A Question of Haste, Part Duex

Way way back (Patch 2.4 even) I looked at haste and how useful it was for Elemental rotations at the time. Today I plan on doing the same sort of thing, but for the post 3.1 Elemental (and this may be applicable to other classes as well).

There's a fair bit of concern out there about how there are "bad haste values". I'll admit part of this is my fault, due to early work on modelling how the rotation would scale with haste (I was forgetting that the dot timer would start at the start of the Flame Shock GCD, rather than at the end), but there are also those that advocate fixed rotations (ie: FS, LvB, 5xLB, LvB, 4xLB, and repeat). Certainly these sort of viewpoints have some value, but they are also some incorrect assumptions being made.

The biggest assumption is that since Lava Burst has a higher Damage:Cast ratio than Lightning Bolt, it's important to get that cast off as soon as possible. This then leads to the conclusion that haste values that result in a whole number of casts in the 8 second cooldown (ie: 8/1.6 = 5) is best. Unfortunately this is not correct.

The comparison needs to be made between rotation values, not individual values. This essentially boils down to comparing rotation X over time Y with rotation X+1Lb over time Y+1Lb-Z, where Z is the time that would have been wasted inside the 8 second LvB cooldown. This boils down to Damage over (Cast time minus Wasted Time) vs the normal rotation DPS. By using this sort of comparison, rather than an cooldown-centric viewpoint, we can see there are a few haste ranges (1-3% usually) where it's better to wait for Lava Burst than cast another Lightning Bolt, but the definition of "better" comes down to a sub 60 dps value.

The following graph is a comparison between two "ideal" rotations. The first, Base, is casting the first available spell without consideration for when the Lava Burst cooldown is finishing, while the "Logic" rotation is using the previously discussed method for choosing whether to cast an additional Lightning Bolt or not. As you can see, the graph has been adjusted to show a 1,500 dps range from 5,000 to 6,500 on Y, with the haste rating on X (assuming that you have Wrath of Air and the moonkin/ret haste aura). The difference between the two rotation methods ranges between 0 and 54 dps (that's 0.99% at most). Note that this model uses both of the T8 set bonuses. (The spreadsheet can be viewed here)



This minor difference, combined with the fact that after the hit cap it is better to stack spellpower than anything else, means that the haste rating limits/targets that are prevalently recommended can effectively be ignored. If haste becomes more valuable later on, than this may have to be revisited, but at the moment you do not need to worry about how fast your Lightning Bolts cast, aside from the obvious Faster = Better.