Superhub : Lag Compensation - MW3 PS3/Xbox Gamers

[Daveoc64]

Originally Posted by qasdfdsaq

While I don't play MW3, in MW2 this was not the case. Host was determined by lowest ping, and bandwidth was not a determining factor whatsoever. In fact, this is the case in most games.

I don't see how that can possibly be the case.

Lowest ping to what?

In a peer to peer system with ~16 clients, you cannot measure the ping to some arbitrary destination and decide that the lowest measurement is the best host.

15ms ping to the MW3 server = great, 200ms ping to one of the clients in Australia = useless.

Such a system wouldn't work well as the "best" host could change every time a new player joined and would provide an erratic experience.

Hence most games on Xbox LIVE use two metrics:

Upstream speed
Packet loss

[qasdfdsaq]

Originally Posted by Daveoc64

I don't see how that can possibly be the case.

Well it is.

Lowest ping to what?

To the clients connecting

In a peer to peer system with ~16 clients, you cannot measure the ping to some arbitrary destination and decide that the lowest measurement is the best host.

The client measures the ping to every available host and joins the one with the lowest ping.

15ms ping to the MW3 server = great, 200ms ping to one of the clients in Australia = useless.

This is a good thing as it prevents you joining hosts in Australia.

Such a system wouldn't work well as the "best" host could change every time a new player joined and would provide an erratic experience.

It does exactly that, and in doing so provides close to the best possible experience (in theory, how well it works in practice is debateable)

Hence most games on Xbox LIVE use two metrics:

Upstream speed
Packet loss

If that were the case, we'd all be joining hosts in Korea and Hong Kong with their >1Gbps symmetric connections. And 200ms+ of lag. You've basically contradicted yourself with every other statement.

[Daveoc64]

Originally Posted by qasdfdsaq

The client measures the ping to every available host and joins the one with the lowest ping.

This isn't how matchmaking works in most Xbox LIVE games. Players are matched first and then a host is selected - some people will already be in a "party" together and could be located anywhere in the world (e.g. in Australia and the UK). Many host-based exploits utilise this design to work.

Your suggestion also implies that any console could be a host, despite the possibility that some people won't have enough bandwidth to reliably support a 16 player game, when others in the game would be able to provide a better experience.

Originally Posted by qasdfdsaq

If that were the case, we'd all be joining hosts in Korea with their >1Gbps connections. And 200ms+ of lag. You've basically contradicted yourself with every other statement.

No, because the matchmaking system doesn't work that way. The system prefers people being located closer together first. If there aren't enough people located close by, then it goes further afield. I've played games with people in Japan, Korea and Northern Europe with very good connections.

[qasdfdsaq]

Originally Posted by Daveoc64

This isn't how matchmaking works in most Xbox LIVE games. Players are matched first and then a host is selected - some people will already be in a "party" together and could be located anywhere in the world (e.g. in Australia and the UK). Many host-based exploits utilise this design to work.

Incorrect.

Your suggestion also implies that any console could be a host, despite the possibility that some people won't have enough bandwidth to support a 16 player game, when others in the game would be able to provide a better experience.

Not everyone has enough bandwidth, but it determines that on a yes/no basis and disables them from ever being hosts. Speed makes no difference on your chance of becoming a host, other than disqualifying you to begin with in the same way as having a misconfigured NAT.

No, because the matchmaking system doesn't work that way. The system prefers people being located closer together first. If there aren't enough people located close by, then it goes further afield. I've played games with people in Japan, Korea and Northern Europe with very good connections.

And how exactly does it measure close by?

... By ping.

[Daveoc64]

Originally Posted by qasdfdsaq

Incorrect.

Please provide some evidence to support this, because what you say contradicts what Infinity Ward (Call of Duty), Bungie (Halo) and Epic (Gears of War) have said in the past about host selection.

Originally Posted by qasdfdsaq

Not everyone has enough bandwidth, but it determines that on a yes/no basis and disables them from ever being hosts. Speed makes no difference on your chance of becoming a host, other than disqualifying you to begin with in the same way as having a misconfigured NAT.

This is not correct. Statistics are collected on a continual basis and used to rate each client in the game when determining the host. This ensures that the data used to make a choice is always accurate and not old.

Some games (MW3 being one of them) will change the host mid-game if the initially selected host actually has a poor connection during the game.

Originally Posted by qasdfdsaq

And how exactly does it measure close by?

... By ping.

No, by using the country of the account. This is available to developers and shown in some game lobbies (e.g. Gears of War 3's "players in your region" count for each playlist).

A region is not defined by ping.

[colin25]

I have to favour qasdfdsaq interpretation...having played MW3 et al

It measures by ping, and doesn't change host mid-game, unless the host loses connection or just quits

[qasdfdsaq]

http://lmgtfy.com/?q=modern+warfare+2+host+selection

[Daveoc64]

Originally Posted by colin25

It measures by ping, and doesn't change host mid-game, unless the host loses connection or just quits

To quote a developer on the game:

"We don't want to be constantly switching the host in the middle of the game, but, if necessary, if the game gets so bad at some point because somebody who had a great connection when it started now doesn't, we'll switch it in-game."

Halo Reach does the same thing.


---------- Post added at 19:44 ---------- Previous post was at 19:44 ----------

Originally Posted by qasdfdsaq

http://lmgtfy.com/?q=modern+warfare+2+host+selection

I don't think we're talking about that game.

Nice try with the childish lmgtfy

I'd also note that the search does not produce anything other than pure speculation.

[craigj2k12]

sorry qas, MW2 measures it on download speed alone

[qasdfdsaq]

Lol. I guess that's why I was always host on my 50mb connection with 1.5mb upload, and never on my 45mb connection with 15mb upload!

[lowei]

As a massive call of duty fan myself, i can play on virgins not so good gaming connection, and when connected as host it runs like there is a lot of lag compensation.
But also have infinity, which i use now for the gaming side of my connection,When host it runs like a dream and all who play with me.

[craigj2k12]

Originally Posted by qasdfdsaq

Lol. I guess that's why I was always host on my 50mb connection with 1.5mb upload, and never on my 45mb connection with 15mb upload!

yeah I could never work it out myself, I was on ADSL back then, got changed to fastpath and it made no difference to how often I got picked as host, it made a massive difference when I was host, but it seems to check download speed only from what I could tell running the ps3 through wireshark anyway, when it says "connecting to matchmaking server" thats when it runs the DL speed test, and thats what it bases the host on

[Chrysalis]

dumb system then.

[craigj2k12]

Originally Posted by Chrysalis

dumb system then.

yeah its completely stupid - thats why you end up joining a game and the host cant even cope with a 4v4 match sometimes, I can only guess at the problem, but apart from high pings, most of it looks like not enough upload speed, and packet loss