Multiplayer Player Cap

Discussion and feedback on Construct 2

Post » Tue Feb 25, 2014 10:47 pm

Is there a cap for how many players can be in a match together?

What's the largest amount of players that the multiplayer test can handle?
B
22
S
4
G
1
Posts: 153
Reputation: 1,938

Post » Wed Feb 26, 2014 7:02 am

awmace5 wrote:Is there a cap for how many players can be in a match together?

What's the largest amount of players that the multiplayer test can handle?


Are you talking the new multiplayer that is coming to beta soon?
Very Little is known at this stage
or a custom made one with websockets?

Personally would never put more than 6 to 12 people together. Ideal would be 2 to 6 playing together. There is a pokemon demo running somewhere with 100,000 - see it on twitch using normal chat box so anything is possible.
B
65
S
16
G
9
Posts: 1,429
Reputation: 12,718

Post » Wed Feb 26, 2014 8:54 am

I can give you an answer on this, but it's probably not really going to answer it that easily.

There is no theoritical cap to the number of players can be in any given game world, instance or whatever. There is however latency, memory and CPU caps and that's what's going to get you. This will be determined by the following factors

1. How much memory your game can allocate from the system mem
2. Your internet connection quality and bandwidth
3. CPU
4. Your capability at writing logic handling
5. The rate of play(turns are low cost, intense realtime ala Street Fighter are high cost)

From what I gather based on hearing gamers playing on Shards of MMO's. The common on a standard good US broadband internet is likely capable of supporting apx 200 players on medium intensive action(ie World of Warcraft). Where as the average computer on an intenseive game say like the Showdown Effect, probably can't handle more than 8 to 12. While the common FPS can probably support 16-64. All depending on your coding skill.

Now of course you should probably notch all of this down a little for it all running in JS.

Now will there be a hard cap on the plugin?
possible, but that doesn't determine a potential. A lot of online games actually are run across a series of comptuers and your game connection is usually shuffled around internally and a lot of invisible and disconnects and re-connects. Those times you zone and get an image on the screen. That's usually the player getting disconnected and then reconnected to another computer; all managed by a more complicated structure.
B
90
S
18
G
9
Posts: 2,455
Reputation: 15,038

Post » Thu Feb 27, 2014 1:06 am

My question is regarding the multiplayer test Scirra released with the latest blog entry. I saw a screenshot of the test on their Facebook page and it showed 15-20 players. Just wondering if there has been a higher figure.
B
22
S
4
G
1
Posts: 153
Reputation: 1,938

Post » Thu Feb 27, 2014 3:45 am

Yes and it for the most part has been answered. Sorry, but MP doesn't really work as a question of "how many players". It's all relative to the design of the game server architecture.

If you want a hard numerical cap I suppose I could give you some form of answer, but it's not an answer you could effectively use to answer your question for effective game play. but WebRTC on Chrome has a cap of 256 for the listen() function.

Unless you going to do a turn based game. I doubt anyone is going to manage 256 players.
B
90
S
18
G
9
Posts: 2,455
Reputation: 15,038

Post » Thu Feb 27, 2014 4:09 am

In a real world.. Largest action multiplayer action game (FPS) I have played was MAG on PS3 .. it had play modes upto 512 players (two teams 256 players each) on a single map and it worked quite ok. No the perfect though, the games were much better when there were like 32 or at max 64 players on same map. But that was some years ago.. I have no idea what's the limit these days.

Of this C2 beta? I have no idea.. Results depend a lot on what kind of game you are planing to do. There are some real world limitations (bandwidth, data loss, ping times) that can't be avoided, so a lot depnds on server and networks.
Need help? Upload .capx file with your question.
B
16
S
4
G
2
Posts: 233
Reputation: 2,560

Post » Thu Feb 27, 2014 1:00 pm

The bottleneck is almost certainly the upload rate on the host. Our multiplayer plugin imposes no specific limit, but due to connection limits the service will seriously degrade after a certain point (which will depend on the connection).

The problem is the host has to send data for N players, to each of the N players. So if each player needs say 10 bytes of data, then each message will have a size around N * 10, and then that message will have to be sent N times. This creates an N^2 bandwidth requirement. For example:

10 players = 10 * 10 * 10 = 1000 bytes per update
20 players = 10 * 20 * 20 = 4000 bytes per update
30 players = 10 * 30 * 30 = 9000 bytes per update
...
100 players = 100,000 bytes per update

As you can see even though the player count increases linearly, the bandwidth requirement increases proportional to the square. This means even with a significantly more powerful server or a much more compact update, it won't get you many extra players.

The engine sends updates 30 times/sec by default, so the last case takes 3mb/sec up. Note then also that if the server has a 5mb/sec up capacity, then it will take 600 ms to send 3mb of data. This adds latency to every player. The game would likely become unplayable long before it was adding an extra 600ms of latency.

So not only do you need lots of bandwidth for lots of players, to keep it responsive you need to significantly over-provision the bandwidth - say, by 10x.

It's possible to make very large games with spread-out players able to send data only for nearby players (e.g. MMORPGs). This is one workaround which we might look in to with our engine.
Scirra Founder
B
399
S
236
G
89
Posts: 24,519
Reputation: 195,351

Post » Thu Feb 27, 2014 5:45 pm

@Joannak
Well I remember this game. However this was Sony's baby. So
A. Sony can afford powerful computers
B. I only gave a sample that 1 Map/zone is 1 computer. in fact you can divide a map into server nodes. This would require that the player connect to more than the current section server and also connect to the nearby sections. As players transition from one section to another they connect and disconnect. This will give the illusion of persistent connection and stability.
C. Sony can afford mega bandwidth. I'm pretty sure that MAG wasn't a PS3 client hosted game even if the player could create rooms.

So for MAG in theory they could have structure each battle to be composed of 3 computers.
2 computers to handle 1 per side(possible 2 computers a side). These computers would then coordinate there actions on a map server.

There are many many many ways to get around all sorts of numerical limitations. It all comes down to the points I made above.

bandwidth
structure
hardware

in pretty much that order
B
90
S
18
G
9
Posts: 2,455
Reputation: 15,038

Post » Fri Feb 28, 2014 2:08 am

I mentioned MAG since I'd consider it the proven limit achievable with big money. It was technological breakthrough, but for us who were part of the beta tests (from early closed betas) the problems were obvious with those largest maps. latencies, keeping 512 players active and participating,with some coherency and coordination. It was not all about tech issues, but those did not help the gaming.

I know that computer power and network bandwidths have increased a lot since those days, but the network latency are still there and in any action game the Delays set the final limit. This is one of the reasons those big gaming companies have distributed servers on various continents. Usually there are at least 3 sets (Americas, Europe, Asia), and hefty systems.

So if you want to use one server to serve all world, be prepared to big problems.. Even though data pipes are quite wide, the latency will hit hard when clients are overseas.

For local multiplayers (on same lan/wifi) I'd agree on the BW/SW/HW.
Need help? Upload .capx file with your question.
B
16
S
4
G
2
Posts: 233
Reputation: 2,560

Post » Fri Feb 28, 2014 2:29 am

I was following up with just some theoretical designs on that could be achieved; not specifically arguing against you :P
B
90
S
18
G
9
Posts: 2,455
Reputation: 15,038


Return to Construct 2 General

Who is online

Users browsing this forum: Baidu [Spider] and 9 guests