Network (split)

[Jona33]

Demongornot: I think there's a problem your end. The graphics are absolutely fine on high even on my fairly crappy rig. Also I see none of these multiplayer problems except for like I said, occasional crashes.

 

I don't see how you think that VOIP will make people act more responsibly. People will act however they see fit. If they want airquake they will fly airquake. I've been guilty of it in FC2 because at the end of a day I want to piss around and have a laugh with people who want to fight. If I want co-op play our group will get a session running. The last time I saw lag was in a server with eight of us playing while trying to tank in the clouds. The server was in London and several people were on the other side of the Atlantic. It really wasn't that bad. FC2 was quite bad, as far as I understood it the gap between the server registering anything and the client moving his stick could be large, the game then warped the plane to it's corrected position. In DCS I don't notice this at all. I believe that a new system was put in place that moved the client aircraft smoothly to the right position. Please stop complaining about everything and fly the bloody games!

[EtherealN]

4 screenshots of different network test with two who test the packet lost and the result : 0%, and not only in my country.

 

And again, you are not understanding the point.

 

Read what you were told again.

I ran similar tests when I had trouble with the server the other day, and guess what? Results were perfect! But it was still my connection that was causing the issue! Go back and re-read.

 

Seriously Demongornot, you need to take the time to read what people tell you, and then attempt to understand it. All you have done now is say the same thing again, and the same response still applies.

 

All you have shown is that your connection does not have any issues connecting to those servers that run those testing sites. Which is fine and dandy, but since the game server isn't running on those test sites...

[Bearitall]

And again, you are not understanding the point.

 

Read what you were told again.

I ran similar tests when I had trouble with the server the other day, and guess what? Results were perfect! But it was still my connection that was causing the issue! Go back and re-read.

 

Seriously Demongornot, you need to take the time to read what people tell you, and then attempt to understand it. All you have done now is say the same thing again, and the same response still applies.

 

All you have shown is that your connection does not have any issues connecting to those servers that run those testing sites. Which is fine and dandy, but since the game server isn't running on those test sites...

 

I agree game servers and test servers are two different worlds...

[EtherealN]

Bearitall - yup. :)

 

Especially in cases where a game server is relatively often sitting in someone's apartment. You can't always expect a quality connection there - especially since the OTHER guy's internet connection can also affect you. If he has connection issues - guess what, it'll affect you even if your end is fine, because the data has to go from you, to backbone, to him, and then back. If anyone has troubles there, it can mess you up.

 

But I give up.

[Demongornot]

Lol i just have to read both of this testimony :

http://forums.eagle.ru/showpost.php?p=1429087&postcount=10

http://forums.eagle.ru/showpost.php?p=1429326&postcount=23

http://forums.eagle.ru/showpost.php?p=1429353&postcount=25

And laugh when i read that the problem don't come from the software but from the connection :D

Maybe too much packets loss in the Lan cable... :D

 

More seriously.

@Jona33 : The graphic that depending, you will be just happy if we have more but for you its enough, some people take pleasure to play with very old game and don't need any visual rending for immersion, its your case, other like me need realistic graphic rending for immersion.

The problem are simple : the effect are horrible (look at explosion) the smoke are not really good, same for the flame, it we forget all effect who are missing like vapor on wing or trans-sonic vapor cone and a lot of other important and other who are only a simple details but who can do the difference between a bad or a simple graphic rending compare to an excellent rending.

The color are totally false at 100%, the cockpit look plastic, compare with cockpit addons in Falcon 4 who are totally photorealistic.

Look that http://www.dsogaming.com/wp-content/uploads/2011/10/air_507a_005.jpg and told me without lie that you don't want to see the same cockpit color realism in DCS, realistic basic color and light reflection ! Its just awesome and hard to see if its a true cockpit on a sim, the its not only a texture color problem (but if since the begin realistic color was using we now don't need to change it and it will be simply better) but the problem come too from the light reflection who are totally false in DCS, the sun color too, when i play the sim that look like if i have a yellow filter front of my eyes, only good points, that offer a good render with the yellow windshield mods, but that's all.

Ground texture are so pixelized (with max setting) and the color are more false than what we have in the cockpit, same thing for the grass, the grass model are nit good, look at Rise Of Flight, the grass are awesome, maybe only a 2D animation but still be 100 more realistic, and if we just modified it a little for see it always front of us we don't will see the 2D effect and that will be just perfect and at high speed the rotation effect will not be perceptible, anyways we can tweak that, same thing for the tree (who are not collidable i remember you this "details") who are really bad, same thing for building and house texture...The ground miss so bad of diversity, no one little bush, no little flower, nothing who give a little Summer color, over that we have a sky and the water with a false color too, the clouds too are not really good, look at this screenshot :

http://i264.photobucket.com/albums/ii172/Soya4me/TV%20Setup/FSX/FSX%20Resized/CLS%20DC-10%20Part%201/2009-5-23_21-29-9-303.jpg

http://i264.photobucket.com/albums/ii172/Soya4me/TV%20Setup/FSX/FSX%20Resized/CLS%20DC-10%20Part%201/2009-5-23_22-6-35-708.jpg

http://i264.photobucket.com/albums/ii172/Soya4me/TV%20Setup/FSX/FSX%20Resized/CLS%20DC-10%20Part%201/2009-5-23_22-24-9-113.jpg

http://i264.photobucket.com/albums/ii172/Soya4me/TV%20Setup/FSX/FSX%20Resized/CLS%20DC-10%20Part%201/2009-5-23_22-54-31-303.jpg

 

Look at the diversity in the ground of Arma :

http://img32.imageshack.us/img32/9196/arma2ingamescreen040931.jpg

http://img32.imageshack.us/img32/2134/arma202helicoptercombat.jpg

http://img32.imageshack.us/img32/9271/arma2int200808141223113.jpg

http://img29.imageshack.us/img29/8766/armaiiscreenshotpcall43i.jpg

Yeah its great ! that look like a real life, unlike DCS where we have 5 things : grass with false color (for every season), river, city/road, forest and mountain, nothing more...

Add to that the bad size effect, look that :

http://img41.imageshack.us/img41/8766/armaiiscreenshotpcall43i.jpg

http://img29.imageshack.us/img29/7623/armaiiscreenshotpcall43n.jpg

In DCS we hare a guy of 3m height who CAN'T come close of the ground and who have eyes who have the size of two car wheel...

DCS = capable to show at 10Km (more with Arma 3 who have better graphic rending in fact) and i have already explain that over 5Km (except for giant building or antenna) its useless to show 3D object cause in real life or in a video game we CAN'T see difference between 2D and 3D object at more than 3Km, 5Km its really enough...Over this distance shot ONLY 2D map don't will take big power, less than what the Arma 2's graphic engine take with 5Km view, and now we can tweak it cause a lot of details are useless for an aircraft simulator, we just have to integrate this tweaked beautiful 3D engine with a 2D sphere (like a lot of game do, example with X3 Terran Conflict or RC plane simulator where the map are awesome, use the rest of the computer power for calculate color changing to this 2D textured sphere and we obtain graphic close to Arma 2 from close range :

http://img25.imageshack.us/img25/6284/armaiiscreenshotpcall43.jpg

And the same rending in medium/long range than what Aerofly FS have :

http://aerofly-sim.de/wp-content/gallery/aerofly-fs/aerofly-fs-discus-suisse-04-20111216-193230.jpg

Both look extremely nice and both together can permit to show giant map scale like DCS with EXTREMELY perfect rending and a lot of possible visual effect.

Look that : http://img708.imageshack.us/img708/8205/21229582.jpg

http://monimag.eu/upload/878/afs-04.jpg

http://monimag.eu/upload/878/afs-05.jpg

 

Yes all 3 are ingame, and don't try to told me that its not a PERFECTLY nice graphic engine who are totally different of DCS and who look just perfect, well ok maybe its not realistic due to shadows and other atmospheric condition who are not synchro with the simulator, but we can using the graphic power of the computer for modificate in real time color effect and adding preprogramming shadows on the mountain (anyways the mountain not will change of position :D its useless to always calculate the mountain shadows on itself...)

 

Same thing for the ground texture color and the aircraft color who look so many realistic that its really hard to believe that its not a simple real picture...

I will stop here cause the list of example are GIANT and its not the thread here...

 

But yes DCS are full of problem and the graphic rending can be 10000 time better than actually, mainly when we considerate the HUGE power of our actual computer (including the lastest graphic card who can run the lastest game who are close to real time photorealistic with 3 screen with nice FPS number).

If only for this "far to be good" actual graphic DCS don't need big power for run, ok i not will say anything, but use 100% of an HD6970 for show ugly result i can't accept that without react...

And just for information i have to a test with my I7 920 and my HD6970, both without overcloack, i have push my CPU to +15% and my GPU from 880 for GPU and 1375 for memory to 925 for GPU and 1410 for memory, and i was really surprise to see how many FPS more i get in DCS...->0

Same thing one day during a FPS/Eyefinity Test (without eyefinity) i have one time get MORE FPS with MAX graphic setting than the SAME scene (the A10C presentation scene) with MINIMUM graphic, sorry but that only shot how many its totally BAD and have NO ONE optimization, just a basic graphic engine optimization and NOTHING more.

Best example, for another flight Sim (FSX AND XPlane) this addon carrier in FSX http://www.referenseo.fr/presse/wp-content/uploads/2012/02/USS_Nimitz__USS_Eisenhower_2.jpg and http://royalefrenchnavy.gratisim.fr/Navires/USS_NIMITZ_IKE_PORT.jpg

and this carrier in XPlane 10 http://www.youtube.com/watch?v=GPQxWLo5PZU both full of aircraft and with animation...Both take LESS performance than the DCS's carrier who have NOTHING on the deck, less animation and poor texture...

Now try to told me that its normal than in both simulator we get better visual for less FPS...

 

 

I'm really sorry to insist about graphic but all i have say its totally true here and proof that the problem don't come from me...

Maybe for people who don't need graphic rending for immersion can be happy but more than half (including people who don't have buy the simulator due to this fact) it actually bad or horrible, i have test with a guy who know a lot about programming, who are really nice for photography and he have say, its horrible that look like if we play at Simcity...(two people who don't know the simulator world have say the same thing, proof that people who know it are just accustom to see it but for every other the result are always the same, its bad and ugly (except maybe people who have never see graphic from computer like me the first time i have see a simulator that was FSX).

 

If DCS was really optimised when we was on an airport we must be able to see the same rending than that (minimum) http://i264.photobucket.com/albums/ii172/Soya4me/TV%20Setup/FSX/FSX%20Resized/CLS%20DC-10%20Part%202/2009-5-24_2-7-20-322.jpg

http://i264.photobucket.com/albums/ii172/Soya4me/TV%20Setup/FSX/FSX%20Resized/CLS%20DC-10%20Part%202/2009-5-24_2-31-18-520.jpg

http://i264.photobucket.com/albums/ii172/Soya4me/TV%20Setup/FSX/FSX%20Resized/CLS%20DC-10%20Part%202/2009-5-24_2-40-12-635.jpg

Cause when we are on the ground and that we don't move fast the simulator must integrated that and be optimized with different parameter for every airport depending of what we see from different airport and when we take off use parameter of the low flight optimization, THIS is a real part of optimization who missing in DCS cause that we flight at the fastest speed possible with F10 view (more than mach 10) or that we still static and look always the same thing, we don't have any FPS difference, sorry but its one of the biggest proof of the really missing of optimization...

 

 

And for stop about graphic (anyways its far to be the only part full of bug and of problem, like the fact that every patch add new bug or the AI and a lot of things)

 

No VOIP don't will make people piloting like if that was real, but VOIP + the huge number of missing feature who make DCS not a real simulator but a realistic game (cause the aircraft itself are perfect but the rest of the sim are bad) like the actual mission, my Battlefield idea its a good example : http://forums.eagle.ru/showthread.php?t=84001

Actual mission are only based to unrealistic thing, my dynamic battlefield idea can permit to embody A SINGLE pilot for a long period with OUR aircraft, OUR country, OUR assignment and a lot of things who will finally don't do a single flight without any suite, but a true conflict where every act count and where if we do stupid things (for avoid troll) that not will affect our country but we must have to pass again flying class for flight again on this server, same thing for ejected pilot who really need to be rescues, base manager will assign several aircraft and helicopter (or ground/sea vehicle) to this rescue mission and spy aircraft will REALLY be useful for the first time unlike today where its only here for make a role play of theirs use...

 

With a lot of feature like that who are EXACTLY like real life and who augment immersion we will see more serious people, and less serious people will anyways keep flight on the standard server like actually.

 

 

 

@EtherealN & Bearitall

I read what you write, the screenshot are just here for shot that i have a "normal" connection speed, its the same ping when i play on any game server, the same band switch when i download and same for upload, the test i have made its just for show that i don't say : "i don't understand why that lag i have a good network connection" and finally have 300 of ping, less than 1MBPS in DL and less than 0.2 MBPS in Up, cause in this case yes its sure that my connection was the problem, but its not, and for game server its easy to see what is my performance, i just have to play with several online game, in the list :

Arma 2 (all extensions), Assassin's Creed Revelation, Battlefield 3, Dirt 3, Fear 3, Portal 2, Rise of Flight, Splinter Cell Conviction, Supreme Commander 1, 1FA and 2, Unreal Tournament 3 and several other and in ALL this game, that i play with people in France, with Hamachi or in the official server with people who host at the other side of the earth its the same thing, i have a nice connection its totally playable and that don't lag, and when that lag its soft and hard to see it, its FAR to be like what i have in DCS, i don't care that every game/software are different, its always the same type of information who are send : position, attitude and actions, if other game can without lag show the position of any unit without have just a little lag, DCS can do it, for example in a game like Arma, we have the position of the player in the map like DCS, the attitude (vehicle, aircraft or where the player look) completing things like where the player look like and where the gun are aim, what weapon he actually carry, wish is in in their hands, the body attitude and more, like in DCS, nothing else and nothing more to transmit, if ANY OTHER game can do it, DCS can do it too...

 

I have better quality of fluidity in the game with any other with : a bad day for network + download and playing with how who have a really cheap connection in another country, than : DCS with firewall disabled, ports open, connection/router restart just before playing (better bandwidths) and playing with people who have a really huge connection and in my country

 

Sorry but when i see that its easy to say after that the problem come from DCS and not from any connection and computer, mainly when we know that the multiplayer know problem during LAN party...

[EtherealN]

for game server its easy to see what is my performance, i just have to play with several online game, in the list :

Arma 2 (all extensions), Assassin's Creed Revelation, Battlefield 3, Dirt 3, Fear 3, Portal 2, Rise of Flight, Splinter Cell Conviction, Supreme Commander 1, 1FA and 2, Unreal Tournament 3 and several other and in ALL this game, that i play with people in France, with Hamachi or in the official server with people who host at the other side of the earth its the same thing, i have a nice connection its totally playable and that don't lag, and when that lag its soft and hard to see it

 

Again, this argument just shows that you don't understand the point I made.

 

I'll try again.

Look at this illustration:

 

This is an abstracted map of the internet. I've marked some servers, and a couple switch locations. Now, let's go through what happens to YOU if there's a problem at the numbered locations:

1) If this one's broken/congested, you'll have issues in DCS, everything else works fine.

2) If this one's broken/congested, you'll have issues in DCS, UT, and the selected PINGTEST* server.

3) If this one's broken/congested, you'll have issues in UT and PINGTEST*, everything else works perfectly.

4) If this one's broken/congested, you'll have issues in everything.

5) If this one's broken/congested, you'll have issues in ARMA, everything else works perfectly.

6) If this one's broken/congested, you'll have issues in PINGTEST*, everything else works perfectly.

 

*Pingtest is however a crap system for spotting such problems, just like all the other similar services, since it normally autoselects the fastest one it can find. Thus, it removes the problem sources from the sample on purpose. You CANNOT use it to find a problem UNLESS the problem occurs at position 4.

 

And keep in mind, this whole thing is extremely simplified. In reality it is even more complex since routings can change ay any time, different things can happen depending on protocols used, packet payload weight etcetera etcetera. The key point though is that if everything except one of these applications works fine, it can STILL be the connection that is the problem. For example, if this map were an actual representation (in actuality there'll be plenty more hops on the route), and you had issues on your ARMA server, but nothing in the other ones, it can EITHER be a problem with the server itself, or the game, or the switch at position 5. Through testing the other ones all you've done is pretty much establish that the problem is not at positions 1, 2, 3, 4 or 6. But you have NOT ruled out problems at position 5.

 

To do that, there are other tools you can use. You can test the route manually, you can sniff your packets and see what's happening to them (though this is hard for UDP since they are one-way only, your computer does not get a "receipt" for them), you can test a different connection and see if that one routs differently (like was the case in the example I gave) etcetera etcetera.

 

Key thing is that if you have these two candidates:

A) Switch position 1.

B) DCS Server

 

...then you cannot simply say "it has to be DCS program". What you have to say is: "my candidates are the local switch, the server, or the server software". That is what we have been trying to explain to you. :)

 

If you want a more actual "map" of the internet, this one should get the idea across: http://blyon.com/blyon-cdn/opte/maps/static/1105841711.LGL.2D.1024x1024.png

 

EDIT:

Also, Demongornot, the subject of this thread is Networks and netcode and your suggestion for "improvments" there. You already have plenty of threads where you talk about graphics. If you want to continue those discussions, do that there. Please keep rule 1.5 in mind - I already split this thread once, I won't do it another time. Thankyou.

Edited April 2, 2012 by EtherealN

[Demongornot]

Thanks for your illustration EtherealN.

 

But the problem is that its a client mode and not a P2P.

 

What about a mix between P2P and client ?

All information of aircraft are send by everyone to everyone in P2P and all other information (mainly the AI and any other information that host server share with client except player position) with client mode ?

 

With this solution we can see maybe one or two plane lag if the connection between them have a problem but we can't see everyone lag due to a problem between us and the server, right ?

 

Not really a mix between P2P and Client but a P2P with a Master, like the difference between seeder and the source in a download with a P2P software...

The Master P2P its the guy who will manage every other information than aircraft position, that not change a lot compare to a standard P2P in fact, just that if we have a problem between us and the server only the information directly relative to the server and the position of the aircraft of the guy who host will lag...

 

And a backup solution will be : if the game measure that so many packet lost between two pair to pass through another pair who have the best network speed between two pair with a problem between them for relay information.

 

Cause if i see the host lag but another pair not its cause the connection have a problem between the host and me, but between another pair and me if i don't have problem and that this pair have a nice free bandwidth, we can using the P2P system for relay the information between host and the pair who both have a problem together and pass through a people who don't have a lag problem with both.

 

This system can permit if the host crash to anyone to take the relay and finally don't make everyone disconnected, same if the host have a bad connection speed, the information that the host transmit can pass through another pair who have a nice bandwidth with the host and relay it to everyone and finally people with nice bandwidth will profit of it by helping everyone who have a connection lag between them but without loose MP quality finally, that will profit to everyone, including in case of "server" crash

Same thing if one of the pair who are using like a relay disconnect, the relay will be switched to other people with nice bandwidth, and if its not possible cause no one have a nice bandwidth for be a relay that will lag but anyways without my dynamic P2P idea that will lag too and with everyone (or a lot of people) and not only between two people unlike my Dynamic P2P do.

The possibility for people who create a really big mission to pass the server host to anyone who want can be nice too and can permit to see a server who not will be end before the not not cause of crash but cause of voluntary disconnection of the host.

 

And if ED decide to include a VOIP that will be better than a Client server mode and for people who have slow download speed block VOIP when its not on the same frequency can permit to finally don't loose bandwidth for something that we don't use...

 

What do you think about my idea ?

I hope it can help.

 

And sorry for Rule 1.5 i was answer "Jona33" last time i will post my answer in a thread who talk about that and just link my answer rather than right it here, do it respect the rule ?

Edited April 2, 2012 by Demongornot

[EtherealN]

What about a mix between P2P and client ?

All information of aircraft are send by everyone to everyone in P2P and all other information (mainly the AI and any other information that host server share with client except player position) with client mode ?

 

Not practical at all.

Let's say I have a 1mbps internet upstream connection. Normally we could expect to need something like 128kbps to be used in DCS. (Approximately - I'm relying on memory of some testing we did way back and I might be wrong.) Normally I send this information to the server and the server handles the rest for me.

 

In a P2P model, if there's 20 people on the server, every single user is liable to need 20x0,128Mbit=2,56MBit upstream. As we can see from your previous post, that means you are not playing because your internet connection LITERALLY isn't good enough. ;)

 

Now sure, you could segregate which information get's sent by type, but you'll also have the issue that there is no-one that is an authority about what your plane is doing. You can easily end up dead on one client and alive on another. Not to mention the very serious threat of cheats.

 

Further, P2P models require everything to be extremely deterministic. Starcraft 1 had problems with this in the beginning - one player could see a completely different game going on that the guy he was playing. They solved this by ramping up the deterministic nature of everything that happens in the game (and in SC2 they gave it a final solution through moving to a fully server-based system through battle.net). The problem is you really can't make a simulator like this deterministic enough, due to granularity of the simulation (the simulation runs in frames too, not just the graphics).

 

And a backup solution will be in the game measure so many packet lost to pass through the pair who have the best network speed between the host and the people for relay information.

 

You cannot measure packet loss for UDP traffic. There is no handshakes, no sessions. To "measure" packet loss, you need TCP. But TCP is slower, way slower.

 

And if ED decide to include a VOIP that will be better than a Client server mode and for people who have slow download speed block VOIP when its not on the same frequency can permit to finally don't loose bandwidth for something that we don't use...

 

Voice and game data can use different models. Starcraft 2 does this. However, again, my worry is basically about return on investment. Is implementing VOIP native to the game worth the expense?

 

Remember, these are not hobbyists. These are people that do this as a job. If someone makes a mod on their spare time and decides to give it to you for free, that's fine. But whenever something is to be integrated as an official part of the simulator, being a hobby is just not enough. And then it starts costing money in design, implementation, testing, optimisation, etcetera etcetera.

 

What do you think about my idea ?

I hope it can help.

 

Unfortunately I don't like it, as explained above.

I really think you need to get yourself some experience of handling software development in some capacity - you don't have to be a programmer yourself to get valuable experience (I'm not), but you need to get a workable understanding of the process and the difficulties inherent in the technology. (And learn the humility to listen to what you're being told when someone who actualyl IS a programmer and does know the relevant source tells you something. Everyone is liable to be wrong at some point and the only way we can stop being wrong is to listen and learn. ;) ) What you are currently doing is back-seat driving at the same time as we end up having to explain the most basic internet infrastructure topics to you, and with respect - that's not the time to distribute recommendations on how to handle major software development projects. ;)

 

My suggstion would be to pick up a book on how to program, and make yourself something like a 2-player pong game. That's fairly simple and you should be able to get something functional within a couple of weeks even with no prior experience. Then make this 2-player pong game run over the internet to another computer and see what a huge can of worms just got opened when you had to handle internet traffic even for something as simple as a pong game.

 

And sorry for Rule 1.5 i was answer "Jona33" last time i will post my answer in a thread who talk about that and just link my answer rather than right it here, do it respect the rule ?

 

You can link to other threads, no worries, just try to keep each thing into their own topic. Thanks.

Edited April 2, 2012 by EtherealN

[Demongornot]

@EtherealN

 

Ok P2P will delete the problem of lag but will finally be impossible due to bandwidth that it will require and will cause problem of possible desynchronization.

And we can't measure the packet lost between two pair in UDP, and about that how can i force DCS in TCP ?

 

But what about a mix of Client and my idea ?

Just the "relay part" for people who have huge bandwidth like people who using fiber, an option of "relay" (with a gift like exclusive skin or anything else who can motive people who have a nice bandwidth to enabled this option)

 

Cause we can't detect packet lost, but we can easily detect lag, mainly when its hard, and if we have a problem between us and the server, we will see everyone lag and everyone will see lag and maybe for a little problem between us and the server but with a good bandwidth for everyone.

If lag was detected between us and the server (anyways in a client system the lag can only be here) we don't receive player information from the server but from the relay, people who have nice bandwidth don't will loose anything to be a relay, mainly for people who can reach more than 50 real Mbps, now we receive the same information than what the server are supposed to send us but before coming to us that just pass through another people and we don't need extreme bandwidth, same thing for the server who will always send the same information, just that rather than directly communicate with people who lag, it will send directly to the relay guy, and in this case the only think that the relay guy need its a not bad computer and a nice upload bandwidth.

Other player will don't see anything changed, them will simply exchange directly with the server.

 

maybe we can have a little delay between what the people depending of the relay see and what it is in real, but finally its better than always have a lag and better than force other people to see us lag.

 

If i check, relay guy with huge bandwidth will not loose anything except a little % of power in the CPU and a little part of the upload bandwidth, people who hots everyone don't will see any change, other client who are connected to the server don't will see any difference cause it not will have except to not see lag from a people who are supposed to lag due to a problem between him and the host, and the guy who are supposed to lag don't will see any change except a little delay but no lag...:D

 

What you think about this new idea ? :megalol:

 

 

And if when the official SDK will be release, anyone create integrate and test itself a VOIP system including the quality of TS3 and the advantage of TARS but with the possibility to using encrypting com panel, did you think that, due to fact that its almost already finish (just maybe a little details to correct) ED will automatically integrate it officially into DCS (with the green light from the creator) or them can refuse to integrate new feature ?

 

 

And ok i get it, anyways i have planed to learn before the end of this year the programming cause of a big project and for my work i will need to create myself several software...

[GGTharos]

@EtherealN

 

Ok P2P will delete the problem of lag but will finally be impossible due to bandwidth that it will require and will cause problem of possible desynchronization.

 

... which is usually what lag is.

 

And we can't measure the packet lost between two pair in UDP, and about that how can i force DCS in TCP ?

 

You cannot, and you will never get the option. What a completely horrible idea. There used to be a TCP option for LO1 (IIRC) and unless you were on a LAN, the lag was horrific.

 

What you think about this new idea ? :megalol:

 

... I think you still have no idea about what you are talking about, or what the consequences are. 'A little delay, but no lag' ... just what do you think lag is? Big latency = bad :D

[nomdeplume]

TCP would make things far worse, because if a packet is dropped the entire stream will hang until the drop is detected and the packet is re-sent (TCP guarantees in-order delivery of all packets, has built-in retries etc., which is outside of the application's control). There's a reason almost no real-time apps (games, voip, streaming video etc.) use TCP.

 

You could still measure packet loss with UDP though, especially in a game where you're expecting each client to send you packets at specific intervals. In the simplest case, you could just include a "sequence number" in each packet the client sends, and if the server is expecting packet #10000 from a client and instead receives #10001, it knows one was lost. This would also treat packets that arrive out-of-order as 'lost', but that's probably desirable (no sense updating a game object with old information).

 

I actually think it would be very helpful for diagnosing issues and determining where problems are if the server did collect metrics on player connection quality and reported them in the player list screen. Currently it just performs a ping test periodically and reports the time, but it could 'easily' be extended to provide some actually useful information:

 

- keep results of the last e.g. 5 pings and display the average (mean) value and the amount it varied by so you can estimate jitter. So instead of just showing the ping is 50 ms or 300 ms, it could show "65±12" or "302±132" - which would give you a better idea of just how crap the connection to the client is;

- count the number of packets successfully received and not received over the same interval and report it as a percentage of packet loss - or simply a count of the # of packets lost.

 

That way you could see at a glance if someone's connection is terrible, and if everyone has a terrible connection then it implies something is up on the server side of things (maybe saturated bandwidth or just a crappy link). Whereas if everyone is showing low jitter and 0 packets being lost but there's still weird stuff going on, it'd point to issues with the code.

 

Don't forget the game also has a track replay feature. It's a little bit different to multiplayer, but it still deals with the basic concept of transferring events that occurred in one instance of the simulation to another instance. It doesn't rely on the unreliable internet, but still routinely fails to play back an accurate record of the events that occurred in the game. So, assuming the netcode is as good as it could reasonably be and that any issues are because of the big bad internet is premature, IMO, when we already have proof that the game can't be relied on to recreate events even when using a lossless transport medium with practically unlimited bandwidth (local HDD).

[Demongornot]

... I think you still have no idea about what you are talking about, or what the consequences are. 'A little delay, but no lag' ... just what do you think lag is? Big latency = bad :D

 

In fact i don't have realize that i use a bad word, English are not my first language sorry...

When i talk about "lag" in fact i want mean the corruption of the aircraft position by packet loss who will do it crazy thing like moving 10 m away back to initial position, slide 5 m left, than 3 m back, be inverted and a lot of thing but in fact the player was immobile on the parking and do start up...

 

And yes in fact the lag are not that, but in France we say "lag" for this phenomena, i have mistake the same world using in France and using in English...

 

In fact we can have a delay between two computer without have a crazy position problem, if the information are not corrupt and send at the same debit not matter if the information come 1 minute or 1 hours after the time where it was send, it will be a giant time lag between two computer but that not will do the aircraft have a bug of position, imagine if a guy use a program who will send the aircraft position information several minutes after the time that his computer generate it, we just will see the aircraft do a fluid flight, but just with a delay between where is the guy in real and where we see him and maybe we flight with a ghost cause the guy have leave the flight since 2 minuts...

 

The problem of the multiplayer in DCS its not the delay between what we see and what other player do, its the position of the aircraft who can make creepy things, the problem don't come from the latency itself, and i never talk about 1 hour latency in my last idea anyways, just MAYBE a little more delay compare to pass directly through the direct server just cause the information will not coming directly from the server but will be relay by another player who are supposed to have a nice connection speed and no problem between him and the server, but that not want mean huge delay, just a little more, but no packets lost due to a problem between us and the server.

 

In fact i want mean, a little latency but no position bug. "and not A little delay, but no lag who finally mean the same thing"

[Bearitall]

In fact i don't have realize that i use a bad word, English are not my first language sorry...

When i talk about "lag" in fact i want mean the corruption of the aircraft position by packet loss who will do it crazy thing like moving 10 m away back to initial position, slide 5 m left, than 3 m back, be inverted and a lot of thing but in fact the player was immobile on the parking and do start up...

 

And yes in fact the lag are not that, but in France we say "lag" for this phenomena, i have mistake the same world using in France and using in English...

 

In fact we can have a delay between two computer without have a crazy position problem, if the information are not corrupt and send at the same debit not matter if the information come 1 minute or 1 hours after the time where it was send, it will be a giant time lag between two computer but that not will do the aircraft have a bug of position, imagine if a guy use a program who will send the aircraft position information several minutes after the time that his computer generate it, we just will see the aircraft do a fluid flight, but just with a delay between where is the guy in real and where we see him and maybe we flight with a ghost cause the guy have leave the flight since 2 minuts...

 

The problem of the multiplayer in DCS its not the delay between what we see and what other player do, its the position of the aircraft who can make creepy things, the problem don't come from the latency itself, and i never talk about 1 hour latency in my last idea anyways, just MAYBE a little more delay compare to pass directly through the direct server just cause the information will not coming directly from the server but will be relay by another player who are supposed to have a nice connection speed and no problem between him and the server, but that not want mean huge delay, just a little more, but no packets lost due to a problem between us and the server.

 

In fact i want mean, a little latency but no position bug. "and not A little delay, but no lag who finally mean the same thing"

 

Hmmm...what I read from your translation of your language into English is you are saying that ED developed a inferrer product...funny I don't have the issue you are having..with a bug as you say it...in the coding...and if so..my question is did you buy the product from ED..or Torrent???? And if it was from Torrent then you got a corrupt copy..and you should not be asking these questions here .. these forums are for paying members...

 

This of course is a speculation on my part until you make it alot more clearer of your copy..since I first using ED products I've never had that issue with it...only in a sense of a player purposely doing the warping by hitting a certain " key "....:music_whistling:

[c0ff]

Don't forget the game also has a track replay feature. It's a little bit different to multiplayer, but it still deals with the basic concept of transferring events that occurred in one instance of the simulation to another instance. It doesn't rely on the unreliable internet, but still routinely fails to play back an accurate record of the events that occurred in the game. So, assuming the netcode is as good as it could reasonably be and that any issues are because of the big bad internet is premature, IMO, when we already have proof that the game can't be relied on to recreate events even when using a lossless transport medium with practically unlimited bandwidth (local HDD).

 

Single-player track replay is completely different from MP game. Historically LO had a feature to switch track to an actual game at any point of track replay. Thus, single player track is a timestamped stream of input events, which relies on deterministic behaviour of the simulator to reproduce correctly. Over the time this system showed to be extremely fragile and hard to maintain. Especially in case of constrained development resources.

Multiplayer tracks are a bit different, they are equivalent to single player tracks for locally simulated entities and are a stream of stored network updates for remote entities.

 

It is possible to make all tracks be a stream of network state updates. We will gain in track robustness and lose in track size and that 'take control' feature.

[c0ff]

In fact i want mean, a little latency but no position bug. "and not A little delay, but no lag who finally mean the same thing"

 

The problem of latency is not that simple as you want it to be. Let's imagine a typical case causing latency complaints: two aircrafts close to each other. In a real world the movement of one object will be detected by another with delay = distance/speed_of_light. As the objects get closer, the delay will reduce, until they touch each other exactly at the same time, just because they are in the same point of space.

Multiplayer game, while pretending to look like a real world is inherently different: subjects are separated by network delays, which do not correlate with simulation at all. And this problem cannot be solved.

 

What makes things harder, in a modern jet sim we are dealing with extremely fast and relatively small objects: airplanes and rockets. Each millisecond of delay translates in 10-100 meters of error = 50-2000% of size of objects.

 

What you can do, is to somehow hide this errors for typical cases. May be, it's time to review that list of typical cases and tune the algorithms where appropriate.

[sorcer3r]

T

What makes things harder, in a modern jet sim we are dealing with extremely fast and relatively small objects: airplanes and rockets. Each millisecond of delay translates in 10-100 meters of error = 50-2000% of size of objects.

 

What you can do, is to somehow hide this errors for typical cases. May be, it's time to review that list of typical cases and tune the algorithms where appropriate.

 

Some tuning there would be nice: http://forums.eagle.ru/showthread.php?t=84037 :helpsmilie:

[Pikey]

Hmm UDP is notoriously flakey on bad networks due to the amount of streaming media using that protocol, but its unavoidable for something as fast as flying.

 

I did a quiet roffle at the peer to peer connections though. Imagine the security risk of clients updating the server where your plane is. Also great for those people wanting to find the IP address of the person that kept shooting them down and griefing their router.

 

Possibilities for madness are endless with that one.