Jump to content


Photo

Modeling Flurry with DW and WF


  • Please log in to reply
38 replies to this topic

#21 berg

berg

    Don Flamenco

  • Members
  • 466 posts

Posted 30 November 2006 - 02:51 PM

I am not sure that I see how you can model the weapons independently. Flurry hastes both weapons. The roundoff is always in the players favor when the buff fades so modeling them separately seems to cheat your dps.

#22 Scregle

Scregle

    Glass Joe

  • Members
  • 20 posts

Posted 30 November 2006 - 03:30 PM

I am not sure that I see how you can model the weapons independently. Flurry hastes both weapons. The roundoff is always in the players favor when the buff fades so modeling them separately seems to cheat your dps.

flurry's tooltip says it increases attack speed by 30% for 3 swings
if that works out to 3 attacks at 30% increase in attack speed, each weapon can be modeled independently.
I did test with /combatlog and it appeared as though it worked that way. I did not do enough testing to verify this, however.

#23 Celandro

Celandro

    Don Flamenco

  • Members
  • 379 posts

Posted 30 November 2006 - 05:22 PM

What the tooltip says doesn't really matter. What matters is how it is actually implemented in the game. The key test if you are willing to do it is take 2 slow weapons of the same speed, and see if they ever do not swing at the same time. If flurry lasted 3 swings, they should eventually not swing at the same time (MH hit, OH crit, mh hit, of hit, mh hit [flurry fades] oh hit). Current theory is that flurry will last 4 swings if neccessary (3.5 on average since it will last 3 if MH crits instead of OH) if weapon speed is exactly matched (eg flurry wont fade until after the OH hit.

Test #2 is you add windfury to the mix. If it is white damage, curent theory states it should eat flurry charges(1), generate rage(2) and glance(3). If it is yellow damage it should not do those 3 things but should be affected by impale(4). 2 and 3 are easily proven, 4 requires some damage meter math which is fairly easy as well (average damage of WF crit should be 2.1 * average damage of WF hit )
1 is a bit difficult as you need to look in the combat log and prove/disprove the following sequences:
- MH crit, OH hit, windfury hit, MH hit, OH hit (flurry fades)
- MH hit, OH crit, windfury hit, MH hit, OH hit (flurry fades)
- MH hit, OH crit, MH hit, OH hit, windfury hit (flurry fades) NOTE: this one is key

Of course modeling this gets a bit tricky to say the least. Feel free to correct me if you see anything wrong :) Ive been known to make mistakes.

#24 berg

berg

    Don Flamenco

  • Members
  • 466 posts

Posted 30 November 2006 - 05:34 PM

my guess is that it would be implemented similar to this

pertinent values:
// mainhand
mh.Speed // weapon speed on tooltip effectively a constant.
mh.Delay // current time waited since last swing
// offhand
oh.Speed
oh.Delay
// flurry is a player buff not weapon it applies to both.
bFlurry
nFlurryCounter // number if swings left before flurry expiration

gameUpdate(float dt) //dt will be something like 0.05f 
{
	if(bFlurry)
	{
		mh.Delay += dt*1.3;
		oh.Delay += dt*1.3;
	}
	else
	{
		mh.Delay += dt;
		oh.Delay += dt;
	}

	if(mh.Delay >= mh.Speed)
	{
		nFlurryCounter--;
		mh.Swing(); // this will calc damage and reset delay timer and potentially reset flurry counter. 
	}

	if(oh.Delay >= oh.Speed)
	{
		nFlurryCounter--;
		oh.Swing(); // this will calc damage and reset delay timer and potentially reset flurry counter. 
	}

	if(nFlurryCounter <= 0)
		bFlurry = false;
}
The important thing to notice is that even if the offhand consumes a final charge of flurry that main hand will still have gained the haste for the duration of its downtime up until that final swing.


The other possible implementation is what you describe where flurry divides your weapon delay constants by 1.3 but programatically it seems nicer to have your target be a constant instead of a changing value.

#25 oldmandennis

oldmandennis

    Don Flamenco

  • Members
  • 375 posts

Posted 01 December 2006 - 05:58 PM

Stormstrike
Stormstrike is weird. It doesn't seem to proc Windfury anymore. However, when it comes to Flametongue and Frostbrand, it seems to be very odd. If either weapon has one of those enchants, it seems that there are up to three proc chances: one for each weapon with those enchants, and an extra one that just seems to magically come with Stormstrike itself. That is, if you have Flametongue on one of your weapons and use Stormstrike, you get two Flametongue procs. If you have Flametongue on both and Stormstrike, you get three procs. It's very strange, and I really couldn't figure out why it was happening.

I was testing a bit last night, and it was even weirder. FT OH was one proc (WF main) but FT MH was 3 (WF OH)??

quick edit: PTR, not beta

#26 Skiace

Skiace

    Don Flamenco

  • Members
  • 324 posts

Posted 02 December 2006 - 02:10 AM

in regards to weapon enhancement mechanics, does flametongue only do damage on a melee hit/crit/glance? i would expect this to be the case, but i'm not sure.


edit: assuming it does damage on all hits, crits, and glances, i've implimented flametongue in my table according to Vulajin's post. I also added rockbiter, but according to thottbot/wowhead, the tootip now reads "Imbue the Shaman's weapon, increasing its damage per second by X" for each rank. if it's this simple, then it's very easy to just add that +dps at the same place where i add dps from AP.


edit2:

here's the link to the updated sheet with rockbiter and flametongue added. http://ieng9.ucsd.ed..._statsDWv07.xls

here's a picture of the section i added.

Posted Image

the yellow section is where you specify which enchant on which weapon, with a 1 or 0.

below that is shown the damage done by each enchant. "proc dmg" and "crit dmg" are just shown for reference. the "dps" column will only display the enchant's overall dps contribution if that enchant has been selected in the yellow section above. the windfury calculations are done elsewhere in the sheet, where they were done originaly.

you can basicly turn on all the enchants, and compare the dps of each listed in this green section. the dps of whatever enchants are "turned on" is added to the overall dps under the heading "enchants" which is where "windfury" was previously. resists and armor are not taken into account at all.


as for observations, it takes some extreme-ish stats to make off-hand flametongue put out better dps than windfury. with a 2.8 speed main hand, 1.3 speed off hand, and 10% spell crit, i'm seeing for example 73dps from FTOH, and 69dps from WFOH. change the off-hand speed to 1.4, and you get WF dps unchanged, but the FTOH dps drops to 68. at higher AP levels, WF continues to scale. at higher spell-damage levels flametongue would scale a bit also, but that's another mechanic that i have yet to implement.

with 2.6/1.4 MH/OH at level 60, i had to drop the AP down to about 500 (unbuffed) to get flametongue to come out as the better dps option. at level 70, this threshold was about 1000AP unbuffed. I'd like to add spell damage to the calculations though, to make it easier for people to compare gear.

#27 Vulajin

Vulajin

    Vula'jin the Void, blessed by the loa

  •  Patrons
  • 5924 posts

Posted 02 December 2006 - 02:54 AM

in regards to weapon enhancement mechanics, does flametongue only do damage on a melee hit/crit/glance? i would expect this to be the case, but i'm not sure.

edit: assuming it does damage on all hits, crits, and glances, i've implimented flametongue in my table according to Vulajin's post. I also added rockbiter, but according to thottbot/wowhead, the tootip now reads "Imbue the Shaman's weapon, increasing its damage per second by X" for each rank. if it's this simple, then it's very easy to just add that +dps at the same place where i add dps from AP.

Thottbot's not entirely accurate anymore. Rockbiter now says "each attack deals X to Y damage. Slower weapons deal more damage per swing." Both Rockbiter and Flametongue proc only on glances, hits, and crits. Rockbiter and Flametongue also each have independent chances to crit (i.e. Rockbiter doesn't necessarily crit if your weapon crits, and a Rockbiter crit may occur on a weapon hit/glance).

below that is shown the damage done by each enchant. "proc dmg" and "crit dmg" are just shown for reference. the "dps" column will only display the enchant's overall dps contribution if that enchant has been selected in the yellow section above. the windfury calculations are done elsewhere in the sheet, where they were done originaly.

you can basicly turn on all the enchants, and compare the dps of each listed in this green section. the dps of whatever enchants are "turned on" is added to the overall dps under the heading "enchants" which is where "windfury" was previously. resists and armor are not taken into account at all.

My spreadsheet does things similarly, and I've implemented armor (assuming a typical mob has 20% reduction), but I don't yet have a good understanding of spell mechanics, so I can't quite implement the proper mechanics for Flametongue. Also, I have no idea what the proc rate on Frostbrand is. If you don't mind, I'll probably rely heavily on your spreadsheet to figure out some of the mana regeneration and usage stuff to implement the spell components of total DPS.

Yep, still a fucking idiot.


#28 Skiace

Skiace

    Don Flamenco

  • Members
  • 324 posts

Posted 02 December 2006 - 03:07 AM

another issue i just noticed, the raw dps comparison numbers for the enchants don't take into account the effect of windfury on flurry. thus the true dps contribution of windfury is higher than what's displayed in that section. regardless, when you actualy compare the total dps output between dual WF and WF/FT, the flurry contributions are taken into account and you can see which one comes out on top.

also, do you have the currently accurate high/low numbers for rockbiter?

does anyone know the scaling factor of flametongue with regards to spell damage gear?

#29 Vulajin

Vulajin

    Vula'jin the Void, blessed by the loa

  •  Patrons
  • 5924 posts

Posted 02 December 2006 - 03:17 AM

another issue i just noticed, the raw dps comparison numbers for the enchants don't take into account the effect of windfury on flurry. thus the true dps contribution of windfury is higher than what's displayed in that section. regardless, when you actualy compare the total dps output between dual WF and WF/FT, the flurry contributions are taken into account and you can see which one comes out on top.

also, do you have the currently accurate high/low numbers for rockbiter?

does anyone know the scaling factor of flametongue with regards to spell damage gear?

Here's how I handled that Flurry thing. The very first thing I calculate is the overall effective speed of both weapons (i.e. how much they get accelerated by Flurry, based on the user's crit rate and which -- if either, or both -- of their weapons has Windfury). Then, I use the base stats of each weapon to determine the "per swing" effects of the various procs, including Windfury, and divide those by the effective speeds to determine the actual DPS. The same thing is done for Unleashed Rage. Thus, the effect of the crits taking place is taken into account before the damage of the crits is actually calculated.

Regarding your other two questions, I'm afraid I don't have answers to either of those. I'm basically waiting on the patch next Tuesday, because I don't have time usually to get on the PTR (I also don't have my own level 60 shaman...mine's still 39).

Yep, still a fucking idiot.


#30 Skiace

Skiace

    Don Flamenco

  • Members
  • 324 posts

Posted 02 December 2006 - 04:05 AM

I forgot to apply Elemental Weapons to flametongue and rockbiter calculations. that makes a big difference. it looks like off-hand flametongue may be dps-competitive after all. hmmmmm.

#31 Vulajin

Vulajin

    Vula'jin the Void, blessed by the loa

  •  Patrons
  • 5924 posts

Posted 02 December 2006 - 04:14 AM

I forgot to apply Elemental Weapons to flametongue and rockbiter calculations. that makes a big difference. it looks like off-hand flametongue may be dps-competitive after all. hmmmmm.

Remember that Rockbiter procs are reduced by armor, offhand Rockbiter procs are reduced by 50% (the offhand penalty), and all offhand Windfury swings are also reduced by 50% (the offhand penalty). Offhand Flametongue damage per proc is based on your main hand speed, not offhand speed (this may or may not be a bug, but it definitely works this way on the PTR), and it isn't reduced by 50%, nor by armor.

Yep, still a fucking idiot.


#32 Skiace

Skiace

    Don Flamenco

  • Members
  • 324 posts

Posted 02 December 2006 - 04:22 AM

I forgot to apply Elemental Weapons to flametongue and rockbiter calculations. that makes a big difference. it looks like off-hand flametongue may be dps-competitive after all. hmmmmm.

Remember that Rockbiter procs are reduced by armor, offhand Rockbiter procs are reduced by 50% (the offhand penalty), and all offhand Windfury swings are also reduced by 50% (the offhand penalty). Offhand Flametongue damage per proc is based on your main hand speed, not offhand speed (this may or may not be a bug, but it definitely works this way on the PTR), and it isn't reduced by 50%, nor by armor.

got all that now. also did a bit of searching here and found that flametongue procs get +10% of spell damage gear on each hit, so i added that as well.

playing around with the player input stats, i'm seeing flametongue as being the dps winner for the off-hand in all but the highest end of gear. it looks like windfury scales better, but it takes some pretty high AP to catch up to flametongue. if this is true in reality, i wouldn't be surprised if we see flametongue get nurfed such that the off-hand speed determines off-hand damage.

#33 LordVoid

LordVoid

    Von Kaiser

  • Members
  • 39 posts

Posted 02 December 2006 - 03:02 PM

http://forums.elitis...pic.php?id=9408

For now, I keep it simple by only considering white damage. Basically ALL you have to do to figure this out is to make your flurry formula take into account the effect that an extra attack has on it. So the chance that your current swing is 30% hasted is equal to the chance that either the last attack crit, or the attack before last crit and the next attack was not a SS proc, or the attack 3 attacks ago crit and the following 2 were not sword spec procs. Sword spec procs do NOT make use of haste. If your formula for calculating flurry takes into account the effect extra attacks have when they do not crit, then simply take your number of swings, multiply by the new flurry formula, and multiply by the additional percentage of swings you get from extra attacks, and you are golden.

I think that I may be a tad unclear in that post, but I'm confident my calculations have addressed some very pressing questions with how flurry and extra attacks interact.

#34 Vulajin

Vulajin

    Vula'jin the Void, blessed by the loa

  •  Patrons
  • 5924 posts

Posted 02 December 2006 - 09:26 PM

Excellent work in that thread, LordVoid. I'd just like to point out, though, that Windfury swings are slightly different from Sword Spec swings, because they don't eat Flurry charges. Still, I wonder how one might incorporate your formula to model the effect of, say, Hand of Justice.

Yep, still a fucking idiot.


#35 LordVoid

LordVoid

    Von Kaiser

  • Members
  • 39 posts

Posted 03 December 2006 - 04:13 PM

Hand of Justice consumes flurry charges just like sword spec does. The only thing I'm unsure of is if it is posssible for sword spec and HoJ to proc on the same swing (or off of one another). If neither is true, then you just find a new value for s in my formula. Assuming they cannot compound proc, off the top of my head I'd think it would be 1 - ((1 - S) * (1 - H)), where S is your sword spec chance and H is your HoJ chance. Assuming fully specced sword spec, this comes to 6.9%

Edit: If you only have HoJ and no Sword Spec, then make the s in my original formula .02(2%).

Edit: Incidentally, if Windfury does not consume charges then there's no need to consider it in the Flurry formula.

Edit: Actually, yes we would have to take into account its ability to refresh flurry.

Are we talking about Windfury now or windfury after the patch? After the patch, isn't windfury a single swing that hits for double damage? Isn't it also yellow damage? Furthermore, does it always occur simultaneously with the regular swing that procced it? If so, then essentially that swing has two separate chances to refresh/activate flurry, and we can calculate accordingly (with relative ease).

#36 LordVoid

LordVoid

    Von Kaiser

  • Members
  • 39 posts

Posted 03 December 2006 - 07:33 PM

Since in 2 days windfury's current behavior will be strictly irrelevant, I will focus on the 2.0 version of Windfury.

I'm assuming that:
Windfury is a single hit for double damage.
Windfury is a yellow attack.
-yellow attacks do not consume flurry charges.
-yellow attacks refresh flurry when they crit.
Windfury occurs simultaneously with the swing that procced it.
Windfury can crit.

c = crit percentage.
w = windfury chance.

So I like to talk in terms of swings. You can convert it later by multiplying swings by average damage per swing. The chance that any given attack is flurried is equivalent to the chance that any of the last three swings themselves crit or procced a windfury that crit.

I've worked this chance out to be: 1- ((1-c) * (1-(w * c)))^3

Multiply this by .3 and you should have your average haste provided by flurry.

With 30% crit, 20% windfury proc rate, and 60 swings a minute before flurry, we have:

21.45329064% average haste and 72.871974384 swings a minute for white damage. 20% of this amount of swings is the number of WF procs.

#37 Skiace

Skiace

    Don Flamenco

  • Members
  • 324 posts

Posted 03 December 2006 - 07:52 PM

Since in 2 days windfury's current behavior will be strictly irrelevant, I will focus on the 2.0 version of Windfury.

I'm assuming that:
Windfury is a single hit for double damage.

are you sure about that? i thought they changed it back to two extra yellow hits. surely someone on test can confirm this?

other than that, the mechanics you outlined are the ones i used in my sheet.

#38 LordVoid

LordVoid

    Von Kaiser

  • Members
  • 39 posts

Posted 03 December 2006 - 08:29 PM

Since in 2 days windfury's current behavior will be strictly irrelevant, I will focus on the 2.0 version of Windfury.

I'm assuming that:
Windfury is a single hit for double damage.

are you sure about that? i thought they changed it back to two extra yellow hits. surely someone on test can confirm this?

other than that, the mechanics you outlined are the ones i used in my sheet.

If windfury is two extra yellow attacks then we're looking at (cut and pasted for ease):

1- ((1-c) * (1-(w * (1- (1 - c)^2) )))^3

Multiply this by .3 and you should have your average haste provided by flurry.

With 30% crit, 20% windfury proc rate, and 60 swings a minute before flurry, we have:

22.54848835032% average haste and 73.529093010192 swings a minute for white damage. 20% of this amount of swings is the number of WF procs.

#39 Vulajin

Vulajin

    Vula'jin the Void, blessed by the loa

  •  Patrons
  • 5924 posts

Posted 03 December 2006 - 10:12 PM

Windfury is two extra yellow attacks.
All your other assumptions on Windfury are correct.

Yep, still a fucking idiot.





0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users