If a tree falls in the forest...

[mewkus]

Along the same lines as the age old question about a tree falling in the forest without anyone there to hear it, If no one is in Thurgadin, do the bar patrons still have their little fight? Does other various dialog still happen?

[Dunthor Warsmith]

heh ... would save processor cycles and other resources, and the electricity needed to run them, to disable it while no one was in zone. That is assuming of course that whatever routines are needed to notice no one is there, shut things down, notice someone is arriving (zoning in?), restart it all, does not wind up on average taking more resources/time/electricity than what was saved while idle.

I would imagine the little guys go through all their routines even when people aren't there, just because it would be far simplier to program and debug, and the gains from the more complex version would be questionable and even if they were positive they would probably be pretty small (and almost definitely insignificantly minute on the overall EQ budget).

[octagon999]

Old one?

If your ranger posts an opinion in Greater Faydark and your sig other is AFK, are you still wrong?

[Aandaie]

yes or no.

The stuff in those areas are automatic. Besides, it is very likely they are set to play on YOUR computer only. That would make more sense.

So the server probably isn't saying "do bar patrons noise." More likely when you are near the bar, the patron noise goes on.
So I'm not sure how they do it.

But they move and do the canned text whether people are there or not.

[Kcalehc]

You could probably test it if you are really interested. Just sit around where it all happens and time the whole thing and the intervals between each event. then wander off and come back a bit later at a time you know an event should be occuring and see if it is or not.

or some variation of that idea will probably work. Slightly boring and utterly pointless but it can be tested.

[QuteCleric]

It's really a question of whether or not the zone exists when someone isn't in it.. the answer is obviously yes since no mob spawns differ in that time span....

Qutie!

[Dunthor Warsmith]

It's really a question of whether or not the zone exists when someone isn't in it.. the answer is obviously yes since no mob spawns differ in that time span....

Could potentially just check respawn timers when the zone was recreating itself, and repopulate the appropriate locations. Making wandering mobs not appear at their spawn point on zone rebuild would be a bit more difficult, but not undoable either.

[Lyron]

A very interesting question, and I don't know the answer. I doubt SOE would give us one either, since it would definitely be a trade secret.

Of course, you do realize that all a zone is, is a list of mobs and PCs, and their attributes, like location, health, mana and buffs. What makes a MMORPG possible is that everything hard is basically done clientside, all the rendering, etc. The world you look at isn't going to lok exactly the same as the eorld someone else looks at , depending on exactly which models are turned on, etc. So it's not that much overhead to keep all the mobs and their positions listed for a zone, but it is some overhead.

However, it can sometimes be a LOT worse to try and recreate a zone, especially if zones are bouncing back and forth from empty to non-empty. (Think of someone bouncing back and forth between north and South Kaladim.) Saving a zone to a file (or in some sort of slower memory), then bringing it back up would be hard. Those disk accesses will kill you in terms of overall speed and resource use. And if you don't store it on disk or in slower, less expensive memory, then you haven't really eliminated much overhead, have you? Only a few processor cycles at best, and I doubt that those are the limiting factor. My guess is that, rather than trying to program this to work properly, if the servers get crowded, they just add a few more boxes to the cluster running each server.

[Dunthor Warsmith]

So it's not that much overhead to keep all the mobs and their positions listed for a zone, but it is some overhead.

The impression I get is that the resource intensive aspects (that would be true for an idle zone) are mob pathing. Its by far the most complex thing going on constantly, trying to figure out how to get from point A to point B on this float grid.

Combat is probably reasonable resource intensive too, as that involves both more calculations, more RNG calls, and a lot more traffic to/from the client.

I don't know if its true for EQ, but one thing I'm continually surprised about in games is the amount of bandwidth and processor time donated to making sure all people *do* see the same graphics. ie if the little sparkle of my fell perfectly on the bag on the ground from my perspective, everyone else sees that same sparkle hit that same point on the ground (as opposed to each client individually randomizing the sparkle patterns). Its nice for sharing "oh wow, that looked cool!" moments, but seems really odd use of limited bandwidth and often limited processor power.

[Saleani]

In my opinion, the hassle factor of saving and restarting a zone is not worth the savings you could get from doing it. Allocating memory is not instant, retrieving stuff from a database is not instant, determining the next move for each of the npcs from stored values is not instant. As a programmer, it seems like spending $1000 to fix a $10 problem.

[Dunthor Warsmith]

Originally posted by Saleani
In my opinion, the hassle factor of saving and restarting a zone is not worth the savings you could get from doing it. <snip> As a programmer, it seems like spending $1000 to fix a $10 problem.

Agreed. But still an interesting theoretical question nonetheless

oops. double posted.

If you go to edit the post, there should be a "Delete Post" section at the very top (above the text box) with a checkbox and a confirmation button, allowing you to delete your own posts.

[Lyron]

Originally posted by Dunthor Warsmith
The impression I get is that the resource intensive aspects (that would be true for an idle zone) are mob pathing. Its by far the most complex thing going on constantly, trying to figure out how to get from point A to point B on this float grid.
power.

Well, a lot of mobs seem to just follow set tracks until molested, and then resume them when the molestation disappears. That's not too terribly memory intensive. It's certainly different than trying to run an algorithm to get from point A to point B (a lot of us did that in 2nd semester comp science, and the algorithms do require a few processor cycles .)

Of course, if pathing is set, then why the heck would mobs ever get stuck in walls (which seems to happen most when noone is there)? So maybe there is some sort of (broken) dynamic system at work. Conversely, if zones DO go down, and mobs are stuck in walls or under the world, then why the heck would you put them BACK like that when recreating the zone?

However, assuming dynamic pathing when no players are in the zone, if you plan to faithfully reproduce the zone, you might need to run the pathing algorithm for thousands of iterations on every mob in the zone. **shudder** And getting all those NPC interactions right would also be difficult. NPC's interact all the time without players present - witness the recent patch message about froglok hunters and foragers getting themselves killed without player intervention. (Of course, those incidental combats WOULD involve more overhead, which would leave to more savings on a shutdown, IF you weren't worried about being exactly faithful.)

So there are arguments on both sides. But IMHO, any way you look at it the savings of resources versus trying to recreate a zone, and making sure you get it right, just don't add up.

[Dunthor Warsmith]

Well, a lot of mobs seem to just follow set tracks until molested, and then resume them when the molestation disappears.<snip>It's certainly different than trying to run an algorithm to get from point A to point B

That's one of the things I'm rather curious about. Do static patrols (wandering mobs) just call on the standard pathing system, or do they each have a unique optimized pre-determined path they follow? Creating a unique path would definitely be more up front work to implement (since utilizing the pathing system is taking advantage of work that needs to be done anyhow), but would probably conserve significant resources at runtime (not having to recompute that path over, and over, and over again).

I'm rather doubtful they do that though, for the very reasons you've pointed out - all the problems with even static npc pathing. Getting stuck on objects, going through the world, and whatnot. Though that could be a sign of a static path gone awry (how I have no clue), and hence why its repeated each respawn ....

If [...] mobs are stuck in walls or under the world, then why the heck would you put them BACK like that when recreating the zone?

Judging by the player code, they haven't figured out a reliable way to figure out if something is under the world or not. If a player falls under the world, sometimes you get caught, get a message something along the lines of "Detected below the world. Moving to safe point, please wait..." but plenty of times you don't, and either take a falling death or get to stay around in nothingness until you get help. I would assume a similar system deals with the npcs, and hence it also doesn't realize that they're under the world.

Figuring out when an NPC is stuck on a wall or object seems like it would be an odd problem to detect. Purely by position, being rammed against an object is not an invalid location. You'd have to take note that the npc is trying to move, and has made zero progress in the last t time. They started to mess with this back in Velious, purposely warping mobs that were excessively far behind aggro'd players (stuck or not). They never seemed to go very far with the system though, and its somewhat unpredictable whether or not things will warp or just get stuck.

Back on the Conquest warder exploit-kill, they mentioned that the warder was 'trying to get to the PCs with aggro but couldn't' so they have some means of detecting npcs that are having trouble getting to their destination. Its unclear if that type of logging was constant or only because of the amount of staff watching the battle (and hence probably also running more detailed logs), and if the detection is only from a human looking over the output, or if the code was aware of it as well. Assuming its an aspect of standard logging and the code is aware, it does seem odd that more is not done to correct "stuck" npcs. Probably either too resource intensive, or fearful that players will complain even more about warping npcs & npcs ignoring zone geometry, or that the above assumptions are wrong.

[eniamn]

erudin would only get about 5 minutes of sun a day. just saying.