This is part two of a series on stats and, by remarkable coincidence, also part two of a post on the haste stat. How very convenient. The previous post dealt in some depth with the ins and outs of haste as a stat, particularly the question of whether it was subject to “caps” or potential devaluation. This post will talk about the effect of haste on elemental’s “rotation”.
The goal of the elemental shaman is to do damage. Oh sure, we do other stuff too – we bring great buffs that we have to be careful to use correctly, we’re fantastic off/emergency healers, we can kite well and dispel stuff and purge other stuff and all sorts – but most of the time our main concern is to do as much damage as possible as quickly as possible.
This is why we develop the idea of a priority rotation of spells to use: we want to use the highest damage spells as often as possible. Higher damage spells have a higher priority. Actually that’s the elemental priority system right there – just use the highest damage spell available.
This wonderfully simple idea is unfortunately complicated a bit by two questions:
- Which spells are actually the highest damage? The game doesn’t really make it obvious.
- What about spells with cooldowns?
As today’s post is about haste’s effect on a rotation, the main question we’ll be looking at is that second one. We’ll talk about how the elemental rotation is founded on the Lava Burst cooldown, look at some misconceptions that are floating around about what this means, and then tidy it up into a few simple, pithy principles.
Or, that’s the theory… in reality I’m probably going to waffle on for three pages about something entirely unrelated before getting distracted by those adorable handfish again.
Time as a resource
Right, so, DPS – that is, “damage per second” – is generally the thing we try to maximise as a damage spec. As casters, we only do damage when we cast a spell to do damage. Rogues use an energy bar and warriors use a rage bar as the limiting factor in what abilities they can cast, but we are not limited by our mana bar; rather, our primary DPS resource is time. Every second that passes in-game is a second we have to choose what spell to be casting. If we’re not casting, we’re not doing damage. This is where the old mage ABC comes from: Always Be Casting.
To do the best damage we can, we want to use our time as effectively as possible, which means casting the spells which do as much damage as possible for the time they take to cast. This concept is referred to variously as damage per cast time (DPCT), damage per execute time (DPET) or damage per seconds casting (DPSC). Working out DPSC is really easy – you just divide the final damage done by a spell by its cast time, or – in the case of instant cast spells – the time you spend unable to cast because of the Global Cooldown triggered.
DPSC vs DPS
If you open ZAP! and click on the Calculators tab, you’ll find a wee box headed “Damage per cast time”, and a list of spells and fire totems. This list is updated for your stats and set bonuses (assuming you entered them!) and lets you compare the ‘efficiency’ of our spells’ damage on a single target. (AoE is a special case that requires a bit more thought, but before we can talk about AoE it’s important to first have a solid grasp of the single-target concepts.)
You’ll notice that the highest DPSC comes from our two main fire totems – Fire Elemental and Searing. Then, depending on set bonuses and glyphs, you may find that Flame Shock has the next highest DPSC, or it might be Lava Burst. Below that will be Chain Lightning, and below that will be Lightning Bolt. Frost Shock should be next on the list and at the very bottom will be Fire Nova.
Searing Totem (and Fire Elemental Totem) has such a high DPSC value because it only takes a second to cast but lasts for 60 (or 120) seconds. This makes it a very efficient source of damage, but it’d be pointless to try to ‘spam’ it – that is, chain-cast it repeatedly again and again and again – because the actual DPS it does is quite low. Its value lies in its efficiency – you spend a tiny amount of time casting it but it continues to do damage for ages afterwards.
Flame Shock is the same idea, because its high DPSC comes from the fact that you only spend a single global cooldown casting it, but it continues to do damage for quite a while as you cast other spells. Flame Shock is also important because it causes our Lava Burst spell to do more damage when it’s up.
So while these are our most efficient damage spells, they’re also not ‘spammable’ – so our goal is to keep them active as much as possible while casting them as little as possible. In the meantime we weave in other spells to do damage while these tick away.
The other spells on the list are all direct-damage spells which do their damage immediately you cast them, which means that their DPSC is the same as their DPS. The highest direct-damage spell in our arsenal by a significant margin is Lava Burst. So obviously we want to cast Lava Burst as much as possible… but we can’t just spam it either, because it’s got a cooldown. So while we wait for Lava Burst cooldowns we have to cast other stuff to fill in the time.
That’s the elemental priority system in a nutshell. A very large, boring nutshell.
The thing is, Lava Burst does so much damage that we want to really make sure we do cast it as soon as possible. But we also remember that time not spent casting is time not spend doing damage. The tension between these two principles is what gives rise to two common misconceptions in elemental circles:
Misconception 1: If the time left on Lava Burst’s cooldown is less than the cast time of Lightning Bolt (or Chain Lightning), it’s better just to wait for it to come off cooldown and not cast anything in the meantime.
Misconception 2: Because not casting = bad, we should try to get a precise amount of haste to allow us to perfectly time a certain number of Lightning Bolts (and/or Chain Lightnings) between our Lava Burst cooldowns without any cast time overlaps. E.g. 8 second cooldown divided by 5 Lightning Bolts would require a 1.6 second cast time, so I need as much haste as will give me 1.6 second Lightning Bolts. If I’m lower than that haste figure I desperately need to get more, and if I’m higher I desperately need to get less.
The only reason we know that these are “misconceptions” is because if we take data from the game and work it through mathematically (whether formulaically or through simulation) we can quantify all the messy, complicated inter-relationships and end up with a clear picture.
Lava Burst’s cooldown
So the elemental rotation is based around this priority system which focuses on keeping Flame Shock up and casting Lava Burst every time it’s available. The more often Lava Burst is cast, the higher your DPS will be. If you only cast it every 20 seconds, your DPS will really suffer. If you manage to begin casting it the moment it comes off cooldown, your DPS will benefit. But only provided you are always casting something.
Ages and ages ago I evaluated whether it was ever a DPS increase to not cast anything and instead wait for Lava Burst’s cooldown to expire. The answer was a qualified yes, which actually means no.
Consider this diagram (click for the full-sized version). It’s a timeline with one-second intervals marked, beginning with a Lava Burst cast and showing when Lava Burst’s cooldown starts and ends. Ideally we’d want to cast Lava Burst again at the second red arrow.
Now consider this diagram. It’s the same one, except this time “filler” Lightning Bolts have been added. You see that the Lava Burst cooldown expires while LB is being cast, so we can’t cast Lava Burst until we finish casting the Lightning Bolt.
If we want to cast Lava Burst sooner, we have two options. Either we wait and do nothing for a while as in this diagram:
Or we can try to better ‘fit’ the time by using a spell with a shorter cast time than Lightning Bolt. The next highest DPSC spell we have is Chain Lightning, which has a cast time equal to the Global Cooldown. Applying that to the diagram results in the following:
Now obviously in this simple and not entirely accurate diagram it’s this last option which is best. But what if the gap is smaller – maybe just a few fractions of a second?
There are situations where, when the cooldown remaining on Lava Burst is only around 0.2 seconds or less, it is mathematically very slightly higher DPS (like, single/double figures) to not fill the time with another spell and push back that Lava Burst cast. The trouble is that doing this puts you adrift of the spell queue mechanism and introduces a sudden spike of human reaction time into the equation. In other words, that 0.2 seconds delay suddenly becomes a reaction time + network latency delay, which for most people will be about 0.3 seconds in addition to the actual gap. So the 0.2 second delay becomes a 0.5 second delay, and the ~10 DPS gain becomes a ~100 DPS loss.
This is why it’s basically always better to follow the ABC rule: always be casting!
But this post is meant to be about haste, right? Right! So here’s the skinny. In formulation it’s possible to chart DPS against haste rating, and see how our damage output changes as we add a little more haste, a little more haste, a little more haste, and so forth. Well, what happens when we go from a situation like this, where Lava Burst is coming off cooldown immediately our last filler spell finishes:
…to a situation where we have just a little more haste and a nasty gap has appeared?
The answer is actually very simple. Because of the ABC – always be casting – rule, and because of what we observed above about lag, all we do is add in an extra “filler” spell.
Filler spells & Chain Lightning
Now if we just add in another Lightning Bolt, we’d see a tiny dip in the DPS graph of a few DPS points as we suddenly delay Lava Burst compared to what it was like before. The dip is only small because the effect of casting all your spells faster from the extra haste offsets the loss from delaying Lava Burst. But if we add a Chain Lightning – or even two! – our DPS continues to scale pretty smoothly upwards.
This is ultimately the reason why we do or do not use Chain Lightning in a single-target rotation as elemental. The spell’s DPSC is only a little bit higher than Lightning Bolt’s and it has complications with the spell queue at high haste, but it can be incredibly valuable because of its ability to shorten the gap between Lava Burst casts while obeying the ABC rule.
Now the situation I described a few paragraphs ago is actually almost entirely impossible in real gameplay, because it’s very very rare to gain only a couple of points of haste at a time and it’s very very unlikely that if you do gain just a couple of points of haste it’ll cause you to need to modify your “rotation”. Normally when we find ourselves able to fit “extra” filler spells it’s after gaining a lot of haste, and the sheer amount of it makes it a big DPS increase even if we’re now delaying our Lava Burst casts by a bit more than previously. Additionally, it’s pretty astonishingly rare to get just a tiny bit more haste without also picking up some spellpower and/or crit. This is why it’s a bit misleading to see haste vs DPS graphs with tiny dips at certain haste ratings – so called ‘black spots’ – because those dips are just an artefact of the formulation process. They don’t really exist in reality.
So let’s review what all this means:
- Always Be Casting – if you’re not casting, you’re not doing damage. (Which, so we’re clear, is Bad™.)
- It’s never a DPS increase to wait for Lava Burst’s cooldown instead of casting a spell.
- Minimizing the time between Lava Burst by intelligent use of filler spells – Lightning Bolt and Chain Lightning – will maximise your DPS. The potential gain is in the region of 50-200 DPS.
- Chain Lightning is most valuable when it helps minimize this gap, and can actually be a DPS loss if it increases the gap.
- Always remember that the elemental “rotation” is really a dynamic priority system – you have to adjust your spellcasting to the situation at hand. Movement, lag, spell pushback, casting Flame Shock, dropping totems, and various fight mechanics all prevent us from simply counting out filler spells after each Lava cast.
- Our instant cast spells – notably Frost Shock and Fire Nova – are lower DPS on a single target than both Lightning Bolt and Chain Lightning and should only be used while moving. They’re not helpful to reduce the time between Lava Bursts because their effective cast time is identical to Chain Lightning’s.
Above all, what this means is that getting more haste is always always always a DPS increase for your character, even if it requires you to change your rotation a bit.
And that’s it for haste as a stat!