I want to HOST large games, 10 player FFA (more than 10 if private servers) on large single worlds(1000 - 1300 diameter), or large world with a single small moon; but also control the lag in the game. I have done some forums and internet research but have not found an all-encompassing source for this type of information and most of the information is out dated. I have also made some of my own observations on hosting many games described above. Below are some of my assumptions and questions. I have tried 1300 earth type planet with three 300 small moons with no metal. Goal was to create a regular RTS map feel but still enjoy some of the benefits of tossing around planets. The 1300 planet will create lag. In 10 player games I have had a few here and there than ran smooth for me, but most the time mid to late game, approximately 30-40 minute in, the game drops to a crawl. It seems people with lower than 12 gigs ram complain faster than I see problems with 12 gigs of ram. Some people complain profusely about lag while I barely see any. Is this something they see on their side that I do not… or should we both be seeing the same thing? I noticed games with players spread father away from each other lag the worst. E.g. Australian player, German player, and several US players in the same game. Since the last patch, I have created a 10 player 1000 tropical planet with no moons. I added snow caps to reduce trees. I experimented with lakes to create an environment with no water spawns yet still have enough water to have ships, control territory, and create sand (which also reduces trees). In these games I have seen 6 out of 10 games run smoothly, with slight lag almost as if the game was designed to play at that speed… meaning unless you play many small scale 1v1 you would not really notice the lag. The other 4 out of 10 games starts to really slow down about the 35 minute mark. Often times in these games, if I am still around, I have noticed that deleting a plethora of land mines, unused bots, or simply eliminating a large player increases the game's performance almost immediately. I know, duh to most the above right? I have tried adding cracks and mountains in strategic areas to create some natural border but at the same time elimination structures on the rest of the map to help with map size and pathing problems. It appears to me that cracks cause the most issue with pathing, often times loosing units inside the crack. I have tried different biomes. I find that 1000 moons do lag less than earth/tropical planets but at the same time, even though they are the same size they appear smaller in perception? They are also more difficult, at least for me, to gain bearings without land marks. I prefer the green(white and red are bright for me) planet with some small oceans and polar caps to help out the perspective. Is there any way to have green planet but simply thin out the trees? Assuming trees actually do create more lag? Would renting a private server for hosting be more fruitful, in the lag department, than using a standard PUBLIC server from UBER? Short questions: Does patrol create more lag? Does pathing issues create lag, any specific things to try to avoid? Is one BIOME actually better than another when thinking about lag? Is one large planet at 1000, more or less lag than 2 planets at 500 each? Is there anything related to the mass, placement from sun, or other setup parameter, when dealing with one large planet, than can affect the lag? Is there any way to tell someone’s ping in lobby to predict whether they may introduce lag for a game? If playing on a large planet, is there any analysis on 4, 5, 6, through 10 players and how much lag will be introduced? Do lowering or raising graphic settings for one player have any effect on lag for other people? E.g if I play uber can any slowdown I see affect the other player’s game experience? And visa a versa. If I am recording a game to post on YouTube, can that affect other’s experience with lag? Thanks for any input! B8factor
Yes, every time a group of units has to do pathfinding it puts more stress on the server. The more units moving around on their own the worse it will be. Every time units are bumping into each other or something else it creates more lag. Yes, biome makes a big difference. The moon biome causes the least lag, the earth biome the most. What matters is all the stuff on the planet, like trees and such. It is the surface area that matters (I belive) and one 1000 planet has a much bigger surface area than two 500 planets. Not that I know of. Because of the client/server architechture PA uses the performance other people experience has no effect on the lag you experience. What matters is how fast the server is, how fast your internet is and how fast your computer is. The number of players don't matter (at least not with only 10 players). What matters is the number of units total. One players graphics settings has no effect on any other players performance. No, only your own.
I have played many games over the past months, always 10 player FFA with the same system. I have noticed that games with some regulals from Russia do indeed lag more than games with all US regulars?
So I am wondering what would be ok stats for a p to host a game? I have an i5 and 16 gig ram, would that outperform the uberservers??
The main requirement for hosting is lots of bandwidth. Uber have a cap of 2mbit per player (upload) so you need 20mbit up to host a large 10 player game successfully... That's the main issue preventing private servers atm.
Some games that go faster and other not is I think because a server hosts multiple games. If one server has a couple of games which are all getting pretty big then the server might slow down already. Would renting a private server for hosting be more fruitful, in the lag department, than using a standard PUBLIC server from UBER? You've got three types of lag where as the third one isn't actually lag but people will call it still lag. People who complain about low framerates or that it stutters a lot is all about the system. Not your concerns really. If people complain early game (I think first 15 minutes) then its probably their system hogging up. You can also have units glitching from one spot to another. This means there is a lot of network lag. ( client doesn't gets the data in time so the units stay at spot A instead of moving to B and then suddenly it gets another data package and in that data package the units is already at spot C so the client correctly updates that the units is at C right now. In the end game the servers time dilation kicks in which makes everything goooooo reeeeaaawwwly slooooooow. the more units the slower it gets which is something that happens in your hosted games eventually. Renting a private server could increase the performance lategame however its expensive. I know that @cola_colin did a lot of research with hosting a private server. The thread https://forums.uberent.com/threads/setting-up-a-community-dedicated-monster-of-a-server.65080/page-4 In the end the conclusion was that it wasn't viable to get the performance they needed for the money they got. However the goals was to host 40 player or at least 20 player playable games. Hosting a smoother 10 player FFA could be possible but you'll need a couple of criteria. Server needs a processer which has a really good single thread performance ( overclocked haswell). Server needs multiple cores ( one to run the actually simulation and others for example sending and recieving network data. A server with good bandwith speeds. Not sure how much. BTW I really do enjoy those FFA games of yours. Thank you for rekking me last game .
Thanks for the info! It may be something I try soon… It really depends on the cost of the servers and if I can find a server that fits the PA criteria. Hehe, you were NOT easy to REKK, that is for sure! I was shocked how many units you had. If it had not been for the mines I had lay between us you would have cleaned me out for sure. I meant to go back and watch from your perspective…. I watched some of your units shoot some mines but then an inch in they just started running over them. I was wondering if your radar coverage was at its max range? You had 3 or 4 times my unit count at that time. I had some luck on my side… having played that map enough times I can tell you your spawn was one of the worst, against mine which was one of the best. You are butted up on 2 sides behind you and I had nobody for ages behind me. Plus I have easily expandable mexes where you have to fight for yours. I think it is time to make a new system similar but different. This one has worked out well, IMO, but now needs some improvements. When I made it I had a 3 major points in mind: 1) No water spawns 2) Even mex at start with even mex in-between starts 3) Make it difficult to turtle to win, must conquest and control more map. That game lasted quite a bit longer with Prime and I bashing each other’s economy for another 30 minutes and then the Blue player just up and WON. Prime is a EXECELENT RTS and FFA player. He had me but then up and quit…. Looking at econ during the replay I cut him down to size but he still had the advantage… I am not sure why he quit, I do not speak Russian. Maybe the porgies were done?
My internet is good, but not good enough to host form HOME. I have 50 down 5 up. By the looks of it I would suffer on the up? If a server is rented I suppose the server location would need to have at least 10 up to host 10 player, more for more?
Thanks alpha link to cola's initiative made it clear no regular hosting PC will fix the time dilation late game. A pity pa is slow late game just like supcom. Only now the server hosted by uber is the bottleneck in stead of some player. The dream of truly massive pa is still a while away it seems.
I suppose there are many perspectives as to what people want to achieve… LARGE player counts does sound fun, but for me, and the games I enjoy playing (FFA), 10 players is about perfect. Most my 10 player games last between 30 minutes and an hour which seems to be a good time limit for me. Any longer and the average attention span on playing, staying to watch the rest of the game, or UTUBER watching a recorded match, dwindles fast. Sure there are die hards, count me in, that want more MORE M O R E, but in reality sometimes more means less interest. I am reminded of a quote I really liked, “overspecialize, and you breed in weakness. It's slow death” *I will talk more on this in another thread. Even in 10 player games I can only keep up with 3-4 other players when scouting and calculating my next step. I have to assume the other side of the world is taking care of itself and the winner of that 3-4 skirmish will eventually be dealing with the winner of my 3-4 skirmish. So from my perspective, I want to be able to host 10-15 player games on 1-2 large worlds well for as smooth as a game that can be played for about an hour’s duration.
OOOh that explains a lot hahaha. I actually thought your sniper bots were doing all the work at my army but it was those mines. You we're lucky then I guess as I completely destroyed my eco trying to get t2 up. So my radars probably had no power and thus tanks can't shoot those mines anymore. But yeah if purple, brown or black went all in on me I'd definately be screwed (thank god they didn't) but thats always the case with an FFA. Prime probably had to leave or wasn't interested no more as he also complained about the time dilation? I think he could have killed you more easily with a nice sniper bot + AA army teleported near your base instead of using units cannons (I don't think slammers are that good) but I guess thats just my personal luv for the Gil-E bots. Anyways yeah he is an excellent player . I thought he'd won it haha XD . Was a really good game though as expected . If you're going to reedit the map then I'd recommend using the custom spawn placement now. Should make a big difference in gameplay. Just let me know if you need any help with that. I wouldn't worry about it too much. There has been some responses from Uber where they acknowledged that there still is lots of performance improvements that can be done. It just takes resources and time to get there. The vision of Uber still is 40 player games. And I'm positive they will try to get there eventually. Supcom is already a fully developed game whereas the engine+pa is right now only 2 years old! Just think about that.