Xoya

Well, for warriors that is definitely true, galzohar. The problem for shamans at least is that fast offhands consume more windfury weapon procs (compared to your mainhand) than slow offhands do.

 

galzohar

As a shaman don't you want to use 2 slow weapons for stormstrike as well? should be a no-brainer ;p
Warriors are where things get more complicated.

 

Tornhoof

Yes, I'm following the thread, I currently have the problem, that a closed form description of flurry does include the uptime produced by WF and Stormstrike procs, but at the same time, the proc rate for WF is partially dependent on the weapon speed.
That is kinda ugly and I will need to find a way around it, before I can successfully produce a closed form description of flurry with wf for now.


As for a closed form description for the shaman, I think flurry uptime needs to be splitted into MH and OH flurry uptime and I'll need to integrate WF and stormstrike into it.
Maybe I can get around the wf proc stuff, by assuming "each 3.15s wf will proc with dw", but we'll see.

 

• Disquette

Tornhoof, I think you have the hardest job in WoW, making a closed-form of this :-)

 

Tornhoof

Edited: Not yet ready

To get Flurry working with WF and SS, I need to atleast model 4 swings for flurry, since with Dual 2.6s weapons, the attackspeed will be below 1s, but I'm close to presenting a flurry model for 3 charges which works with WF and Stormstrike.


Currently it looks like this, under the assumption of a 3s Cooldown, for faster weapons with <=1.0s attackspeed, for example one string of swings.
1.0 attack speed example
F* flurried
NF* non flurried,
WF wf
SS SS

F1,F2,F3,NF1,NF2+WF+WF, F4,F5,F6,NF3,NF4+WF+WF+SS1+SS2,F7,F8,F9

Now for the first wf proc, NF1+NF2 can proc WF, for the second F6, NF3 and NF4, SS1,SS2 can proc flurry too

The exact swing count necessary is dependent on the weapon speed, for slow weapons (>2.6s) it should be safe to use 4 swings, for faster weapons you'll need more, for daggers certainly more than 6.

Another approach might be, calculate the swings necessary that flurry is up 97.5%, let's say 10 swings, now calculate the amount of white swings in there and calculate the the probability that flurry procced multiple times for those swings, then calculate the probability that we can get let's say 7 charges from 10 swings and take it from here.
A proper simplification would be to say 9 charges and say what is the probability to crit 3 times in 10 swings (simple combinatorics now and substract the probability to crit 3 times in a row), this will underrate flurry a bit but will cope for missing stormstrikes or windfuries and should be fairly correct and since we just said 97.5% it should work out nicely.

 

T.K.

Sup, everyone. I did have some points i wish to share since this post was introduced in here, but coudn't register before^^

As i checked the options u guys made about why the procs where working this way, it seems to me that u guys forgot something very importante about WoW theory in general:

The programers boundaries and even the programing language limit.

My point into this is that, all the available options that i saw in here where made as if the WoW universe could be created with any possibility, what is not true when u build a world based upon a programing language.

Into a very basic level, first and foremost, about the 4 (and even more) procs that we sometimes get... i think this could be answered with a basic programing concept like true/false check:

Moreso, all skills ,i think, work like a server active command -> client passive response, but i'll get into this a bit more futher in this post...

An example of the lol flurry procs with x number procs:

1.0 MH crit hit *Flurry starts*

1.1 OH hit
Hit, so take 1 charge out. Charges left 2
Server performs check with something like this: IF when MH hits, u have zero or minus charges, THAN stop flurry state. True or false? False = KEEP DA HASTE DAMMITT!!!


1.2 MH hits ( I won't include crits or whatever besides normal hits for now)
Hit, so take 1 charge out. Charges left 1
Check => MH hit (true) = zero or less charges (false) => False = HASTE, HASTE, HAAASTEE!!


1.3 OH hits
Hit, so take 1 charge out. Charges left 0
Check => MH hit (false) When it gets a OH hit, with the programing script, it cannot serve as a ruler for ending or not the charge state, so it skips this check.

1.4 MH hits
Hit, so take 1 charge out. Charges left -1
Check => MH hit (true) = zero or less charges (true) => True = WTH!!! STOPIT U DAMMIT!!!!!!!! STOP FLURRY!! NOOOWWWWWWW

ok, although my server seems like someone about to have a heartache, this is more or less how i see the check it performs to stop or continue flurry. It get's the hand that made the flurry to pop up, and than leaves it as the ruler to decide to stop or not.
I made the example with the MH as ruler, but is the same if u get a crit hit with the OH, and so it's it that rules the checks from that moment on.


That said, what i think it's causing lot's of procs with just a simple crit proc, would be that to prevent hacking and whatever, blizzard made the programing of the game with the client only being able to inform stuff and the server taking care of decisions. With this, server only receives information and send information back, while the server procs em and see what happens everytime.
I really think this to be the way how things work, as every single game that the client allows for abitrary commands or decisions has exploits or hacks available, because most of the time they are built with these already in, deciding tools.
This would lead us to a situation where for any reasons varying from connection lag, lost of information into the transmition, or whatever u guys might think, the client will keep the exact same situation and commands the server sent, while the server gives no other command. As such, if the server command to stop flurrying, or to reduce a charge in mid flurry state got lost or somehow delayed, this would lead to an increase into the number of procs, in part cause of the first programing limits i said and in part for this.

Right now i'm really late to get my gf, but if anyone have any interest into this or think might have any clue to solve, i might elaborate a more complex scenario with delayed information grouped with programing check limitations to see if in agrees with the results we got into those posted tests up the post.

 

Xoya

One thing to try, then, based on the most recent post: can we somehow prolong a flurry state inevitably by removing the weapon which crit before flurry falls off? Not that this would help our DPS at all, but it would certainly be an interesting result regardless.

 

T.K.

This wouldn't change the way the procs are counted, as although u are hitting much much less with a bare hand, u're still hitting with your MH and OH whatever the cause could be

 

Blightlenox

T.K.'s theory sounds pretty valid after re-reading the combat logs from the above posts.

 

Duranthor

So does this mean then, that if you had a 2.6 MH and a 1.3 off hand and you crit with your main hand, you could then expect to see as many as 6 swings with your off hand under the influence of flurry before the server tells it to stop for three main hand swings?

Could we then claim that the closer in speed your off hand weapon is to your main hand, the less chance you would get for extended flurry strings?

In other words, fast off hands do not eat up flurry charges for warriors because the the MH holds all the counters.

Is this what the evidence supports then?

 

T.K.

To be honest, i really couldn't try that options, although i imagined them. At these days i had so much problems with my stupid guildmates, trying to prove them what for my enhance shammy can be good.

Basically a retarded resto shammy estarted complaining that at our VR fight he had to burn all his mana, plus Tide totem, plus 4-5 pots just to keep me alive while the other fights where my shammy wasn't in he didn't even use 80% of his mana pool, with just rogues on the group to be healed... But whatever, i'm very pissed but this is not for this topic...

If someone could get 2 very different speed weapons, with a very fast offhand and see how much strikes while flurried u can get, would help a lot. Myself, i discarded all fast weapons i got.

If anyone already tried this, could please post results?

 

• Disquette

I'm sorry, somehow I missed this post earlier. My contention is that weapon speed has a lot to do with % of flurried attacks. This is because of the amount of yellow attacks in a 10 second period compared to the number of white attacks in the 10 second period.

 

Petit_Mal

I would suggest that there is a very simple solution to this question.

Buffs last, at minimum, until the end of the current, serverside, second.

That is- once a buff has been consumed, it does not fade until the next second. In each of the cases where you were getting 6 attacks after gaining flurry, pairs of your subsequent attacks occurred, each time, within the same second. During each of those seconds, the server registered that you got "at least one" attack, and so a single aura of flurry faded at the end of that second.

This mechanic can amusingly be seen with the warrior ability "Spell Reflect" wherein being struck by a spell procs a hidden "reflection" buff that lasts for 1 second. If you get hit by more than one spell during that second, you reflect all of them.

Given general buff mechanics, this seems like the most parsimonious explanation.

 

• Disquette

Interesting, thank you for the insight and example :-)

 

Petit_Mal

np. I was just glad to have something to contribute to these forums. ;-)

(The name of the hidden aura gained is actually "Reflect Spells." My bad.)

 

Tornhoof

I even think that this (next second) is actually a dynamic timeframe, load dependent timeframe, e.g. higher in instances on dedicated servers and lower on the outside

http://code.google.com/p/wowequipoptimizer/

 

p o j

I realize this topic has fallen from recent interest, but was wondering if anyone has done any WF totem testing other than Tornhoof or in addition to his. More than likely the obvious guess is the answer, but I feel it's worthwhile to just double-check.
Have any discoveries or insights been made as to server lag between WF procs (or Sword Spec procs) affecting consumed flurry charges? Since lag seems to be the fallback safety response for the extended flurry state, is there any chance it could work against a player? Or is the game coding such that doesn't allow this sort of mix-up?

Obviously this is of interest as a dps warrior gets a significant boost by both flurry and WF.

You can't stop the love train, baby.

 

anticide

Isn't time-to-next-swing 77% of the normal length? Assuming 30% haste flurry.

3.0 weapon hasted 30% swings (3.0 / 1.3) = 2.30. (2.30 / 3.0 ) ~ .77.

 

Machinator

I looked through the thread and the question was asked if dodge, parry, block, miss take a flurry charge and I didnt see it answered. I read that miss might not, so I want to double check if they use charges or not.

"Information is ammunition."

 

WazAg

I've never been a proponent of the "Fast OH consumes charges" school of thought, because, when doing my own study -- which is, admittedly, less thorough and intricate than the current thread's research -- I found the only occasion this could be a true statement, given a logical (i.e. 3 straight swings) flurry count is if you were to equip an offhand that was more than twice as fast as your mainhand, to have your mainhand crit and be instantaneously followed up by a sequence of three offhand attacks.

Given that, at best, I can think of one weapon combo outside of the Kael fight that achieves this (Dragonstrike (2.7) + Searing Sunblade (1.3), and taking into account the one-second aura theory, there is no reason not to use a fast offhand, and, in fact, extra duration for each flurry proc speaks to a fast offhand being even better.

I've always followed this logic, but hopefully these arguments will keep warriors, at least, from reducing their dps via. faulty assumptions.