|
Author
|
Topic: clan war sunday feb 22nd 2004
|
Irascible Member
|
posted February 24, 2004 07:20 PM
I'll take your word for it Matrix. I've never studied the subject of netrate closely. And to be clear, that is NOT what I'm busting 666's chops over. So if you wish to clarify, then tell me this. Wouldn't the server's setting's override the client's? In other words, wouldn't the netrate be set at which ever one (client or server) had the lowest setting?IP: |
KONG Member
|
posted February 24, 2004 08:01 PM
good job Coax. you hit it on the nose. I dont believe there is a way to turn that off. ------------------ "My style? You could call it the art of fighting without fighting.'' Homepage
IP: |
CyberCannon Member
|
posted February 24, 2004 08:40 PM
OK, my two cents worth. I could care less whether or not a weapon makes a scorch mark or not, this is nothing more than "eye candy", the deformation brought on by things like the machine gun also don't do much, at least not on their own. I have more than once though gotten myself stuck with the drill, which I have to admit is part of the hazard in the game, now that has only happened when I started out in a hole to begin with. I would be careful about what weapons have their deformation taken out, but the scorch I could care less about.Dealing with the netrate, I just play to the best of my ability, at a netrate of 3K I believe, and just deal with the situation as it comes. Sometimes dealing with it is by leaving, because it is "unplayable", by my standard at least. Now I suspect that my idea of unplayable may not be the same as others, since I have heard people complaining and not seen anything different than normal. So what ya'll do with the netrate I will just play as I can. ------------------ An Irish Blessing May the road rise to meet you. May the wind be always at your back. May the sun shine warm upon your face. The rains fall soft upon your fields and, Until we meet again. May God hold you in the palm of His hand. CC IP: |
coaxs Member
|
posted February 24, 2004 08:44 PM
The issue really isnt netrate. Netrate is the fine tuning you do after you address the main problem. The main problem is simply having to update all the new scorches and craters. And the greaest source of them are the tanks themselves. Since they leave tread marks and craters by simply driving!------------------ Tread Marks Ladders
IP: |
Irascible Member
|
posted February 24, 2004 09:18 PM
It was pointed out to me that clan maps increase weapon selection. More nukes and cruise missles will fly because there are more opportunity for them to spawn, and people NEVER pass them up. So you will get more of the large scale deformation that Killer spoke of. AND since driving itself creates traffic, the increased driving that has been a fully acknowledged side effect of clan maps also adds to the bandwidth problem.Of course, I don't propose any changes to map selection as we've previously agreed to. But the irony is just too thick to not point out. IP: |
666 Member
|
posted February 24, 2004 11:03 PM
Yo...Dudymas...where are you...you need to put a humorous spin on all of this. IP: |
Irascible Member
|
posted February 25, 2004 02:01 AM
fyi I'm going to let this one go and see where the chips fall. Ultimately the best way to measure this is to see real world results. Whether we make changes now or not, sooner or later there will be a clan war with 20 players on clan maps. If there's lag, we can say that the map didn't help. Of course, the funny part is that that's already happned, but... meh. If there is reduced lag on some future game with 20 players on a clan map, you can only claim that the maps did it fairly IF we don't implement any other tweaks. If we do implement other tweaks, we'll have to wait and see how things perform with the tweaks with 20 players on a stock map. None of this is to say I endorse 666's attempt to ameliorate his hatred of spawn fragging by forcing his maps up my colon, but what the heck. What happens, happens. IP: |
BamZipPow Member
|
posted February 25, 2004 02:06 AM
Don't ferget one other factor...That computer/server also hosts other servers fer the other LDA games...  So if you really want to slam the TM server, get a bunch of other people to play the other LDA games online at the same time...  IP: |
KiLlEr Member
|
posted February 25, 2004 02:41 AM
K, one last thing on netrate.I'm quoting this from Dreamer's old post: the client netrate is the maximum the client is willing to handle. the server netrate is the maximum rate the server communicates with each client at. -> If the client netrate is lower than the server netrate, the client netrate takes precedence. If the server netrate is lower than the client netrate, the server netrate takes precedence. If the sum of all precedent netrates used in the communication between server and clients hits the limit the servers network connection can handle, problems arise. Example: Yesterday we had between 25 and 30 participants. If each of those had their netrate at 10k (kBytes per second), the server would have had to prepare, pack and send up to 300kB per second - equivalent to a 2.5Mbit line, not counting the computing power needed to calculate that much data on the fly. ..... That means that either the server netrate has to be dropped for huge matches such as those we had last night (WMan said it was set to 10k, if I remember right), OR everyone has to lower their client netrate until both the server AND the line can handle the bandwidth. Another problem is that some people tend to keep their netrate at 10k, even though their internet connection can't handle 10kBytes per second (equiv 80kBit per Second - more than ISDN or dialup can handle!) Those people always cause and experience lag on populated servers. Yet another problem is that even if your line should be able to handle the bandwidth, it may in fact not be at the very moment you need it. Internet traffic on exactly those nodes your packets are routed through can cause slowdowns. ---------- 4k is a little jumpy, but pretty safe unless you're using a 28k modem. (32 users at 4k need a bandwidth of just over one Megabit per second (a good connection can handle that) and 32 is the max number of players allowed by TM, afaik.)
IP: |
=DNX= Matrix Member
|
posted February 25, 2004 05:21 AM
quote: It was pointed out to me that clan maps increase weapon selection. More nukes and cruise missles will fly because there are more opportunity for them to spawn, and people NEVER pass them up. So you will get more of the large scale deformation that Killer spoke of. AND since driving itself creates traffic, the increased driving that has been a fully acknowledged side effect of clan maps also adds to the bandwidth problem.
This leads to an altogether worse problem..... Paranor drudging across the plain muttering things like : %£"*(*&3 Ramdrills..... Dirtballs..... Ramdrills..... Dirtballs..... lol ... IP: |
666 Member
|
posted February 25, 2004 10:34 AM
Ras and I had an enlightening talk last night. And he is correct, I do know something about how different maps behave, but I have not yet effective communicated those experiences. What I did for Sailor was to test over a 100 maps, with 40 tanks, 4 teams, to 99 frag count multiple times at the godly setting to weed out the maps that did not work. My finding are that some maps played better then others, due to the amount of action on the map, terrain of the map, and location of the spawn points. Therefore my experience is that the map determine the playability of the game. The current thought that is being discussed that the quantity of the tanks determines the amount of lag. If that was true then every map that I tested should have lagged out with 40 tanks leaving 40 treadmarks and 40 tanks rutting the map all at once. That did not happen, some maps were laggy and some played well. The maps that did not lag out were the maps with the action that was slowed down and with less rutting. These maps had terrain that would slow the action down and spawn points not in the middle of the battlefields. I do not buy the argument that the quantity of the tanks determines the lag of the play. I think that it is a good guess, but not the complete answer. We all have had the mutual lag experience when someone is joining late to a game. The updating of the crater has been agreed upon as the cause, but that does not explain why there is lag in the game when someone is not joining. Then there is my experience in ranked events with maps with few spawn points. The action and map deformation would increase within a concentrated area of the map and lag would start to enter the picture. Then there is my experience with too many plasma mines on the map. I use to run a one-weapon server for a while and the plasma mine version would lag out. One would even start to teleport on the server with only 6 tanks and all those mines on the map. Then there is harmless and my experience with the Xmas fun ddm on the clan wind chill map. I love the Xmas tree and elf mesh decorating the maps. The clan wind chill map was the only map of the other 2 maps that would lag out. And the lagging was predominate when you had the vista view of the windchill map with all the different meshes on it. We both thought that this was very odd since there were no more than 6 tanks on the map and no mods causing lots of graphics. I think the answer came at the LDA mod forum that would make sense of all these lag experiences on TM. TB1tiger asked about the poly count in a frame. There is a limit of the poly count for any given frame generated by the TM engine. Max out that limited and you start to have lag. This would make sense of all the above experiences that I have had with the map testing, the plasma mine, and Xmas ddm. It was not the amount of tanks, but the number of events generating the polys. More poly events within ones frame such as fast action and lots of map rutting, or lots of bouncing plasma mines, or lots of meshes in a vista contribute to a player own experience of lag. Therefore I do not think that the lag experience by some is experienced by all at the large gatherings outside of someone joining late. There is a simple test to verify if this is true. I will set up the one weapon server again, have the AI’s launch a bunch of plasma mine, then I will go hop over to mirror land. If my guess is correct, once in mirror land my lag should drop off when I am no longer viewing all those mines. ------------------ OH, Community Modded pack is out…is this going to be stock? Race Ladder Results & Downloads [This message has been edited by 666 (edited February 25, 2004).]
[This message has been edited by 666 (edited February 25, 2004).] IP: |
=DNX= Matrix Member
|
posted February 25, 2004 12:53 PM
Which is why i play with 500m or 750m view distance.... im used to 250 so its no big deal for me. I play in 16bit and 800x600 with 256 texture with 7000 tris - i never see ANY lag on ANY map - i do make a point of putting all this in the read me's of my map releases, a damn shame no bugger ever reads em.[This message has been edited by =DNX= Matrix (edited February 25, 2004).] IP: |
coaxs Member
|
posted February 25, 2004 01:29 PM
666: Then the solution is for the end user to lower their settings to 250meters not installing the clan maps.I post twice now, so people actually read my posts. ------------------ Tread Marks Ladders
IP: |
coaxs Member
|
posted February 25, 2004 01:30 PM
666: Then the solution is for the end user to lower their settings to 250meters not installing the clan maps.I post twice now, so people actually read my posts. IP: |
Irascible Member
|
posted February 25, 2004 02:11 PM
Yes, but what should the view distance be? And how can we be sure? IP: |
coaxs Member
|
posted February 25, 2004 02:18 PM
Well that is up to the user to decide. If he lags at 1000m;. Reduce to 750m, if he lags at 750m reduce to 500m and so on.------------------ Tread Marks Ladders
IP: |
=DNX= Matrix Member
|
posted February 25, 2004 02:34 PM
There are 4 fields of view, 250m, 500m, 750m, 1000m - the LOD (Level of Detail) is affected by the distance settings and tris.example with 1000m and 5000 tris: Range 0 to 250m - Most Important: Contains 40% of the tris This is where most detail is needed in front of the users eyes. Range 250 to 500 contains 30% 500 to 750 contains 20% 750 to 1000 contains 10% (least important) when you go down to say 500m viewing distance, 0 to 250 in front of you contains 60% and 250 to 500 contains 40% So why do i use % - because if u change to 2000 tris and 1000m, you either need a flat map or the ability to drive on a wobbling landscape. The tris setting tells tm to assign that many poly tris over the desired viewing distance. The closer the ittem or land mass is to the eye, the more polys its assigned in order to draw it properly. The texture is then layered on top. Killer, im not doing too well explaining this - some assistance plz ? IP: |
Irascible Member
|
posted February 25, 2004 02:36 PM
LOL, I see why you routinely give me smart @ss contradictory remarks Coax. It's great fun, isn't it? Here's the thing 666. First, as I've stated I am not debating you on map selection anymore. I don't need to because you won't be able to make changes without convincing people. And with respect, the above is a start - but it will never do the trick. You and Coax have both had training in the scientific method. You know that what you've listed above is only the first step: observation. You must then collate the observations and hypothesize an explanation for them. You then test the hypothesis to prove or disprove the theory. It's like you've skipped the theory part of it and are testing to prove or disprove the observations. In that case, the fix is in. No one can disprove your personal observations! If there is a logical way by which a clan map reduces lag, then you should be able to present a single concise theory as to why. If it's based on reason, we don't need your personal testing experience to understand it. I don't need a formal mathematical proof to understand that 2 + 2 = 4. But at the moment, you've not given me the 2 + 2 part of it. You're just assuring me that it is 4. 4 is the result (or, the observation of less lag). What formula got you there? Running the test yet again to see the result yet again does nothing to explain how the result was created. Incidentally, the view distance thing is quite the interesting suggestion that I've never heard before. If TM doesn't have to tell me what it's seeing in the distance, it needs less bandwidth. Excellent suggestion Matrix. Sorry Coax, I was unimpressed with your shower revelation because it's of no practical use. Though thanks for the imagery. :^p IP: |
666 Member
|
posted February 25, 2004 03:06 PM
Guess in a sense you could choose any map if you use the minimum setting. Therefore maps that have high action and rutting could be used. Or folks could choose clan type maps that spread out the action in smaller battlefields thereby allowing the higher settings. I think folks just need to tune their setting according to the maps that the clan picks. Ras’ ideals of reducing the scorch and rutting of various weapons will be a benefit for all and should be implemented. Reducing the weight of the plasma mine should be implemented. Setting the netrate to 4K would also benefit and and should be implemented. If there were others suggestions that I have missed, please remind me.
[This message has been edited by 666 (edited February 25, 2004).] IP: |
KiLlEr Member
|
posted February 25, 2004 04:53 PM
k, now this is getting silly. You are talking apples and oranges here.The view distance, deformed terrain, poly count setting, lots of sprites (i.e. smoke, explosions), resolution, color depth, detailed triangles, and trilinear filtering, contribute to video lag. I have a Radeon 9800Pro and I suffer zero video lag at the highest settings. On my other machine I have a Geforce II MX, If I don't play in low quality mode, I can't play at all (it gets 5 frames per second). This lag only affects the one player and does NOT cause lag for anyone else. Then you have lag caused by computing the actual deformations. If you have a slow machine, you will lag. But this does not affect other players online who have faster machines. Lets call this local lag. Then you have lag caused by data flying back and forth between the clients and the server, This is network lag. Any time an entiy is created or destroyed, data is sent out. Players move around, data gets sent out. Damages gets inflicted, data is set out. Some one gets a frag, data is sent out. If you have lots of projectiles flying around (or standing still) data is sent out. Just because something is causing lag, it seems that you guys assume its because its network lag (bandwidth). WRONG WRONG WRONG WRONG You need to find out which of the three types of lag it is so you can figure out what to do about it. Sometimes it can cause 2 types of lag, but you have to see which is type is cause you more of a problem. For example: Lots of projectiles cause network lag whether they move or not. Lots of meshes and sprites (i.e. triangle count and blending) cause video lag. You need to try lowering your visual quality (resoulution, color depth, view distance, max triangles) to see if it helps. If it does then your video card is to blame. If it doesn't help, then its network lag. Craters and scorching cause all three types of lag. Its computationally expensive (it needs to build and rebuild sections of terrain), it causes video lag by increasing the poly count, it causes network lag when someone joins midgame and there is alot of data that needs to be sent out to reproduce the terrain on the client machine and if there is alot of terrain deformation action, you can cause 'updating craters' to show you while your playing. To reduce your overall lag, this is a prime candidate. But it would helps to see if its your machine thats the culprit and not the server. Play a single player game, set to 40 mins with a map that has lots of terrain features (deer hunter for example) and when it starts hit the 'F1' key to bring up the FPS or frames per second. While playing, keep an eye on your fps counter. If that number starts to drop as the game progresses and gets below 18-20 FPS, then try setting the view distance to a lower value. If it clears up, then its your video card. If it does not, its either you machine is too slow, or you video is really old and in need for an upgrade. You have to know what type of lag it is your experianceing. Just saying 'ooo it's soo laggy' then messing with the server settings because you automatically assume its network lag is just plain stupid. Its like bleeding in the olden days as an attempt to cure every possible ailment, meanwhile cause more illness and thats more of a simily than you think. Messing with ent files is a sure way of increaseing network lag, since the king of all of lag is network lag caused by entity synching. You literally cause the server to stand still while its has to send out hundreds of ent files to every person that joins the game. IP: |
kv Member
|
posted February 25, 2004 05:17 PM
are you guys sure you aren't talking about 2 different kinds of lag?There is visual lag (stuttering of the screen) and network lag (the warping effect). AFAIK, a TM server should not be creating visual lag. All the server is doing is relaying tank, weapon, and crater information, and it does this for all items in the game, no matter if you can see them or not. The client takes this info and draws what the server tells it you should be seeing. Example: Clients A, B, C, and D are playing on Server. Client A tells Server "i'm Here, this is My_weapon, this is my Direction_of_travel at Speed. What do you know?" Server tells A: "B, C, and D are Here with these Weapons, doing This. This is the current Terrain_update." Client A then determines if B, C, and D are within visual range, and if they are, draws them. Server doesn't care what anyone is seeing on their screen, only where they are and what they're doing. If you're getting visual lag, you are trying to draw more than what you're system can handle, thus, a client problem. But it shouldn't effect the server, I think (the client should continue sending positional data to the server while the client continues to recieve data and tries to catch up). In this case, the client should stop moving until the stuttering stops and the client catches up with the server. If stuttering persists, the client should lower their video settings. If you see a phantom, or warping, tank, then I think that would be evidence of the server not being able to feed you data quickly enough (a netrate problem?), or it could also be caused by a client that isn't able to upload data quickly enough to the server (client netrate or bandwidth?). I run my client at full settings and I don't experience visual lag, but I do see phantom tanks. From what I understand about how game servers work, my full visual settings aren't causing the phantoms, the server (or lagging client) is. To ruin one arguement, the server is always sending out crater data, so there is no differece between a big hole and a little hole, just the number of them that the server has to update when someone joins mid game. It's up to the client to draw what it sees, the server just tells it that Weapon hit Here and made Impact. IP: |
kv Member
|
posted February 25, 2004 05:19 PM
figures, someone posted the same thing at the same time.oh well. IP: |
coaxs Member
|
posted February 25, 2004 05:31 PM
We are tackling all lag everywhere.------------------ Tread Marks Ladders
IP: |
mELON cRACKER Member
|
posted February 25, 2004 07:27 PM
My contribution to this:I am not going to get in the middle of all this right now except to say this: 10K, 4K, etc.... these are not complete terms. For instance, 56K modems run at 56Kb, or 56 Kilo--bit. However, many other things such as RAM are measured in Kilo--byte. As you know, there are 8 bits in a normal byte, and there are 10 in an analog trasmitted byte, 8 plus the start and stop bit. So 56Kb = 5.6KB. I have always assumed that the TM settings were in Kb. Hope this helps a little. Thanks, [BWL]mELONcRACKER
IP: |
KiLlEr Member
|
posted February 25, 2004 08:34 PM
TM's Netrate is in KiloBytes not kilobits. Its a misnomer in TM's case as communication has always been calculated in bits as the telecommunication is serial in nature and not parrallel. Games count the amount of bytes they need to send before actually sending the data out. This is a very common tactic used in video game. The more modern games use bi-directional protocols and manage netrate dynamically.IP: | |