Hello Jorb & Loftar, what can I do for H&H?

[theTrav]

Is it a recent patch? Loftar changed the rendering of avatars not so long ago to eliminate the long delay when people move in and out of party radius

[Spiff]

No, this was like several months ago.

[Jackard]

this was before loftar released the stuff for making clients. reverse-engineered or something, not quite sure

[Delamore]

The only thing our guy could figure was that it might be related to a bug that removed the highlighting of objects behind other objects in our client.
The client apparently went from 100% usage to 40% on some peoples systems.
What ever it was it vanished the next time it got updated.

[Axehilt]

Maybe you can still help with performance optimization though - it would be really cool if the lag can be reduced or eliminated entirely.

I would not be surprised if the current lag was caused by a bug of some kind rather than improper optimization.

I'd lean strongly towards the current lag being a result of improper pathfinding optimization. I'm sure there's some sort of performance profiling that could be done to help track it down.

The evidence for pathfinding being unoptimized seems pretty strong. I mean every single time I get a mob to run into a wall (a tough pathfinding situation) the game lags.

Plan on recommending the game to all my old game design work friends once perf/lag is solved.

[sabinati]

Maybe you can still help with performance optimization though - it would be really cool if the lag can be reduced or eliminated entirely.

I would not be surprised if the current lag was caused by a bug of some kind rather than improper optimization.

I'd lean strongly towards the current lag being a result of improper pathfinding optimization. I'm sure there's some sort of performance profiling that could be done to help track it down.

The evidence for pathfinding being unoptimized seems pretty strong. I mean every single time I get a mob to run into a wall (a tough pathfinding situation) the game lags.

Plan on recommending the game to all my old game design work friends once perf/lag is solved.

i notice the same thing when they run into water

[Leonon]

The evidence for pathfinding being unoptimized seems pretty strong. I mean every single time I get a mob to run into a wall (a tough pathfinding situation) the game lags.

Pathfinding should be server side lag though, not client side.

I think that a large part of the resources the client uses are because they seem to be pushing Java's graphical capabilities using the "MORE POWER!" method. If someone were to make a client without shading (day and night cycle) and with a better way of layering graphics (hive behind a hive behind a hive...) it would likely end up being a fairly light program.

[loftar]

Thanks for the offer Phazorx, but as has been said elsewhere, having more people work "with us" would probably be more work than it'd be worth. Jorb and I work well together because we now each other and can meet regularly IRL and what not. It would be a bit weird to share the server code with someone we don't even know IRL and so; you can probably understand that.

The client obviously needs a lot of work, though, so if you're up for that, there's a ton of things that need to be done. :) First and foremostly, I'd really like it for someone who actually knows OpenGL to look through the drawing code and tell me why it's so slow. When I've been profiling it, it seems almost all CPU time is spent inside OpenGL, but I can see no reason why it shouldn't be faster. That task didn't exactly seem to match the areas of expertise that you listed, though. :P

I think that a large part of the resources the client uses are because they seem to be pushing Java's graphical capabilities using the "MORE POWER!" method. If someone were to make a client without shading (day and night cycle) and with a better way of layering graphics (hive behind a hive behind a hive...) it would likely end up being a fairly light program.

I cannot help but wonder if you're actually basing that on something or if it's pure conjecture. :) The day/night shading takes almost no time whatsoever on any of the computers that I've tested the client on; and neither does layered graphics. From what I've been able to tell from my profiling of the client, the cycle time is spent mostly in 1) actually drawing tiles/sprites to the screen (and as mentioned above, I cannot really understand why it takes so long) and 2) somewhere outside the drawing of the actual game view; that is, somewhere in the HUD or something, and I haven't gotten so far as to discover the exact location.

As for the server-side lag, it consists of a couple of issues:
* Right now, there's too little RAM in the server while it's trying to do too much (as mentioned in the "Lag" thread and elsewhere, it is also running Apache and a couple of other things).
* As a consequence of the lack of RAM, there's a constant trickle of harddrive I/O, which causes loading and saving of maps to slow down a lot, causing cycle overruns.
* I am indeed not using the pathfinding algorithms very well right now. Many of the greatest lag-spikes are due to animals trying to find a path across an entire system of rivers, so I need to limit it and make the animals realize that they should give up.

Apart from those three main issues, there are a couple of lesser things as well, but they matter not a fraction as much.

[Axehilt]

i notice the same thing when they run into water

Well yeah...by "walls" I mean any obstructing object. Rivers are obviously part of that.

server needs more RAM

Time to break out the item shop and sell non-combat/gameplay pets! (I'd buy one just to support the game )

[kajiura]

Is it for certain that client is so slow because of some wrong opengl code, and not because of java? Also, afaik, there are people with stable high fps regardless of number of objects on the screen, maybe they have something common, or maybe they just have too fast machines...