 124 will fix these performance issues.

[ENO]

Viper? Any feedback at all? Dev input? Anything?