Bandwidth Usage for online lobbies in Gran Turismo 5 - with tests

  • Thread starter steiner
  • 23 comments
  • 2,897 views
171
Germany
Saarland
Steiner84
edit:
i ran some tests as fixed host, results in post 11 and post 16

Lobbies without fixed host VS Lobby with fixed host, tested as a single participant not host in post 24

________________________________________

Why this thread you ask? Here's why:

The purpose of this is, to see what connection you need to host a full lobby because fixed host lobbies seem to be more stable. you can read about it here
https://www.gtplanet.net/forum/showthread.php?t=168334
Thats why I initialy started this thread.

In essence it sais that:
a) in lobbies without a fixed host (normal open lobbies and the coffee cup lobby) the network type used by GT is a mesh network. Like a p2p network. Every player sends data to every other player.
That is why you so often see people in lobbies that cannot see or hear each other, you see people not beeing able to join the lobby due to incompatibilites between networks or people getting dropped mid race. that sucks.

b) in an open lobbie where you tick the box "fixed host" GT uses a Star Type network and the player who opens the lobbie acts as the central game host for every player.

So in case a) where every player needs to establish and maintain multiple connections to each other participating player in the lobby where some players have nat type 2, some nat type 3, some exposed host, some here some there some everywhere - basicaly networking is all over the place and a total mess. Hence there are so many problems with visibility and stability of the race. One player that lags can cause significant loss of race quality for every other player.

In case b) every player only needs to maintain one single connection to the hosting player. Like you connect to a central server and all player information and voice data you get sent from this single server. Once you established the connection dropping out or haveing problems is significantly less likely to happen as long as your and the hosts connection is not oversaturated.

Therefore these tests.

over the last two days I hosted two fixed lobbies for combined 7 hours. in this time arround 30-35 people joined and played in the lobby, a core part of arround 10 people for 3-4 hours minimum and I didnt see a single drop or any other trouble like people not able to see each other or the like.


Also i think for people with a slow internet connection it is better to connect to fixed host lobbies because you get a single constant stream of data that should be less than in a p2p lobbie. This is more likely the case the more players are in the lobby.


_______________________________

Original Topic:
Hello


I searched the web but couldnt find any useful information regarding bandwith usage in GT5 online lobbys.

Say I host a 16 players online lobby with race and voice quality set to very high, how much upstream should I have? How much on high or normal? How much does voice chat add? And so on.
Is there any official information from PD on that matter?


i might add: what exactly is the difference between the race qualitay settings? whats different when set to very low or very high?

regards


edit: I've read a bit more and found theres a severe difference between fixed host and non fixed host, but failed to make out how I can set up a private lobby for my friends but with fixed host. Fixed host does only seem to be an option for open lobbys??
But there surely has to be a way to be the fixed host for a group of friends?
 
Last edited:
When ever I make a lobby my room settings are always...
Race Quality- High or Very High
Chat Quality- Low.

Reason for this is that the cars will look smoother with a higher race quality, while voice chat will have less lag with it set lower than the race quality.

If you have the race quality and chat quality set at high, you increase the load on your connection causing mic and room lag period.
 
Chat quality to high doesn't really make a significant enough difference to sacrifice room latency over anyway in my opinion.

Unfortunately I can't help you with bandwidth figures, I run unlimited so I've never cared to look into how to measure such things.
 
Also when you scroll thru the community click on your own name and click on the coffee cup and create your own lounge. From what I've heard that's as close to a private lobby as your gonna get, also fixed host is convenient if you do not want someone to host the room when you leave. Note fixed host will sometimes, randomly drop people.
 
Lock2Lock: Thanks, ive allready managed to get that voice quality should not be higher than standard.
I figured a higher quality setting would cause more lag for the voice, but its new to me that it affects the whole room latency.

Also thanks to UrieHusky for the input on this.


UrieHusky: I also run on an unlimited connection. The thing is i would like to host for a group of friends because I have a 70/5Mbit connection but wanted to know if thats enough to host a 16 player lobby on certain quality settings.

ryzno: the problem with the coffee cup is that you cant set your own lobby (the coffe cup lobby) to fixed host.
 
Lock2Lock: Thanks, ive allready managed to get that voice quality should not be higher than standard.
I figured a higher quality setting would cause more lag for the voice, but its new to me that it affects the whole room latency.

Also thanks to UrieHusky for the input on this.


UrieHusky: I also run on an unlimited connection. The thing is i would like to host for a group of friends because I have a 70/5Mbit connection but wanted to know if thats enough to host a 16 player lobby on certain quality settings.

ryzno: the problem with the coffee cup is that you cant set your own lobby (the coffe cup lobby) to fixed host.
That's more than enough.

Don't set chat quality too high, because not everybody has such a good connection as you. And it affects everybody, sometimes massively.
Race quality 4/5 is enough, no need for 5/5. 3/5 is ok too, but I wouldn't use 1 or 2.
It's not all about the hosts connection, the clients can suffer from quality settings being too high, even though the host is fine.
 
That's more than enough.

Beat me to it :lol:


Yep, that's plenty 👍 just keep in mind that GT5 uses P2P (I think?) so if one user is lagging it tends to stuff up the entire room.
 
ok thanks again.

As far as I've learned GT only uses p2p if you play in a private lobby or an open lobby without fixed host.
If you set fixed host all players connect to that host and get their gameupdate from him. That should increase the bandwith for the host quite a bit. Thats why im asking.
 
Beat me to it :lol:


Yep, that's plenty 👍 just keep in mind that GT5 uses P2P (I think?) so if one user is lagging it tends to stuff up the entire room.
In this aspect you're clearly from the wrong country. :lol:


GT5 without fixed host is P2P/mesh, with fixed host a star network.
Most people don't like fixed host rooms, because it kicks everybody as soon as the host leaves or disconnects.
 
In this aspect you're clearly from the wrong country. :lol:


GT5 without fixed host is P2P/mesh, with fixed host a star network.
Most people don't like fixed host rooms, because it kicks everybody as soon as the host leaves or disconnects.

Ahh yep, I can see where that confusion stems from then - I can only think of one instance of being in a fixed host room.
 
fixed host has the advantage if one has high bandwith while others struggle.

So I've made some tests :D


host with this connection
2793985966.png


Timetable of Events:


18:00 I created an open Lobby, fixed host, race quality very high, voice quality normal
18:09 me + 1 guest
18:14 me + 2 guests
18:27 me + 3 guests
18:31 me + 4 guests
18:36 me + 5 guests
18:45 me + 4 guests
18:59 me + 5 guests
19:01 me + 6 guests
free roaming up until now
19:03 Start of a race on circuit de la sarthe; me + 6 guests
19:08 end of race

a graph where x is time and y is bandwidth used, up and down
chart2tcjsp.png

the big spike at the end is the race

a screenshot of my routers online monitor at the time of the race and after (the drop in upsgtream indicates the end of the race
racefb2uk3g.jpg


as you see, download is not even shown, its too less
upload ranges in the 500 - 700 kbit region

detailed table of bandwidth usage:
Code:
Datum Zeit 	Traffic out (Volumen)	Traffic out (Geschwindigkeit)	Traffic in (Volumen)	Traffic in (Geschwindigkeit)	Ausfallzeit	Abdeckung
  1 bis 121 von 121  
24.06.2013 19:13:10	620 KByte	170 kbit/Sek.	563 KByte	155 kbit/Sek.	0 %	100 %
24.06.2013 19:12:40	603 KByte	164 kbit/Sek.	315 KByte	86 kbit/Sek.	0 %	100 %
24.06.2013 19:12:10	625 KByte	171 kbit/Sek.	665 KByte	181 kbit/Sek.	0 %	100 %
24.06.2013 19:11:40	581 KByte	159 kbit/Sek.	634 KByte	173 kbit/Sek.	0 %	100 %
24.06.2013 19:11:10	259 KByte	71 kbit/Sek.	327 KByte	89 kbit/Sek.	0 %	100 %
24.06.2013 19:10:40	334 KByte	92 kbit/Sek.	545 KByte	149 kbit/Sek.	0 %	100 %
24.06.2013 19:10:10	65 KByte	18 kbit/Sek.	67 KByte	18 kbit/Sek.	0 %	100 %
24.06.2013 19:09:40	70 KByte	19 kbit/Sek.	130 KByte	35 kbit/Sek.	0 %	100 %
24.06.2013 19:09:10	79 KByte	22 kbit/Sek.	76 KByte	21 kbit/Sek.	0 %	100 %
24.06.2013 19:08:40	487 KByte	133 kbit/Sek.	296 KByte	81 kbit/Sek.	0 %	100 %
24.06.2013 19:08:10	1.678 KByte	458 kbit/Sek.	456 KByte	125 kbit/Sek.	0 %	100 %
24.06.2013 19:07:40	1.656 KByte	452 kbit/Sek.	464 KByte	127 kbit/Sek.	0 %	100 %
24.06.2013 19:07:10	1.665 KByte	455 kbit/Sek.	476 KByte	130 kbit/Sek.	0 %	100 %
24.06.2013 19:06:40	1.691 KByte	463 kbit/Sek.	481 KByte	132 kbit/Sek.	0 %	100 %
24.06.2013 19:06:10	1.706 KByte	466 kbit/Sek.	508 KByte	139 kbit/Sek.	0 %	100 %
24.06.2013 19:05:40	1.842 KByte	503 kbit/Sek.	548 KByte	150 kbit/Sek.	0 %	100 %
24.06.2013 19:05:10	1.830 KByte	500 kbit/Sek.	545 KByte	149 kbit/Sek.	0 %	100 %
24.06.2013 19:04:40	2.304 KByte	628 kbit/Sek.	570 KByte	155 kbit/Sek.	0 %	100 %
24.06.2013 19:04:10	2.367 KByte	647 kbit/Sek.	598 KByte	163 kbit/Sek.	0 %	100 %
24.06.2013 19:03:40	1.078 KByte	294 kbit/Sek.	559 KByte	153 kbit/Sek.	0 %	100 %
24.06.2013 19:03:10	1.638 KByte	447 kbit/Sek.	560 KByte	153 kbit/Sek.	0 %	100 %
24.06.2013 19:02:40	1.547 KByte	423 kbit/Sek.	580 KByte	159 kbit/Sek.	0 %	100 %
24.06.2013 19:02:10	1.223 KByte	334 kbit/Sek.	378 KByte	103 kbit/Sek.	0 %	100 %
24.06.2013 19:01:40	1.147 KByte	313 kbit/Sek.	357 KByte	98 kbit/Sek.	0 %	100 %
24.06.2013 19:01:10	838 KByte	229 kbit/Sek.	255 KByte	70 kbit/Sek.	0 %	100 %
24.06.2013 19:00:40	869 KByte	237 kbit/Sek.	234 KByte	64 kbit/Sek.	0 %	100 %
24.06.2013 19:00:10	1.032 KByte	282 kbit/Sek.	269 KByte	73 kbit/Sek.	0 %	100 %
24.06.2013 18:59:40	864 KByte	236 kbit/Sek.	333 KByte	91 kbit/Sek.	0 %	100 %
24.06.2013 18:59:10	854 KByte	233 kbit/Sek.	345 KByte	94 kbit/Sek.	0 %	100 %
24.06.2013 18:58:40	849 KByte	231 kbit/Sek.	447 KByte	122 kbit/Sek.	0 %	100 %
24.06.2013 18:58:10	821 KByte	225 kbit/Sek.	324 KByte	89 kbit/Sek.	0 %	100 %
24.06.2013 18:57:40	837 KByte	228 kbit/Sek.	524 KByte	143 kbit/Sek.	0 %	100 %
24.06.2013 18:57:10	832 KByte	228 kbit/Sek.	338 KByte	93 kbit/Sek.	0 %	100 %
24.06.2013 18:56:40	762 KByte	208 kbit/Sek.	319 KByte	87 kbit/Sek.	0 %	100 %
24.06.2013 18:56:10	684 KByte	186 kbit/Sek.	272 KByte	74 kbit/Sek.	0 %	100 %
24.06.2013 18:55:40	699 KByte	191 kbit/Sek.	254 KByte	70 kbit/Sek.	0 %	100 %
24.06.2013 18:55:10	514 KByte	141 kbit/Sek.	165 KByte	45 kbit/Sek.	0 %	100 %
24.06.2013 18:54:40	539 KByte	147 kbit/Sek.	198 KByte	54 kbit/Sek.	0 %	100 %
24.06.2013 18:54:10	712 KByte	195 kbit/Sek.	268 KByte	73 kbit/Sek.	0 %	100 %
24.06.2013 18:53:40	804 KByte	220 kbit/Sek.	413 KByte	113 kbit/Sek.	0 %	100 %
24.06.2013 18:53:10	548 KByte	149 kbit/Sek.	240 KByte	65 kbit/Sek.	0 %	100 %
24.06.2013 18:52:40	474 KByte	129 kbit/Sek.	230 KByte	63 kbit/Sek.	0 %	100 %
24.06.2013 18:52:10	499 KByte	136 kbit/Sek.	244 KByte	67 kbit/Sek.	0 %	100 %
24.06.2013 18:51:40	580 KByte	158 kbit/Sek.	252 KByte	69 kbit/Sek.	0 %	100 %
24.06.2013 18:51:10	707 KByte	193 kbit/Sek.	251 KByte	69 kbit/Sek.	0 %	100 %
24.06.2013 18:50:40	653 KByte	178 kbit/Sek.	240 KByte	65 kbit/Sek.	0 %	100 %
24.06.2013 18:50:10	494 KByte	135 kbit/Sek.	244 KByte	67 kbit/Sek.	0 %	100 %
24.06.2013 18:49:40	475 KByte	129 kbit/Sek.	235 KByte	64 kbit/Sek.	0 %	100 %
24.06.2013 18:49:10	359 KByte	98 kbit/Sek.	144 KByte	39 kbit/Sek.	0 %	100 %
24.06.2013 18:48:40	476 KByte	130 kbit/Sek.	261 KByte	71 kbit/Sek.	0 %	100 %
24.06.2013 18:48:10	831 KByte	227 kbit/Sek.	274 KByte	75 kbit/Sek.	0 %	100 %
24.06.2013 18:47:40	545 KByte	149 kbit/Sek.	181 KByte	49 kbit/Sek.	0 %	100 %
24.06.2013 18:47:10	388 KByte	106 kbit/Sek.	120 KByte	33 kbit/Sek.	0 %	100 %
24.06.2013 18:46:40	441 KByte	120 kbit/Sek.	131 KByte	36 kbit/Sek.	0 %	100 %
24.06.2013 18:46:10	819 KByte	224 kbit/Sek.	224 KByte	61 kbit/Sek.	0 %	100 %
24.06.2013 18:45:40	1.033 KByte	282 kbit/Sek.	333 KByte	91 kbit/Sek.	0 %	100 %
24.06.2013 18:45:10	1.013 KByte	277 kbit/Sek.	280 KByte	76 kbit/Sek.	0 %	100 %
24.06.2013 18:44:40	1.309 KByte	357 kbit/Sek.	302 KByte	82 kbit/Sek.	0 %	100 %
24.06.2013 18:44:10	1.186 KByte	324 kbit/Sek.	288 KByte	79 kbit/Sek.	0 %	100 %
24.06.2013 18:43:40	1.010 KByte	276 kbit/Sek.	294 KByte	80 kbit/Sek.	0 %	100 %
24.06.2013 18:43:10	1.023 KByte	279 kbit/Sek.	414 KByte	113 kbit/Sek.	0 %	100 %
24.06.2013 18:42:40	1.242 KByte	339 kbit/Sek.	672 KByte	183 kbit/Sek.	0 %	100 %
24.06.2013 18:42:10	1.273 KByte	348 kbit/Sek.	909 KByte	248 kbit/Sek.	0 %	100 %
24.06.2013 18:41:40	971 KByte	265 kbit/Sek.	268 KByte	73 kbit/Sek.	0 %	100 %
24.06.2013 18:41:10	855 KByte	233 kbit/Sek.	236 KByte	64 kbit/Sek.	0 %	100 %
24.06.2013 18:40:40	992 KByte	271 kbit/Sek.	277 KByte	76 kbit/Sek.	0 %	100 %
24.06.2013 18:40:10	1.211 KByte	331 kbit/Sek.	359 KByte	98 kbit/Sek.	0 %	100 %
24.06.2013 18:39:40	1.181 KByte	323 kbit/Sek.	314 KByte	86 kbit/Sek.	0 %	100 %
24.06.2013 18:39:10	1.009 KByte	276 kbit/Sek.	281 KByte	77 kbit/Sek.	0 %	100 %
24.06.2013 18:38:40	976 KByte	266 kbit/Sek.	275 KByte	75 kbit/Sek.	0 %	100 %
24.06.2013 18:38:10	805 KByte	220 kbit/Sek.	337 KByte	92 kbit/Sek.	0 %	100 %
24.06.2013 18:37:40	825 KByte	225 kbit/Sek.	196 KByte	53 kbit/Sek.	0 %	100 %
24.06.2013 18:37:10	903 KByte	247 kbit/Sek.	237 KByte	65 kbit/Sek.	0 %	100 %
24.06.2013 18:36:40	837 KByte	229 kbit/Sek.	330 KByte	90 kbit/Sek.	0 %	100 %
24.06.2013 18:36:10	761 KByte	208 kbit/Sek.	303 KByte	83 kbit/Sek.	0 %	100 %
24.06.2013 18:35:40	734 KByte	200 kbit/Sek.	277 KByte	76 kbit/Sek.	0 %	100 %
24.06.2013 18:35:10	643 KByte	176 kbit/Sek.	217 KByte	59 kbit/Sek.	0 %	100 %
24.06.2013 18:34:40	624 KByte	170 kbit/Sek.	181 KByte	50 kbit/Sek.	0 %	100 %
24.06.2013 18:34:10	613 KByte	167 kbit/Sek.	183 KByte	50 kbit/Sek.	0 %	100 %
24.06.2013 18:33:40	597 KByte	163 kbit/Sek.	186 KByte	51 kbit/Sek.	0 %	100 %
24.06.2013 18:33:10	454 KByte	124 kbit/Sek.	298 KByte	81 kbit/Sek.	0 %	100 %
24.06.2013 18:32:40	467 KByte	127 kbit/Sek.	180 KByte	49 kbit/Sek.	0 %	100 %
24.06.2013 18:32:10	325 KByte	89 kbit/Sek.	161 KByte	44 kbit/Sek.	0 %	100 %
24.06.2013 18:31:40	287 KByte	79 kbit/Sek.	187 KByte	51 kbit/Sek.	0 %	100 %
24.06.2013 18:31:10	225 KByte	61 kbit/Sek.	138 KByte	38 kbit/Sek.	0 %	100 %
24.06.2013 18:30:40	226 KByte	62 kbit/Sek.	133 KByte	36 kbit/Sek.	0 %	100 %
24.06.2013 18:30:10	176 KByte	48 kbit/Sek.	120 KByte	33 kbit/Sek.	0 %	100 %
24.06.2013 18:29:40	75 KByte	21 kbit/Sek.	42 KByte	12 kbit/Sek.	0 %	100 %
24.06.2013 18:29:10	152 KByte	42 kbit/Sek.	95 KByte	26 kbit/Sek.	0 %	100 %
24.06.2013 18:28:40	232 KByte	63 kbit/Sek.	82 KByte	22 kbit/Sek.	0 %	100 %
24.06.2013 18:28:10	296 KByte	81 kbit/Sek.	189 KByte	52 kbit/Sek.	0 %	100 %
24.06.2013 18:27:40	336 KByte	91 kbit/Sek.	121 KByte	33 kbit/Sek.	0 %	100 %
24.06.2013 18:27:10	246 KByte	67 kbit/Sek.	137 KByte	37 kbit/Sek.	0 %	100 %
24.06.2013 18:26:40	217 KByte	59 kbit/Sek.	110 KByte	30 kbit/Sek.	0 %	100 %
24.06.2013 18:26:10	215 KByte	59 kbit/Sek.	113 KByte	31 kbit/Sek.	0 %	100 %
24.06.2013 18:25:40	220 KByte	60 kbit/Sek.	113 KByte	31 kbit/Sek.	0 %	100 %
24.06.2013 18:25:10	252 KByte	69 kbit/Sek.	146 KByte	40 kbit/Sek.	0 %	100 %
24.06.2013 18:24:40	256 KByte	70 kbit/Sek.	176 KByte	48 kbit/Sek.	0 %	100 %
24.06.2013 18:24:10	255 KByte	70 kbit/Sek.	209 KByte	57 kbit/Sek.	0 %	100 %
24.06.2013 18:23:40	255 KByte	70 kbit/Sek.	146 KByte	40 kbit/Sek.	0 %	100 %
24.06.2013 18:23:10	521 KByte	142 kbit/Sek.	858 KByte	234 kbit/Sek.	0 %	100 %
24.06.2013 18:22:40	336 KByte	92 kbit/Sek.	704 KByte	192 kbit/Sek.	0 %	100 %
24.06.2013 18:22:10	284 KByte	78 kbit/Sek.	335 KByte	92 kbit/Sek.	0 %	100 %
24.06.2013 18:21:40	262 KByte	72 kbit/Sek.	388 KByte	106 kbit/Sek.	0 %	100 %
24.06.2013 18:21:10	81 KByte	22 kbit/Sek.	80 KByte	22 kbit/Sek.	0 %	100 %
24.06.2013 18:20:40	109 KByte	30 kbit/Sek.	81 KByte	22 kbit/Sek.	0 %	100 %
24.06.2013 18:20:10	153 KByte	42 kbit/Sek.	154 KByte	42 kbit/Sek.	0 %	100 %
24.06.2013 18:19:40	45 KByte	12 kbit/Sek.	38 KByte	10 kbit/Sek.	0 %	100 %
24.06.2013 18:19:10	58 KByte	16 kbit/Sek.	136 KByte	37 kbit/Sek.	0 %	100 %
24.06.2013 18:18:40	184 KByte	50 kbit/Sek.	93 KByte	25 kbit/Sek.	0 %	100 %
24.06.2013 18:18:10	187 KByte	51 kbit/Sek.	197 KByte	54 kbit/Sek.	0 %	100 %
24.06.2013 18:17:40	165 KByte	45 kbit/Sek.	66 KByte	18 kbit/Sek.	0 %	100 %
24.06.2013 18:17:10	180 KByte	49 kbit/Sek.	71 KByte	19 kbit/Sek.	0 %	100 %
24.06.2013 18:16:40	196 KByte	54 kbit/Sek.	63 KByte	17 kbit/Sek.	0 %	100 %
24.06.2013 18:16:10	170 KByte	46 kbit/Sek.	67 KByte	18 kbit/Sek.	0 %	100 %
24.06.2013 18:15:40	225 KByte	61 kbit/Sek.	79 KByte	22 kbit/Sek.	0 %	100 %
24.06.2013 18:15:10	171 KByte	47 kbit/Sek.	89 KByte	24 kbit/Sek.	0 %	100 %
24.06.2013 18:14:40	87 KByte	24 kbit/Sek.	58 KByte	16 kbit/Sek.	0 %	100 %
24.06.2013 18:14:10	85 KByte	23 kbit/Sek.	65 KByte	18 kbit/Sek.	0 %	100 %
24.06.2013 18:13:40	113 KByte	31 kbit/Sek.	151 KByte	41 kbit/Sek.	0 %	100 %
24.06.2013 18:13:10	99 KByte	27 kbit/Sek.	196 KByte	54 kbit/Sek.	0 %	100 %



conclusion is arround 100kbit upstream per user on a 7 player race, fixed host, race quality very high.
That could be higher as more players join because the host has to send all information to every player - more players -> more information per user.
But i didnt test that. thats just what i think.

im not sure about p2p lobbies but i guess at some point of numbers of players a fixed host has the advantage for participating users with a slow connection because on p2p you recieve varying data from multiple peers whereas you have a single constant stream of data on a fixed host scenario which should not exceed 250kbit, 300 max.


this was unfortunately a non voice scenario, as noone spoke a word during the test, but its a start i guess.


we also made some latency tests from participating players to my router
it varied from 41 to 60ms depending on the clients connection
my own ping is 22ms according to speedtest.net


here are all files ive collected
http://www.speedshare.org/download.php?id=0C5AA3DC1

feel free to discuss
 
Last edited:
That's interesting data, thanks for sharing, steiner.

I think there has been efforts to document GT5's bandwidth usage in the past, but I don't have the links handy at the moment. It always surprises me just how little bandwidth is actually used. Latency, of course, remains as the most significant metric.
 
yeah I have desperatly searched the web for information but couldnt find something useful.

Latency is always key, sure, but what me interessted was if I can host a 16 player fixed host lobby with certain quality settings.
Unfortunately we have never been more then 7 players, but seeing it was a spontaneous test its at least a start. I will monitor bandwidth in the next few days and hopefully can gather more data with more players and voice chat.
I will post here when I have something to share.
 
ok, more data from today

fixed host open lobby
race quality very high
voice quality normal
races have been 2 and 3 laps on circuite de la sarthe, no time change, no weather.

the last bigger block is always the race mentioned. the huge short drop ist the start of the race. immediatly after that the most bandwidth is needed. During race its a bit less.



9 players, 3 with headset
chart9p3hsvuzpp.png




10 players, 3 with headset, addidtional 3 players in lobby watching
chart10p3hs3ld5snk.png



8 players, 3 with headset
8p3hsgeszh.png



6 players, 1 headset
6mann1voicejgsu5.png




again 6 players, 1 headset, with a 7th joining lobby midrace
6mannag2jz46.png



the more players, the more bandwidth needed for every single player. headset adds to this further.
extrapolating from this i think a 3Mbit Upstream connection should be sufficient to host a 16 player race with a bunch of people using headsets.
maybe 4Mbit for everyone to use headset.

I hope i can record data for a full lobby sometime soon, will post the results if it happens.
 
when i do a speed test there is a massive difference between my PS3 speed test to my PC (PS3 is running at about 40% of PC speed test)
So with all this extra loading we see from the graphs about I can see how we can get drop outs or lag issues if you are only using 40% of the bandwidth :(
 
FYI, I have a 12M down /1M up line (which is pretty quick for Australia!)

but PSN see's 4 to 5M down / .8M up
 
velly: its the same for me, at least in the downstream. ps3 shows merely 20mbit.
For upstream i have to look agian.


GUMP07: thats the whole purpose of this, to see what connection you need to host a full lobby because fixed host lobbies seem to be more stable. you can read about it here
https://www.gtplanet.net/forum/showthread.php?t=168334
Thats why I initialy started this thread.

In essence it sais that:
a) in lobbies without a fixed host (normal open lobbies and the coffee cup lobby) the network type used by GT is a mesh network. Like a p2p network. Every player sends data to every other player.
That is why you so often see people in lobbies that cannot see or hear each other, you see people not beeing able to join the lobby due to incompatibilites between networks or people getting dropped mid race. that sucks.

b) in an open lobbie where you tick the box "fixed host" GT uses a Star Type network and the player who opens the lobbie acts as the central game host for every player.

So where in case a every player needs to establish and maintain multiple connections to each other participating player in the lobby where some players have nat type 2, some nat type 3, some exposed host, some here some there some everywhere - basicaly networking is all over the place and a total mess. Hence there are so many problems with visibility and stability of the race. One player that lags can cause significant loss of race quality for every other player.

In case b every player only needs to maintain one single connection to the hosting player. Like you connect to a central server and all player information and voice data you get sent from this single server. Once you established the connection dropping out or haveing problems is significantly less likely to happen as long as your and the hosts connection is not oversaturated.

Therefore these tests.

over the last two days I hosted a fixed lobby for combined 7 hours. in this time arround 30-35 people joined and played in the lobby, a core part of arround 10 people for 4 hours minimum andc I didnt see a single drop or any other trouble like people not able to see each other or the like.


Also i think for people with a slow internet connection it is better to connect to fixed host lobbies because you get a single constant stream of data that should be less than in a p2p lobbie. This is more likely the case the more players are in the lobby.
 
steiner, I would really love to see tests with fixed host vs. non-fixed host lobbies with the same number of guests. I only run non-fixed lobbies because I think it is selfish to explode the place when I leave.

I would also love to see with mic vs. no mic.
 
Crunch Houston:
I will try do a non fixed scenario when I get the chance.
mic vs. no mic can be a bit tricky to handle but I will see what I can do.

My intension for doing lobbies with fixed host is primarly aimed at events. We are a group of people that do certain events like "2 hour lemans race" that are time fixed. For those scenarios a fixed lobby is a good choice because the event is over when its over and the lobby can be closed.
For just for fun races or open lobbies with random people non fixed is ok I guess.

illustratedtrick: thanks. I will collect more data over the next weeks and post accordingly.
 
Lobbies without fixed host VS Lobby with fixed host
I have not been the host this time, goal for this test was to find out the bandwidth usage for one individual participant in a lobby

Lobby Time is marked


no fixed host
16 players
no headsets
nonfixed16p0hsg1sgq.jpg


fixed host
16 players
no headsets
fixed16p0hstos8e.jpg



as assumed, non fixed lobbies requier more bandwidth for individual players with some nasty spikes whereas in fixed lobbies the bandwidth is lower and much more constant.
This should help players with a bad connections.


next i will try to do non fixed vs fixed as host.
but my guess is that the non fixed lobby result will not differ much wethere your "host" or not as it should be roughly the same outcome for every player, host or not.
but it could take some time to get 2 full lobbies.
ill post when i have something.
i guess ill name them "Dirty drivers wellcome" as that seems to atract alot of people in a short time :ill:
 
Last edited:
Back