Pater posted something showing the possibility of a DPS drop between 1.44 and 1.50 attack speed area. The drop started at 1.50 and went back up to pre-1.50 areas at 1.44%. I'll see if I can find it again. Don't know if its been proven, but some interesting data. But it was done to show the effects of haste to our DPS.

I believe it has to do with the fact that being inbetween 1.4-1.5 means that your 2nd hit will land just outside of the 3s WF cooldown. So if your both weapon speed was 1.49, you could only proc WF every 4.47 seconds since your first hit after procing a WF and triggering the cooldown would land at 1.49 then 2.98 then 4.47.

Right, it's a small window of attack speed that basically guarantees you'll be waiting the maximum amount of time possible for a chance at proc'ing windfury after the 3 second cooldown has expired.

Simply put, if your attack speed is 1.5 it is possible for you to proc Windfury every 3 seconds. If your attack speed is 1.49 a Windfury proc is only possible once every 4.47 seconds. See the problem? The extra white damage is not enough to overcome the delay in Windfury procs for this very specific window of attack speed.

It'd be even worse if your attack speed were say 2.9, in which case you could only proc Windfury once every 5.8 seconds. Luckily that's not possible unless you're using a two-handed weapon.

Anyway, as you get lower and lower below a 1.49 attack speed it gets better and better, until eventually it starts actually increasing your dps again instead of nerfing it. Like I said, it's just a specific 'window' of attack speed around 1.49 that is very bad. Generally faster is definitely better.

Yea I get that, but this is all reliant on the fact that you would proc WF as soon as its off CD, which is definitely not always the case.

It's not really reliant on that fact at all. Over time, if you are attacking every 1.49 seconds, you will average less total damage than you would if you were attacking every 1.5 seconds with the same weapons. The slight gain in white damage is not enough to account for the delay to your possible windfury procs.

To put it another way, at 1.5 speed, it is possible for 50% of your attacks to proc WF. Basically every other attack atleast has a chance to proc. At 1.49 speed, it is possible for only about 33% of your attacks to proc WF. After we take into account WF's 20% proc chance, this means that over 100 attacks a 1.5 speed weapon will average about 10 WF procs, where as a 1.49 speed weapon will average about 6.6 WF procs.

The damage you get from the extra 3.4(on average) WF procs every 100 attacks will definitelty surpass the damage you gain by attacking .1 seconds faster. In the amount of time it takes a 1.5 speed weapon to make 100 attacks(150 seconds), a 1.49 speed weapon will have made 100.6 attacks. That's not even 1 extra attack per 100 attacks.

Some things about weapon speed, the next ideas only apply to high-end raiding gear probably, with a flurry uptime of >75%.

Let's consider two 2.6s weapons, without 100% flurry uptime (and no haste gear), with flurry we have an average swing time of 1.0 = (2.6+2.6)/(4*1.3)

F* = flurried swing
NF* = nonflurried swing
WF = windfury

NF+WF,F1,F2,F3

this means: "Windfury procced from an non flurried critical swing and we gain 3 charges"

NF: 0.0s
F1: 1.0s
F2: 1.0s
F3: 1.0s

in this case, our 3rd flurried swing will be still in the 3s CD for Windfury (assuming no lag).

Now if we have slower weapons, with an average flurried swing time of >1.0, let's say dual 2.8s with 6% haste gear (2.8+2.8)/(4*1.36) = 1.03s, the last flurried swing F3 is not in the 3s Cooldown anymore and can again proc flurry.

So the mathematical difference is basically:

1-(1-crit)^3 vs. 1-(1-crit)^(3+2*0.36)

For 40% crit the values are: 0.784 vs. 0.850

Simply spoken, as soon as we reach a certain crit level, where the above example is true more often than not (37% crit -> 75% flurry uptime), there will be a jump in the flurry uptime if your weapons are slower than 2.6s in combination.

The above problem just increases the weight of slow weapons in both hands.

The above swing time problem, is the only problem, I am aware of for weapon speeds which can be shown with every model currently in use.

The bottom line is:
You should be aware, that with high levels of crit you might not see a linear increase of dps, just because your average swing time is below 1.0s now and the above example holds true.
The dps difference is not much, maybe in the 1.5% range


How often the example "wf cd ends at the same time as your flurry" is true, is more complicated to say, there is certainly a crit rate where it will be above 50% (~37% crit).


Edit:
As for the 1.50s/1.49s discussion, there are no realistic weapon combinatins to reach that with dual wield, your average attack speed is well below and the attack speed for your mh is above, if you want to reach the optimal mh wf count.

assume flurry+dragonstrikeproc+dst = 30%+20%+30.95%

[top] 2.7/(1+0.3+0.2+0.3095)

1.492



Note: this does not mean dst is a bad item, the gain in white dmg is greater than the loss of wf procs for that moment, the damage will be on average just less thant having a hypothetical 2.8s weapon with dragonstrikeproc in the mh.

Thanks.

Any thoughts on at what point dropping below 1.5 speed becomes better?

Hmm, let's see, assume flurry uptime of 85%, dst uptime of 40%, dragonstrikproc uptime of 20% (dunno the exact value here), the probability that all three are up at the same time is, simplified, just the individual rates multiplied.

so we have a 7% probability of getting below 1.5s, so just forget it, the individual worth of the different haste procs is superior than the loss of a bit wf damage for the rare case that all 3 effects are up at the same time.

I have a 2.6 offhand (Netherbane) hasted to 2.51 by passive hasterating.

When flurry is up (80%+ in Raids), DST procc is up (>34%) and mongoose is up (1 of 2 is enough) the speed is 1.46. With dual mongoose up even my 2.7 MH is hasted to 1.49.

What would be the best to do now? Take off the 36 hasterating from my belt ( 1.55 /1.5 with FLurry/DST/1 Mongoose) or try to stack more hasterating? Wouldnt the procc from Dragonstrike make things even worse?

I know the best would be to have 2.8 speed weapons, but syphon doenst drop :[

http://elitistjerks.com/398309-post78.html
If you follow Pater's haste model, you'd see an increase in DPS at 1.41 speed, at a 55% haste speed. (Sorry if I've missed if this has been proven to be incorrect)

Though as Thornhoof says it's a rare scenario to have all effects up at once.

Hello,

I would just like to get a quick 2nd opinion to make sure my math is right regarding gems, [Design: Bold Living Ruby] vs [Item not found!] I did the math, and with motw+kings the 20 ap gem still yielded more ap overall raid buffed, is this correct?

(I added strength totem before adding kings if it even makes a difference)

493 str x2 = 987.58 ap (3x 8str gems)
467 str x2 = 994.78 (3x 20ap gems)

Unique?

What you want to do is to figure out the actual AP value of each gem, you're complicating things a great deal when you're adding MOTW and Kings and your AP, when you can just add Kings to the [Design: Bold Living Ruby].

[Design: Bold Living Ruby]: 8x1.1x2=17.6AP
[Item not found!]: 20AP

So yes, it would be better with 3 [Item not found!], but they're "Unique-Equipped", meaning your gear can only have one of those gems in it.

wow totally missed that, thanks

haha

Liking the work you guys are doing with the Haste rating, definetly lots of food for thought.

The question I have, since my guild is currently working on Kael, is which legendary should I be using?

Devastation vs. the infinity blade MH. Seems like I do HORRIBLE dps with the two-hander, and almost seems worth it to just use my regular set-up (Wicked Edge/2.6 S2 Axe) or switch the wicked edge for an infinity blade. The speed boost seems useful for running from Sanguinar to Telonicus, then to Kael to interrupt that potentially lethal first fireball, as well as running to eggs. On the other hand, I'd be running slower, but doing more consistent and reilable damage.

I have heard Stig say on multiple occassions he uses Devastation, thoughts?

Yes apparently you are correct, that different sources of haste stack multiplicatively and not additive, thanks for pointing that out.

This is also wrong, correct method is...

1) add together all currently active items which apply haste rating (in this case 325+31 = 356)

2) convert total haste rating to a speed multiplier. (here its 356/10.5 = 33.905 ->1.33905)

3) apply all speed multipliers which are active (here its 2.7/1.3/1.02/1.33905 = 1.52)

Your method was very nearly right here, but if you have quite a few seperate haste rating items then applying them seperately will give an answer further away from the truth. There is also no need for that number of brackets when dividing, order is not relevant so it can be all done in one.

I would like to ask something about the 1Str = 1CR equation:
Given that 1Str=1CR, 1% Crit would be equal to 22 Str or 44 AP.

As of my current gear, I have 1250 Ap, 28% Crit and 21.75% Hit; Black Planar Edge MH. My avarage hit would be 536 (MH only for simplicity reason, mainly for me) without migation, at least that's what my charscreen says.

In 100 hits, that would make 53600 + 28x536 = 68608 dmg, but I guess I would miss about 4 hits, so it's 66464.
That is if crit is unaccected by misses, in the way that even with 0 hit and 20% crit, out of 100 hits you would still hit 75 times and crit 20 times.

So
adding 1% crit would give me 53600 + 29x536 = 69144 or 67000 with the misses.
adding 44 AP would give me 44/14 = 3.142 DPS or ~8.4 dmg per hit.
This would result in 544.4 average hit and 54440 + 28x544.4 = 69683.2 or 67505.6 dmg after 100 hits.
That would mean a 500 dmg difference, which is almost 1%.

That is as far as my math skills reach, don't know if this is even worth mentioning, but it seemed strange to me that AP would help me more than crit, as I thought crit would be better. Maybe I just make a mistake and all that is obsolete; but as a result, I would stack AP over crit now.


I would like to ask another thing, as I have no idea how to calculate that:
I have Maulgar's Warhelm and the Desolation Helm. I wonder which one would be better if I had the critdmg meta gem in the desolation piece.

Are haste rating added up and then applied, while flurry\mongoose\bloodlust are applied as their own multiplier then?

I did a quick test with a twohander for a warrior today, and I ended up going from 3.60, flurried to 2.77 and then bloodlusted to 2.13.

I thought at first it would be 3.60/(1+0.3+0.3), but this didnt match the numbers so I did
3.60/1.3=~2.77
2.77/1.3=~2.13

Can we safely say that percentage-based haste buffs are added seperately from the haste rating buffs then?

So it'll end up as

(((((Weapon speed/flurry)/bloodlust)/mongoose)/mongoose)/haste rating + haste rating + ..) = new speed?

You can not use simplified single hand formulas to argue about these values, for example each crit also affects flurry and unleashed rage and also the offhand, while each additional strength (or 2 ap) also affects unleashed rage and the offhand. Additionally these values are for raidbuffed equipment, they do not necessarily apply to unbuffed equipment.

A few values the ideal equipment of my current model calculates:

normal 1377,98
+44 AP 1392.62
+22.1 CR 1392.11
+15.8 HR 1387.48
+10.5 HasteR 1386.98

increase:
+44 AP 14.64 DPS
+22.1 CR = 14.13 DPS
+15.8 HR = 9.5 DPS
+10.5 HasteR = 9 DPS

1 Str = 2 AP = 1 CR = 1.06 HitR = 0.75 HasteRating

HasteRating probably changes to certain degrees with the equipment, as for the other values you can easily ignore the +0.06 for the HitRating. For this equipment each Rating is of equal value, while HasteRating is superior.

I will add equivalence calculations to my model soonish.

Yes, anything with a rating is added together to give a total % haste that comes from rating. Anything that has a flat % effect is applied on its own as a multiplier.