Schnitzelnagler wrote:

Most likely the real bottleneck is the network (data-transfer from your server to your clients). Every other resource usage is game/source code specific and can hardly be discussed or even only granted an estimated guess without knowing details.

^ More specifically, upload bandwidth is the most likely weak point.

Schnitzelnagler is correct on all counts.

Byond games aren't as demanding, so i'd say a 2.3ghz intel quad/duo core, as much disc space as u need, and a good UPLOAD and DOWNLOAD speed.

I understood what you meant to express, though, for the 'average' BYOND game running 10-20 players, even a DSL 2000 line would be fine, upstream wise.
While I agree that this is definitely a needle-hole to be considered, the one most people forget (though it is indeed a bigger issue) would be routing.
You can have all the upload on earth, it just takes one lame overused router between you and the client to completely slow down the transfer.

We need to go into computer then, not internet, my cpu maxes out when im hosting some BYOND games, the code has been tweaked about a million times for things to run quicker and much better coding practices put in, but still only around 10 people, and the cpu jumps up there at max while their actually doing anything.

What would be the best way then without changing the code, since that has been done plenty to host the game better.

I don't have much budget and can run 3D games and all well, and even host them, but not when it comes to BYOND.

I hosted also a 2D game that has all the features this game has, but the other game has an insanely bigger map as well and such, yet my BYOND game still lags to hell & not this other 2D game.

I could likely code using something else like C++, C#, VB, or other coding languages if someone thinks those will get significant improvements over BYOND. Obviously coding languages like these take more work, and you would need to know coding, as I do, a beginner wouldn't be able to do it.

Could an advanced coder with those other languages make a significantly faster game than with BYOND, or do you think BYOND might be pretty close to the max speeds you might be able to get with the other languages?

Not true.

Although a lot can be done to optimize games, BYOND is inherently slow. Sometimes you just cannot make it run fast.

In a game I am working on (LONER) the AI in it is very advanced.
Human enemies can react to sight and sound, they can alert friends, can react to dangers such as running from grenades.
They can use and shoot weapons, reload weapons, change weapons, pick up loot if they're running out of ammo, use and throw grenades, use appropriate healing items, they will to some degree even attempt to hide behind cover when in danger or reloading, and will generally try to use terrain to their advantage.

Some of this stuff is more complex than it seems too. Shooting requires they know the properties of their weapon, and how best to use it, they have to check that there is a clear line of sight between them and who they want to shoot, and that they wont hit any friendly allies. (Shooting also uses pixel based projectiles and collisions)
Throwing grenades requires similar, but also a check to make sure the explosion wont hit themself or allies.

I do a lot to optimize the AI, such as dynamically altering how active it is, from inactive, to very active, based on what players will see.
All of the code and functions are optimized as much as I think anyone could possibly optimize them. (As an example of how optimized it is, certain procs that handle the interface and sorting lists are much, much more CPU intensive than any of the AI procs, they are just called less)

Due to this, the game which was originally intended to support as many players as it wants, now will probably have a cap of 10-20. Maybe even less.
CPU usage when nothing is happening is 0%, CPU usage when 50 NPCs are all fighting it out can rocket to 20-50%, with 100 NPCs it can become unplayable.

Due to this the game had to be designed in such a way that it wont be possible for 50-100 NPCs to all be in combat at once. But with enough players, I could not prevent this from becoming a possibility.

This is an extreme example as I doubt most games will use an AI even remotely as complex as this. But it can happen.
You cannot really say if the CPU usage is too high, it is due to poor optimization, maybe it is, but maybe it isn't.
(Oh yeah, the exact same AI, with combat simplified to nothing but using bump to inflict damage can have about 5000 AIs running at once with only 40-50% CPU usage on my computer)

Depending on the type of game, 128Mb of RAM may not be remotely enough.

A 1000x1000 map in BYOND would use all of that 128MB of RAM, or just under it.
Most games wont ever use a map that big, but a decent sized RPG might get close to that, and without some clever loading/unloading of maps (which may or may not be possible in a multiplayer game) you could get close to reaching that 128MB of RAM usage with all other parts of the game taken into consideration.


Also... Lets me modern shall we? It's not 1990 anymore.
Why would I use my 20 year old PC with like a 166MHz processor and 64MB of RAM, when I am sat on a PC with a 3.2GHz i7 (quad core) processor, with 2TB of storage, 8GB of ram and 2GB of VRAM?

Seriously guys, I know which one I'd rather use, even if I'm not going to make full use of it.

Actually, for some reason, my computer doesn't like isometric mode in BYOND very much.
Using the map editor in isometric mode is murder, and games running in isometric mode tend to be slower than those not running in it (the exact same game that is).

I can strangely run full, commercial 3D games with more ease and faster than BYOND in isometric mode.

When you're writing incredibly advanced AI that is a different story.

Many things can be done to optimize performance of games written by people not actively looking to be efficient.

Many simple things such as changing for(var/X) to for(var/X in RelevantList) can be done to optimize your game.

Even advanced programmers will pass up some things they could do to optimize their systems. With amateurs, you can assume that inefficient mistakes will be rampant.