Curse

My Lipgloss is Cool · Mar 30, 2008, 03:06 AM // 03:06

I think if the build made a difference, the difference would have been much more noticable then what it was.

Devonas big Sister · Mar 30, 2008, 01:19 PM // 13:19

Quote:

Originally Posted by Antithesis

I think you'll find the gwdata file contains only graphics and physics of the engine. This saves us downloading the weighty data like textures when entering an instance.

It doesn't contain loot or AI info, that'd be loaded from the server when entering the instance.

Like mentioned before, if its not in de gwdata, then where is it? in your RAM is ofcourse a big nono. Anyways, the formula IS written down SOMEWHERE.

Another thing crossed my mind: that formula cannot have an endless cycle.
So technically spoken, you can calculate when the forumla refreshes an thus calculate when you get the same drops again.

So...
If you keep farming and farming and you get a drop worth a million ingame gold, let us call that Drop-X.
Write down the area, the foe, and the exact time you entered the area where you got Drop-X.
Wait for the formula refresh, and enter the area again at the exact same time, and yet again, Drop-X is obtained.

Again really farfetched, but whenever I see topics like this my braincells go berserk with ideas theories etc.

Chthon · Mar 30, 2008, 08:04 PM // 20:04

Quote:

Originally Posted by Devonas big Sister

It may be farfetched, but i come up with another interesting theory:

The meaning of "bad spawn" could change drasticly with this. Since you guys are really into science

you should look for this to even consider my theory plausible:

Check the way your spawn at the start is build up and compare it with the quality of the drops. Check Spawn location, patrol directions and professions in the spawn. Maybe you should also check the server IP/ID in the Ping-Dot.

One extra thing is crossing my mind. I know you guys are out there: the GW BigFile Data readers. There is a very very small chance that it is possible to read and decode a formula that determines the drops/droprate/quality.

If we consider the above theory to be true and we can get our hands on that formula (legally), then consider what it could do for the wealth ingame.
It would make exclusive skins available for anyone without really dropping the prices.

Time to pass the baton...

1. All that's happening here is that the server is using the last X digits of its clocktime at entry as the seed for the pseudo-random number function. So every time the game "rolls the dice" in two instances with the same seed, they come up the same. There's no secret "formula" to discover.

2. If I recall right (and I may well not) GW is written in C/C++. So, if we assume that they're using plain old srand(), and not using a custom complier with a funny unsigned integer size, there would be 65535 different seeds, so there would be 65535 different ways your farm could turn out (not counting whatever effect is causing the small variations like the one Nechtan Thaumaturge reported. Loot scale at work I guess...). Now, you could make an index for every worthwhile farming zone that would contain the first 5 or so drops for all 65535 seeds and whether that seed will eventually produce anything worthwhile, but it would require a positively herculean effort. I don't think the benefit you'd gain by skipping bad seeds would ever pay off the effort it would take to build such an index in the first place.

ccruzp · Mar 31, 2008, 12:35 AM // 00:35

Quote:

Originally Posted by Devonas big Sister

Like mentioned before, if its not in de gwdata, then where is it? in your RAM is ofcourse a big nono. Anyways, the formula IS written down SOMEWHERE.

Serve side? no one in their right mind would make a mmorpg where your computer decides how to handle this kind of stuff, you would be able to edit you files and make ectos drop at each pre-searing grawl kill.
Take diablo 2 for exemple, when playing Open character (ie, the server is your computer) w/ memory scan/modifiers you could do stupid things to the game variables and run around godmode, make every monster have 100% chance of droping uniques and coutless other things. But when you played a closed character(ie bnet is the server) there wasnt much you could do messing around w/ the files in your computer, you could MapHack wich basicaly reads te map info that blizard sends you each time you move into different acts and read the memory of the game to show the location of nearby monster(the ones the server already sent you).

To the testers, arent minotaurs in Anvil Rock easy to sync? they are really close to the spawn point, about 1,5 aggro bubbles away from the iron horse mines spawn point, i just dont remember if their positioning varies so you guys can check if its "synced", anyway there are quite a few of then (10-12), you can Fire aoe then all to death(since some people dont consider SoJ as mass killing) if you body block then or run NM and use a SV/sliver armor build to kill then 1 by 1. Here's the spot, http://www.youtube.com/watch?v=r4QRU6q0UDQ (4:00 and on) yes? no? maybe?
For the sake of testing ild try a Fire Ele and a 55/SV necro, different classes and very different kill speed.

Mr. Undisclosed · Mar 31, 2008, 01:13 AM // 01:13

Quote:

Originally Posted by My Lipgloss is Cool

I think if the build made a difference, the difference would have been much more noticable then what it was.

I doubt it, I think for the most part its based on your timestamp and anything else depends on your speed of kill.

Nechtan Thaumaturge · Mar 31, 2008, 01:15 AM // 01:15

On a whim, I decided to try doing a synchronized run with two characters running the exact same build, as a sort of experimental control case. I expected that the drops would be exactly the same, since presumably the slight differences observed in previous trials were due to the fact that two different builds were being used. Things didn't turn out that way, though. In fact, the results were about what they always have been. About 90% of the total drops were shared, and each character got 5-10 unique drops. The unshared drops included items of every rarity (save green items as I did not kill any bosses), collector items, lockpicks, materials, and coin drops -- in other words, both loot scaled and non-loot scaled items.

Also, I did a second run to try to verify my results, and the character that came out ahead the first time received the smaller number of drops the second time around.

Subject to this phenomenon being verified, I would tentatively suggest the following:

1) Until this can be explained away, any conclusion about the effect of differing builds on drops has to be carefully considered. In particular, if one build gets, say, ten more drops than another on a synchronized Fahranur run, you can't automatically say that it is better because plus or minus ten drops would seem to be within the normal amount of variability. A larger difference, something more like 20-30, would presumably be needed.

On the other hand, the original hypothesis, that Shield of Judgment and Spoil Victor give the same amount and quality of drops, would appear to be strengthened, since the trifling differences in those tests could very well be the result of other factors.

2) There's something affecting drops that has nothing to do with either the build or how it's run (at least not over the course of a single run). This may be old news, but at least now there's something akin to proof to back up the supposition.

Quote:

Originally Posted by ccruzp

To the testers, arent minotaurs in Anvil Rock easy to sync?

I'll try this one next time out.

Mr. Undisclosed · Mar 31, 2008, 01:58 AM // 01:58

Nice research Nech.

Ambuu · Mar 31, 2008, 02:28 AM // 02:28

Just a question:

As long as the two people zone at the same time, can they take as much time as they want? Like one person goes first as the other person waits 5-10 minutes then goes?

Armond · Mar 31, 2008, 02:43 AM // 02:43

Quote:

Originally Posted by Ambuu

Just a question:

As long as the two people zone at the same time, can they take as much time as they want? Like one person goes first as the other person waits 5-10 minutes then goes?

Ought to be.

A Simple Farmer · Apr 14, 2008, 05:45 PM // 17:45

Hey Skinny, How small is the margin of error for syncing? The nearest second, 1/10th of a second 1/100 of a second?

The reason I ask is I've been chest running for the inscribable magma shields. They chest only spawns every 6 runs or so, and apparently 1 in 20 chests have a shield. Needless to say it's a lot of runs.

I've been tracking how often it spawns and the time at which I enter (to the nearest second) in the hopes I might see a pattern (same can be said for the Saurian Scythe chest). Then I remembered the 'sync' was hard to achieve. And perhaps all my data gathering has been in vain as it wasn't accurate enough.

While I'm thinking of it, when you were synced with your brother were the chests in the same locations with the same items within?

A Simple Farmer

ps.

Awesome thread!!!!

Skinny Corpse · Apr 14, 2008, 10:14 PM // 22:14

It's really small.. I'd say it's about a 1/10th of a second or so.. can't say exact tho but
it sure can be tricky to hit, and it usually takes several tries.

We had chest spawning in the same locations and with the same drops from them during
our runs.. can't remember if we documented it tho.. might have been the one that I screwed up the screenshots of..

Chthon · Apr 15, 2008, 12:33 AM // 00:33

Quote:

Originally Posted by A Simple Farmer

Hey Skinny, How small is the margin of error for syncing? The nearest second, 1/10th of a second 1/100 of a second?

Computers measure time in milliseconds (1/1000 sec), so the window could potentially be that small. (And that's how I would seed my srand() if I were writing the program...) However, we have to recall that the server is basically running in a loop performing its basic tasks in a cycle, and checking for zone transfers probably only happens once per cycle. So everyone who crosses the zone boundary at any time during the same cycle gets "recognized" as crossing the boundary at the same time. (Well, actually they get recognized sequentially, but probably within the same millisecond.) How long does a cycle take? That depends on the server hardware and the processor load. A cycle may be shorter in magumma stade than rata sum. And a cycle might be shorter if a-net updates their server hardware. And, since it may not be the case that they're using uniform hardware right now, a cycle may be shorter for some zones and/or in some geographic territories that got better hardware than others.

jiggles · Apr 15, 2008, 12:57 AM // 00:57

very intresting decteive work there. would love to see another couple of runs though

on a side note, your 55hp necro has 74hp o.o

Skinny Corpse · Apr 15, 2008, 04:25 AM // 04:25

Quote:

Originally Posted by jiggles

very intresting decteive work there. would love to see another couple of runs though

on a side note, your 55hp necro has 74hp o.o

We both do after killing two bosses along the way.

A Simple Farmer · Apr 15, 2008, 05:42 AM // 05:42

Quote:

Originally Posted by Skinny Corpse

It's really small.. I'd say it's about a 1/10th of a second or so.. can't say exact tho but
it sure can be tricky to hit, and it usually takes several tries.

That's unfortunate I think the data I've been gathering might not be as usefull as I thought

. Short of writting some automated entry bot I don't think I can get the zone entry time acurate enough... Of course, if someone out there did write one and got enough data it would be possible to figure out a chest spawn clock, which is way beyond my turing skillz.

Quote:

Originally Posted by Skinny Corpse

We had chest spawning in the same locations and with the same drops from them during
our runs.. can't remember if we documented it tho.. might have been the one that I screwed up the screenshots of..

Now that's very interesting and probably more usefull. It means that, monster spawns, chest spawns (and their loot) are in the same loot Zone Calculation. Which makes sense because the magma chest will only spawn if the quest is active (tied to the mob).
I'm guessing for the Magma Chest, that since it's spawning a fairly consistant 1 in 6 ratio it's possible there is a specific mob spawn that matches the chest spawn. (Hopefully there's only 6 combos or 12, not 18 or 5052 or some random number times 6!!)
(starting to screen cap my zones, and throughout my run)

As a side, what's the ratio that the raptor chest spawns, or even any non-quest related chest spawn at (any chance it's 1 in 6 as well)?

Quote:

Originally Posted by Chthon

... How long does a cycle take? That depends on the server hardware and the processor load.

I like your idea, but doubt that the zone window is calculated based on the clock cycle, you'd have to have some sort of throttling control (ever run a REALLY old game on a PC where the movement was based on clock cycle? it's like they're on speed). I imagine it's more like the 'drunkard' window (some sort of fancy load distribution).
Or because it is a server client environment, it's done client side every 0.xxxx seconds (which'd be really stupid for them to do security wise).

etek · Apr 15, 2008, 08:41 AM // 08:41

hi,

there is a simple explanation 4 equal drops - do u log on through the same router..? if u do, that's it..

- that way u have the same IP for the incoming and outcoming con... Try to hook on using different con, for instance one through some kind of "air" con (GPRS, satellite,...) and the other one using standard way...

Cya, Kire.

Skinny Corpse · Apr 15, 2008, 02:17 PM // 14:17

Quote:

Originally Posted by etek

hi,

there is a simple explanation 4 equal drops - do u log on through the same router..? if u do, that's it..

- that way u have the same IP for the incoming and outcoming con... Try to hook on using different con, for instance one through some kind of "air" con (GPRS, satellite,...) and the other one using standard way...

Cya, Kire.

I know it has become pretty big thread but we've been over this several times before..
We have different connections and even different ISP's.. we DO live in the same town but that's the only thing we have in common.

Chthon · Apr 15, 2008, 06:12 PM // 18:12

Quote:

Originally Posted by A Simple Farmer

I like your idea, but doubt that the zone window is calculated based on the clock cycle, you'd have to have some sort of throttling control (ever run a REALLY old game on a PC where the movement was based on clock cycle? it's like they're on speed). I imagine it's more like the 'drunkard' window (some sort of fancy load distribution).
Or because it is a server client environment, it's done client side every 0.xxxx seconds (which'd be really stupid for them to do security wise).

Please reread my post, you have misunderstood. I never said anything about the "clock cycle." I did say it's probably tied to how quickly the server can go through an iteration of its main loop. (I probably confused you by using the word "cycle"; I'll use "iteration" from here on out to avoid confusing you again.) Because there is probably only one check for I-want-to-start-a-new-instance requests performed per iteration, such requests are only going to get honored once per iteration. So the duration of the window when you can send a I-want-to-start-a-new-instance request after another player and get the same seed that they get is going to be the amount of time until the server executes around its main loop and reaches the step where it checks for such requests.

(On a sufficiently fast computer (which I don't think a-net has), one iteration of the main loop could be executed in less than a millisecond, so 1ms would really be the size of the window. But I don't think a-net's servers are anywhere near that powerful.)

Zhoul · Apr 15, 2008, 08:24 PM // 20:24

I'm really impressed with the results of the first and third runs... very interesting... great job there!

A Simple Farmer · Apr 15, 2008, 10:15 PM // 22:15

Chthon, I wasn't confused by your post, and I did understand it. I'm just doubtfull that faster hardware = smaller window to sync zones. Again, it was your statement "How long does a cycle take? That depends on the server hardware and the processor load." that I found fault with.

I don't think I was clear, I just doubt that there isn't some "throttle control" for as mundane a task as watching the door. By that I mean 'A cycle is independent of the server hardware and the processor load'.

Thus my example which probably confused you, The old PC game (pre windows 3.1). In these games, faster machine = faster game, to the point where old games were unplayable on a new machine. So programers learned from this and stoped tiing their iterations of loops to the clock cycle.

So that being said, I doubt that they would make any part of the games performance dependent on the speed of the hardware it's installed upon (server side). If they did, that 'loop' would uncontrollably execute that loop as fast as it possibly can (100% load).

That being said, I'm probably wrong & so are you, but I bet we can test this out

.

Testing you theory should be relatively easy to do. Towns under high volume/load would have a larger window to ZoneSync. So it should be easier to sync from USA TOA D1 on a Saturday afternoon on a free entry day than say, UK Piken Square D1 Wed at 3am.

If it's equally difficult to ZoneSync in both towns, then we have learned something else, non?

I think posting # successful attempts out of say 30, with the time, town & district (server IP if possible)?