CPU Core number 3

So i don't need to still compressing my video for CPU/GPU/FPS graphic usage decrease if the team already work on this problem ?

what about cores 4-15 (assuming you're starting with Core 0)

Demongornot: I think we all understand, what you try to explain.

But as is said in this thread and many others: DCS World can not be broken in many parts and (as c0ff said) magically communicate with each other. And I am really sure, that the AI won't get better just because it runs on another core.

I read all of what you have been writing in this thread and still are sure that is not as easy as you think to go a "little" multicore. I am very sure, that ED is on it's way to real multicore support and Nevada will help that a lot. Btw, if you run your CPU on about 4.5 Ghz, it will perform a lot better than on 4.0Ghz Turbo. Thats what I experienced when I installed my new rig with I5 3570k, maybe you want to give it a try.

Demongornot, seriously,

I suggest you sit yourself down and learn programming, and then do this stuff yourself. It will be a very enlightening experience for you when you then turn back and look at all the things you have written.

This is a very serious suggestion, because it looks like you don't even understand that you don't understand.

@Neverminded

First, no, a lot of people just answer wrong cause they are too lazy to read the whole post, i assuming that i have write a long one, so now that i have clarifying things, people do understand.

And no, you never say this, and it can, like i have already say and like REAL THING actually DO and WORKING its possible, like Follow me car for FSX who using a splitted application, still able to work by receiving data from the game and input data to the game, like AI Carrier do, like FS Traffic, AI Carrier, FS Traffic, FS Passenger, PMDG/Ariane Design/VRS and other aircraft do (so don't told me that AI can't be switched into another application, mainly when AMATEURS can do it) so if an amateur who have NEVER even see from far the source code of FSX can get it work (just look at PGDM or VRS, they using their OWN physic/flight model, avionic and anything else and just use FSX asa graphic support and they still working perfectly) why the original coders can't get the sim to work like this ?

It CAN communicate, its called the information export, the same thing that can get the gauge working when someone create a cockpit for any aircraft, data exportation can also work for a lot of things, including AI and visual effect.

@EtherealN

I understand, and what i understand is that people told me that something ALREADY WORKING in another software its impossible to do, not just impossible with this software, but impossible at all, its the same logic than old scientist saying that Asteroid can't exist cause : There is no rocks in the sky.

A lot of people with really nice programming skill including one time, a guy who working for a great video game society (that i wont mentioned the name), a codder, told me that a lot of things that you, guy in this forum told me that it was impossible was finally completely possible and for some example already existing, one of my best friend is actually beta tester for Microsoft and have a really big programming skill and he totally agree with everything i have ever say in previous thread where i get the same reaction, and its definitely not someone who will lets me think something impossible.

Its not cause someone with a "unknow" programming level think that something already working for a lot of software was impossible that it is.

And by the same logic multiplayer is not possible cause after all AI aircraft that we see on MP mission are controlled by the server and not inside our computer, but yes i'm wrong and everyone know that DCS don't have and will never have multiplayer cause it can't work and AI can't communicate data...

And finally no one answer me when i ask if i need to upload to youtube my video for show FPS/CPU/GPU drop issue or if it won't help and i can avoid a long upload time... ?

Yes, it is possible for separately compiled programs to communicate data with each other, but considering that ED has complete access to the source code of DCS, why would they want to? Sans computing clusters, any advantage you might get from having separate programs you could also get (and more) from having a properly multithreaded program.

All of the mods you mentioned are 3rd party MSFS mods. They don't exist as separate executables because it is ideal, but because of the limitations of being a third party developer.

They are aware of visual effects (explosions/smoke) and the issues with FPS they are causing and are working on a fix.

Is it related to the CBUs and MLRS rockets?

Is it related to the CBUs and MLRS rockets?

Yep! Cluster bomb/bomblet explosions hit fps big time. As said unfortunately a bug. I guess too much particles in the multiple explosions or something like that. :dunno:

I guess too much particles in the multiple explosions or something like that. :dunno:

More like too much AWESOME from putting the hurt on a group of enemy armor with a single munition!

:joystick: :thumbup:

Demongornot, the problem is, that you can't change such a technology like flipping a switch.

Currently, we have a rock-solid house built upon a nice basement. Yet there is no cellar. Of course, there are houses with a cellar. Even houses from Microsoft with a nice, giant cellar. People told you it's possibel, even with different rooms, but still this ED-house still has none.

Now the problem is, that this is not a new house. It's a big old house and there are a lot of different things to consider. And you can just lift that house up, build a cellar and put it back on top. It simply doesn't work this way in our physical world and neither does it in software-business.

If you want fundamental changes, you either need a lot of money and manpower and build it from scratch. Or you have to do it with one step after another. And that's exactly what they are doing.

Ok. What is this "multithreading is so complicate" argument about.

I'm not a professional programmer, nor an engineer etc., but I know some basics on mathematic calculations back from school.

Let's look at a normal calculation: ^2 is for squared

2+4^2-8/2-2*2=10 well, this is not that complicated, you can do it in some 15-20 alone.

Now we speed up calculations by distributing it to 4 people (cores), easy right?

Let's see, there are rules that help distributing the task: squared is more important than multiplication or division and plus or minus is last...

So we have 4 tasks according to the people we call them a-d:

a. 2+ result of b

b. 4^2

c. result from a - 8/2

d. result from c - 2*2

Well, that is easy, isn't it?

The real problem is like in a computer bus system only one guy is allowed to go into the room every 5 seconds and put his result on the paper.

So after 5 sec: a enters the room and b's result is not there

after 10 seconds b says I need just some 2 more seconds... sorry, next

after 15 seconds c enters and no result from a

after 20 seconds d enters the room and no result from a

after 25 seconds a enters the room, still no result from b

after 30 seconds b enters and puts his result after waiting quite a while for his next turn

after 40 seconds, and c and d's turns wasted, a enters and has the result from b, starts calculating, but takes 6 seconds for it. So it is c's turn who still waits for the result from a, d waits for c so...

after 55 seconds a puts his results

after 60 seconds c enters and calculates his result

after 65 seconds d enters and finishes the calculation!

Job done with multithreading. More than one minute! Single core time: 15-20 seconds. :thumbup:

That is a slight idea of what is this complication in distributing tasks between cores or threads...:smartass:

It's more like...

Asking person 1 to walk at 3mph

Asking person 2 to walk at 5mph

Asking person 3 to walk at 7mph

Asking person 4 to walk at 10mph

...and yet still be at the same distance from the starting line at exactly the same time at any given moment.

Very simplified explanation:

Games are nothing more than a sequence of calculations and drawings per frame, and for each frame each thread has to be synchronized with the rest of them or else the game will start to exhibit very odd behavior. Some calculations can be done incredibly fast but then that process must pause until the frame is done and other calculations will take the entire duration of the frame to complete. These separate calculations being done at the same time are called threads.

If thread 1 only takes a quarter frame to complete, it cannot simply move on to the next task... it has to wait for the other three threads to complete and the frame to be rendered before moving on to the next calculation. Otherwise, in a very short amount of time (four frames in this example, thread 1 will be a full frame ahead of the rest of them (assuming, for explanations sake that the remaining threads are all running at the same speed). If we assume thread 1 is working on game physics calculations, the result would be a huge amount of incorrect values being applied to everything in the world (bombs would/could have steeper or much shallower arcs on their fall trajectory, mavericks could potentially travel faster than the speed of sound, etc) because the threads are not synchronized.

Now to take a single core game and make it multi-threaded ... it's not as simple as saying "ok make all physics run on core 1, make all sounds run on core 2, offload all graphics to GPU chips"... you pretty much have to go through and rewrite every aspect of the game so that it functions properly and in-sync with the rest of the game. You define your threads and then define what goes on within each thread for each frame of time. Then you have to make sure that they are processing in sync as well.

It is definitely not a simple matter of just "making the GPU do graphics". MUCH more complicated in practice than that.

:smartass:

So we have 4 tasks according to the people we call them a-d:

a. 2+ result of b

b. 4^2

c. result from a - 8/2

d. result from c - 2*2

 

Well, that is easy, isn't it?

That is a good example of something that can't be parallelized. Here a depends on the result of b, c depends on the result of a, d depends on the result of c. So even if you distribute it to more threads, it will end up being done in sequence and nothing is gained but performance is lost.

The real problem is like in a computer bus system only one guy is allowed to go into the room every 5 seconds and put his result on the paper.

Parallelizing only makes sense when you have multiple computation cores actually running in parallel. Actually gaining something from such a setup requires that the problem can be split in multiple independent parts that only occasionally need synchronization across threads.

Lots of problems simply can't be processed in parallel. For those where it is possible, it's hard to do when the code isn't designed to do so from the ground up. Converting an existing code base to parallel execution thus needs a lot of work and introduces a lot of potential bugs in the required synchronization.

And you better hope that in the end you gain more by parallel execution than you lose by the synchronization and communication you have to introduce.

That is a good example of something that can't be parallelized. Here a depends on the result of b, c depends on the result of a, d depends on the result of c. So even if you distribute it to more threads, it will end up being done in sequence and nothing is gained but performance is lost.

 

 

 

Parallelizing only makes sense when you have multiple computation cores actually running in parallel. Actually gaining something from such a setup requires that the problem can be split in multiple independent parts that only occasionally need synchronization across threads.

 

Lots of problems simply can't be processed in parallel. For those where it is possible, it's hard to do when the code isn't designed to do so from the ground up. Converting an existing code base to parallel execution thus needs a lot of work and introduces a lot of potential bugs in the required synchronization.

 

And you better hope that in the end you gain more by parallel execution than you lose by the synchronization and communication you have to introduce.

Exactly! And in a simulated environment, where hundreds of objects are closely related to model a "world" it may not be easy to find tasks that can be done in parallel, at all.

Actually, there are few applications that benefit from multiple cores at the moment. Usually these are math calculations or video rendering of video in asynchronous mode and of course not in real time, like in a game...:D

I understand, and what i understand is that people told me that something ALREADY WORKING in another software its impossible to do, not just impossible with this software, but impossible at all, its the same logic than old scientist saying that Asteroid can't exist cause : There is no rocks in the sky.

Actually, no, that's not at all what you are being told.

The fact that you think it is what you are being told is one of the things where it becomes clear to me that you do not understand the subject.

If you at least tried to listen when people try to explain to you how things like this do work, then arguing with you might work. But you don't. So arguing with you becomes rather pointless. Thus: please take a couple courses in computer programming. Once you get started on some larger projects, you'll start to understand.

A lot of people with really nice programming skill including one time, a guy who working for a great video game society (that i wont mentioned the name), a codder, told me that a lot of things that you, guy in this forum told me that it was impossible was finally completely possible and for some example already existing, one of my best friend is actually beta tester for Microsoft and have a really big programming skill and he totally agree with everything i have ever say in previous thread where i get the same reaction, and its definitely not someone who will lets me think something impossible.

Again, this is where you are just plain wrong.

A) You have NOT been told that it is "impossible". (Well, this particular suggestion right here probably is, but it is difficult to know for sure since most of what you say doesn't make any sense. I suspect this is a language barrier problem though.)

B) That it is possible to make an airplane break the sound barrier does not mean it's a good idea, or even feasible, to make a Cessna Skylane supersonic; to get it there, you would have to pretty much replace EVERYTHING.

Now go back to what c0ff said, read it, read it again, read it a couple more, google a couple of the things he mentioned, and then try to understand.

Finally, I would suggest that if you want to attack people with decades of experience with words like "amateurs" etcetera, you at least START with reaching the level of amateur yourself. QBASIC is a good start. :)

And no, you never say this, and it can, like i have already say and like REAL THING actually DO and WORKING its possible, like Follow me car for FSX who using a splitted application, still able to work by receiving data from the game and input data to the game, like AI Carrier do, like FS Traffic, AI Carrier, FS Traffic, FS Passenger, PMDG/Ariane Design/VRS and other aircraft do.

Unfortunately none of these actually mean it is handled on a separate core, or multithreaded and finally, it for sure does not help with framerates!

Nobody says it is impossible. It has been said it is complex, time consuming and ED is ALREADY working on a new engine for DCS since years...

one of my best friend is actually beta tester for Microsoft and have a really big programming skill and he totally agree with everything i have ever say in previous thread

If he by chance beta tested Windows8 did he notice that operating a touch interface with your arms held up in front for a whole 8 hour workday is totally nonsense and disqualifying Windows 8 for business use...? No, having the touchscreen lying on the table is no option, 8 hours bending your neck down to look at the screen is a health and safety no go, as well!

So these "professionals" you talk about, have they programmed a simulation environment themselves, or do they just talk about "how it could possibly been done", I wonder?

And last but not least, as an inventor you may know that what is possible is not always affordable for lack of time, money, ressources or a combination of these...:smilewink:

Ok ok i have understand, rather than see any good point of my idea you just see all potential negative point (maybe their is no negative point at all in fact but nevermind) and find a way to fix it.

I have understand you don't want to see any improvement before EDGE is release (in a lot of time, will be full of but and will kill performance)

And as an inventor i know that even if the first idea can't be use, we can just learn about the suggestion who was a part of the idea like the improved AI or anything else, but you just say : NO.

So ok stay with the game with the worst optimization of the video game story and the worst AI of all software i have ever see and keep told me that i'm wrong, i will wait for Seven G who like i have just said in another thread, have realistic and beautiful smoke (not a plastic one completely unrealistic who made a creepy and impossible cone at start of the smoke) who don't do any FPS drop and also to Arma 3 where the game even in Alpha have 200 time less bug than DCS and who have a support of almost a patch per day with one of the best graphic engine of the history (for a team far to be big like big game company).

The only thing who working correctly (and still have some bugs anyways) is the AFM and the avionic.

And you will say probably : at least it work...

Yeah but a car "simulator" with a perfect car reproduction who operate on planet like Mars (without oxygen, or just simple traces) with a lower gravity and where all the surface is water (you have understand that i don't talk about Mars but a planet with all other property except water the temperature) will be NOT a simulator, mainly if with the big wind of Mars like plane the car was not affected by weather and the graphic engine is not optimized to show water, it what the actual DCS look like.

So yes i'm totally wrong and DCS is a master piece that codder mastering at 100%, the landing gear who broke always at slow speed for FC3 aircraft (and where no update was done with the autoupdater supposed to be here for this) and the simulator who have FPS drop by reducing the use of CPU/GPU and who have bad performance without even using the whole computer power or low computer config who get more performance than the biggest one or even the fact that is common to get more performance in the game maxed out than minimum setting is the perfect proof that the simulator work well and that i'm totally wrong, the proof was that no one have the clue to talk about it and prefer just see the only "bad" point of my idea without even give the proof that it can't work, at least i have give example of SIMILAR working idea.

Still defending the game i will still laugh, and i will never give any idea for any optimization or graphic or anything else, like the graphic engine i have propose, that several codder who actually know a lot about graphic engine told me that it will work and then here people still told me that it can't work just cause they always act like this when someone talk about graphic engine or optimization of DCS, cause any optimization or bug change make people nervous and they don't want to hear about...

People are fun...

Anyways i will still give idea about gameplay cause now that DCS are release no other company see the need to create a new sim and we are stuck on this one, and i want to see GOOD and REALISTIC possibility even if we can't have a correct graphic engine and cpu/gpu management.

For answer to shagrat, for the last time : I have not say to make the game use a single apps who will use several core including one for AI and effect, i was talking about make a separate apps for EFFECT (who actually consume for a couple of explosion and smoke more performance than the whole game) and AI for have the possibility to make them use more power and get something better than just the "worst AI of all the story of video game" who have ridiculously simple script actually and no possibility for anyone to make them less stupid, and for people supposed to work on their own graphic engine who will handle multicore, (and who will magically communicate) who have already done a separate core for sound and who are supposed to mastering C++ and maybe low level codding, it wont be hard, mainly when AMATEUR can do it for other software without access to source code...

But nevermind.

And actually food tester do, for things like cuisine you can't judge without know it, that was not true for everything, in computer even a kid who have never hear about the word codding will easily see that DCS are optimized by using the feets on the keyboard, it contain more bug than working feature and the visual rendering and the effect are just ugly.

And without ear about FPS he will understand that explosion and smoke cause a ridiculous fps drop like no other game have ever do in the history of all video game.

So yes beta tester of microsoft are capable to coding, understand how work the game and how it was codding and capable to modifying some part of windows, at least for the one i know, and like everyone do when someone talk about graphic engine or optimization, they just talk about what they want and even when they use all the whole threat point per point they just make it sound like they want by carefully choose quote the part of the text they want, i have also talk about someone working for a big video game company that i have already told him this kind of idea and who totally agree, and at least the game he and the team of this company don't do more bugs than feature...

Anyways useless to talk about it anymore, you get what you want i will no more give idea for any kind of optimization and just still laugh at how the sim "try" to working correctly with all this bug, leak of feature and the total miss of optimization that can't get good FPS on computer than actually NO GAME are hard enough for consume all the computing power of the modern machine.

This post was writing with the : sarcasm and criticism engine 2.5 !

It's not that your idea is bad, using all available resources is a good thing. But not only is it not a new idea and has been brought up many times, you also make it sound a simple thing to do which it is not.

My dear "Inventor",

Please invent a way to make an existing legacy 1.5 million lines C++ code run more efficiently by using 4-8 cores and you have hit the Jackpot. You will be very rich and famous by solving a problem that the greatest computer science minds of our time still cannot solve.

Good luck,

A humble Software Architect and Programmer

you also make it sound a simple thing to do which it is not.

It pays to remember that when designing multi-core code, you may well run into unforeseen problems where one thread has to wait for another to complete, and thus end up running SLOWER than an analogous algorithm executed sequentially on a single core only! And all this of course gets even _more_ hellish if you're modifying old code - which, BTW, is something you shouldn't even DO unless absolutely necessary, because it's often easier (and therefore cheaper) to just junk the old codebase and start anew. Trust me - been there, done that, and I've still got the ulcer to show for it :chair:

G19 keyboard, G700 wireless mouse, G930 7.1 wireless headset, Razer Tiamat 7.1 Headset, Saitek pro flight rudder, Thrustmaster HOTAS Warthog, TrackIR 5, Saitek X52 Pro

Guys, don't mess with Demongornot. He's got 2 heads and 4 arms.

Yep! Cluster bomb/bomblet explosions hit fps big time. As said unfortunately a bug. I guess too much particles in the multiple explosions or something like that. :dunno:

Just wondering because that MLRS 'bug' has been around a really long time and still hasn't been fixed. Makes you wonder if ED is really looking into it or is it lip-service. CBUs have been decreasing FPS to a slideshow since DCS World was released a while back and the issues still haven't been resolved.

Guys, don't mess with Demongornot. He's got 2 heads and 4 arms.

Actually i have 3 joystick.

The Warthog the X52P and a simple Cyborg.

And i have 3 headset with Razer, Logitech and also a Tritton Ax Pro, so i have 3 head and 6 arms... :megalol:

And if you want i can also explain you why i have all this headset and joystick.

@Sorin i have a solution, change ED's coders...Maybe we will get something who work and no new bugs with new patchs...

Anyways like everyone know its a really old engine and anyways EVERYTHING need to be change, and if a solution to make the game itself running on multicore can be found, it will be a waste of time, like make exploding a nuclear weapon for making some Wind turbine turning.

And again (even if now i stop to talk about my idea) i have NEVER say to make the game ITSELF using a multicore, just using external application for a thing who take too much performance and another who need performance...

But when i see that DCS are the ONLY game i have ever see who rather than give us extremely ugly visual effect (for the release date of DCS World, NO other game have relative to the date a so bad effect like this, also true for graphic rendering itself, compare with ROF or Aerofly pro FS for example) and its the only one in the world who rather than make ugly effect, make them kill all your performance. :noexpression:

So yes i will still think what i think about ED codders until a see a REAL change, i'm not pretending to be actually able to do this, but every other codder with their (supposed) level of knowledge about codding who make game do better in all side...

Anyways enough talk about this too, everything is already writing. :(

Demongornot, I'm quite surprised how patient people have been with you over the past 5 pages of this thread but please show a bit of respect with regards to your wording towards the ED programmers.

I've done my best to try and figure out what you're trying to propose and as multiple people have pointed out, it's just not possible.

It'd be like me coming here and telling you to plug in all your headsets and joysticks and trying to use them all simultaneously with DCS World...or any other program for that matter. go ahead, give it a try...

Ok ok i have understand, rather than see any good point of my idea you just see all potential negative point (maybe their is no negative point at all in fact but nevermind) and find a way to fix it...

Man, are you trolling or are you really too stupid too understand?

1. ED is already doing what you ask for since years, like Microsoft is always developing the next windows...

2. You're idea is as old as Black Shark, but still it will not be useful to get some AMATEURS on board, to fix the MultiCore problem because they will start from scratch, run into all the mentioned problems and take some 5-8 years to get a working "new" multicore engine done, if they don't quit.

3. You know there is this brilliant solution that the people in 3rd world countries won't need to suffer from thirst or hunger. We simply freeze fresh, clean water in the industrial countries and ship it there, as well we can do with our leftover food... Well it is a very old idea, even AMATEURS can do it... freeze the ice ship it, voila! Fresh water for Africa.

I leave to your own imagination why nobody(!!!) has realised this easy solution until this day!

To get things done a company needs money to pay their people. So who is paying for this development, oh, wait that is us.

So ED should raise the price for a module? Let people pay for DCS World? Have DCS World 2, DCS World 3, DCS World 4 released, rather than patches for free? I guess the people whining about $50 for a module will not like this idea... Or we raise the price of a module to $1000 to cover the costs for development?

Even YOU must see, that it is not just asking 2-3 AMATEURS to fix that code from ED in some 2 Weeks... Ridiculous! :doh:

Last: ever done a project? You know throwing people into the project does not speed up things. Make an amateur course for project management and they will tell you why, if you are not brilliant enough to find out on your own.