The goal of this section is to study the impact of passive haste on a warlock dps.
For simplicity's sake, it is oriented toward a shadowbolt-spamming warlock using only Shadow Bolts and Connexions.
Standard considerations
At first we could consider that the dps gained from passive haste would be directly proportionnal from the amount of haste used. The only limitation would be the "haste cap", the point when your main nuke's cast time would be reduced to the Global Cooldown.
However the main nukes of a warlock are known to be pretty mana-inefficients, thus forcing said warlock to make a massive use of connexion. The problem is that connexion is instant, and thus already "haste-caped".
Basic assumptions
As most of the problem here is related to the time lost during "mana regeneration", we can't assume a "mana free" shadowbolts
-
spam. However, we will assume an "infinite" health pool, considering hat the life lose via Life Tap is replenished throught HoTs and such.
We will assume that the caster is hit-capped and we won't -for now- consider its critical strike rating. So when refering to a shadow bolt "damage", we will consider that it is its "average" damage.
To improve readability, the formulas used in those pages are using the following conventions:
- manaCost is the cost of the spell, talents and buffs included. For a shadow bolt (rank 11), manaCost=420.
- castTime is the real cast time of the spell, talents and buffs included. For a shadow bolt (rank 11) and using Bane
-
, castTime=2.5.
- manaGained is the is the mana gained from a life tap. For Life Tap (rank 7), manaGained=580 + 0.8*dmg.
- tapCastTime is the is "effective" cast time of a life tap. It is equal to the Global Cooldown, thus tapCastTime=1.5 sec.
Standard fight
A standard fight could be decomposed in two phases. In the first one, the warlock will be able to chain-cast its shadow bolts without mana considerations.
During the first phase, a shadowbolt's dps is equal to:
During the second phase however, the warlock will have to pay a "health and time" cost instead of a real mana cost. To calculate its "real dps", we will then have to include this new "time cost" into the formula:
dps = damage / (castTime + tapTime)
The tapTime can be defined as follow: It is the fraction of a Life Tap's cast time used to cast one shadow bolt.
tapTime = manaCost / manaGained * tapCastTime
Let's throw Haste in there. Since version 2.2, you get 1% haste for 15.7 haste rating. The dps of a shadowbolt in phase 1 will then be:
dps = damage / hastedCastTime
where
hastedCastTime = castTime / (1 + haste / 1570)
Note that this won't help you cast more bolts in phase 1, whis will just reduce the duration of this phase.
In phase 2 however, the gain is lowered, since tapTime is not reduced by the haste rating. The new dps will be:
dps = damage / (hastedCastTime + tapTime)
.
Phase 2 haste-to-dmg conversion
We will have to expand the formula of the previous chapter by remplacing everything inside it and using "real" damage instead of an average value.
For consistency sake, here are some new conventions:
- baseDmg is the (average) base damage of the spell, critical strikes non included. For a shadow bolt (rank 11), baseDmg=572.
- dmg is the spell damage used by the character, spell, buffs and talents included (like Demonic knowledge
-
or Fel armor
-
).
- crit is the spell critical rating of the character, buffs and talents included. (22.1 crit = 1% crit chance)
- untalentedCastTime is the base cast time of the spell. For a shadow bolt (rank 11), untalentedCastTime=3.
Using those values, we can define the dps of an unhasted talented hit-capped shadowbolt as:
dps = (baseDmg + dmg*untalentedCastTime / 3.5) * (1 + crit/2210) / castTime
The formula for an hasted talented shadowbolt becomes:
dps = (baseDmg + dmg*untalentedCastTime / 3.5) * (1 + crit/2210) / (hastedCastTime + tapTime)
= (baseDmg + dmg*untalentedCastTime / 3.5) * (1 + crit/2210) / (castTime / (1 + haste / 1570) + manaCost / manaGained * tapCastTime)
So, if we want the +dmg equivalent of some haste, we can compare a "damage-amplified" bolt and an "haste-amplified" one:
Where
dmg' is the spell damage needed to gain the same dps as
haste.
Given this equivalence, we can express the haste as:
haste = (castTime / (((baseDmg+dmg*untalentedCastTime/3.5)*(castTime+manaCost*1.5/(580+0.8*(dmg+dmg')))/(baseDmg + (dmg+dmg')*untalentedCastTime/3.5))-(1.5*manaCost/(580+0.8*dmg))) -1)*1570
Disclaimer
I'm particulary bad at simplifying equations. The haste = f(dmg') above could be wrong.
If you want to peek at its development, here it is: sheet1, sheet2.
If you find any problem in it, please contact me.
Results
The following table can be used to search the point where haste is better than damage:
| haste = f(dmg), with arbitrary base spell damage: |
| mana cost |
420 |
|
|
|
|
|
|
|
|
|
| mana gained from life tap |
580 |
|
| untalented cast time |
3 |
|
| cast time |
2.5 |
|
| base damage |
572 |
|
|
| damage equivalent \ base spell
damage |
900 |
1000 |
1100 |
1200 |
1300 |
1350 |
1375 |
1400 |
1500 |
1600 |
| 10 |
13.85 |
12.82 |
11.92 |
11.13 |
10.44 |
10.12 |
9.97 |
9.82 |
9.28 |
8.78 |
| 20 |
27.75 |
25.67 |
23.86 |
22.29 |
20.90 |
20.26 |
19.96 |
19.67 |
18.57 |
17.58 |
| 30 |
41.69 |
38.55 |
35.84 |
33.46 |
31.38 |
30.42 |
29.97 |
29.52 |
27.87 |
26.39 |
| 40 |
55.68 |
51.48 |
47.84 |
44.67 |
41.88 |
40.60 |
39.99 |
39.40 |
37.19 |
35.21 |
| 50 |
69.71 |
64.44 |
59.88 |
55.90 |
52.40 |
50.80 |
50.04 |
49.30 |
46.53 |
44.05 |
| 60 |
83.79 |
77.44 |
71.94 |
67.15 |
62.94 |
61.02 |
60.10 |
59.21 |
55.88 |
52.90 |
| 70 |
97.91 |
90.47 |
84.04 |
78.43 |
73.50 |
71.25 |
70.18 |
69.14 |
65.25 |
61.76 |
| 80 |
112.08 |
103.54 |
96.16 |
89.73 |
84.08 |
81.51 |
80.28 |
79.08 |
74.63 |
70.63 |
| 90 |
126.29 |
116.64 |
108.32 |
101.06 |
94.69 |
91.78 |
90.40 |
89.05 |
84.02 |
79.52 |
| 100 |
140.55 |
129.79 |
120.50 |
112.42 |
105.31 |
102.08 |
100.53 |
99.03 |
93.43 |
88.42 |
| 110 |
154.85 |
142.97 |
132.72 |
123.79 |
115.96 |
112.39 |
110.69 |
109.03 |
102.86 |
97.33 |
|
|
|
|
|
|
|
|
|
|
|
|
It seems that the borderline where haste begins to be better than pure damage is around +1400 damage.
Note that all raid buffs are included in those 1400 damages, so it's not "that" high.
Item Budget
To compare two items, we still need to take something into account: one point of spell haste rating don't have the same "cost" as one point of spell damage.
As I don't know the cost of spell haste, I'll evaluate it using two items: [Wristbands of Divine Influence]
-
and [Bracers of Nimble Thought]
-
. Those are convenients because they share the same iLevel, the same item slot, and don't have gem slots.
The [Bracers of Nimble Thought]
-
gains 28 spell haste rating, 34 spell damage and 3 stamina over the [Wristbands of Divine Influence]
-
, but loose 1 intellect, 28 spirit and 62 healing.
Assuming the numbers provided by WoWWiki are correct, those 28 spell haste rating are worth 26.139 stat points
1 + 28 + 0.455*62 - (3*0.667 + 34*0.855)
. So, the StatMod of spell haste rating equals
0.934
26.139 / 28
. This means that spell haste is 9.2% more expensive than spell damage.
Converter
As the conversion formula is absolutely atrocious, here is a small javascript tool to convert spell damage to haste rating:
| Damage to haste converter: |
|
|
Links