Oct 8 2015, 12:21 pm
In response to Zasif
|
|
Kamuna
 |
The webclient has the potential to be faster and has more portability, among other things. The fact that it is currently in a browser says little to nothing about the software (opinion). You can easily make an executable of a browser game. Having a game in a browser is a design choice, usually for easy portability and reaching to a greater audience. You'll be surprised on how many people will not play a game if you have to download and install it, (disregarding one-click download and install solutions) and in some cases, they won't even play if it is not using a one-click download, install, AND play solution. Installing and playing a BYOND game for the first time is even more complex than the average installation (not counting BYONDexe), and people (even "real gamers") can get easily confused in the process. The webclient is a great way to change that without having to give everyone BYONDexe.
|
I have a post of it running "100 players" (101, actually), all in combat. I've optimized it a bit more so it's more like 150-180 players now. Next time I get my hands on that laptop (Wednesday, probably), I'll find out a better approximate number and screenie it. My own PC can run 90-100 players just fine with a cheap AMD processor.
I'm running all the player logic I can on those NPCs (plus their own AI), so any CPU difference in connected players vs NPCs is negligible, and it's even possible that the NPCs use more CPU than players would. Le post: http://www.byond.com/ forum/?post=1911651&page=48#comment16863178 | |
Sorry to be the bearer of bad news, but the difference between a connected player and a simulated player is actually pretty significant.
Every frame, the server looks through the view, screen, and images list of every connected client. It iterates through every object in the view, screen list, and images list every frame to check for appearance changes. I'm not trying to be a downer here, but NPCs pretending to be players are just right out not the same thing. I'm curious how the two test cases would compare. | |
Yes, but I was lead to believe that the CPU usage for each connected player is very minimal. In any case, it should be much faster than calling view() and such (it's doing it on the lower level, right?). Would it make a big difference, you think? Even if the CPU usage were to double, that's still 75-90 players per instance on a regular modern laptop with the optimized code. That's not bad at all, especially considering the methods I could use to hold more players.
| |
Yes.
Monologue: I wouldn't be using the software if I didn't have some sort of community inside of it. Well, that and the fact that I enjoy making server-sided code more than client-sided code, and that client-sided logic when making online games is much easier to get sidetracked on, especially since you have so much power. BYOND isn't that bad if you use it right. That's also a reason why I'm using BYOND, I want to prove that. I can always convert the game from DML to C# or another C/C-like language after the game is finished if I really need to. Another reason I am using BYOND is that it is so quick to use for prototyping, and this game only needs to hold 16 players per instance max. So yeah, make the game first, know my scope, worry about problems in my scope. Everything else is just extra, if my scope suddenly expands, I don't mind re-creating the game in another language. | |