Elemental, Shaman, Theorycraft, WoW

Elemental stats: Haste caps and relativity

This is the first post in a series on our primary DPS stats – which are Hit Rating, Spellpower, Haste Rating, Critical Strike Rating and Intellect.  Yes, they’re all yellow stats except Spellpower.  No, I’m not dealing with them in that order.  Yes, Zing already talked about all this in a much more concise post.  Let’s move on…

Let’s talk about haste!  I think across all the forums I read, haste is the number one topic of questions or the number one subject of well-intentioned misinformation in answers (I think number two is Chain Lightning use – something I also want to revisit soon).  It’s a stat that confuses and alarms some and that others just can’t get enough of.  The major issues that confuse people relate to how haste affects our rotation, the existence of otherwise of a “haste cap”, and the relative value of haste compared to other stats.

That’s a lot of ground to cover in a single post, so this will focus on the idea of “haste caps” and the concept of stat relativity.  While I’ll touch briefly on the effect of haste on our spell “rotation”, it’s not really the focus of this post and will probably be dealt with in more detail in the future as I continue the series on our primary stats.

So let’s get started!

Haste Caps

Is there a haste cap?

Simple answer: No.

Actual answer: Well… Yes, sort of.

There is basically no achievable point at which haste becomes a junk stat for Elemental, and similarly there is no real disruption in this basic stat value priority:

Hit > Spellpower > Haste > Crit > Intellect

If you’re looking for a simple answer, that’s it.  Spellpower is better than haste and haste is better than crit.  Nothing that follows will change the reliability of this basic rule.

The “soft cap”

Sometimes you’ll hear other players referring to the haste “soft cap”.  Oftentimes folks actually mean different things by this.  Here are some of the definitions I’ve heard:

  1. The point where Chain Lightning and Lava Burst hit 1 second cast times (50% haste)
  2. The point where Lightning Bolt hits 1 second cast under Heroism/Bloodlust (53.9% haste)
  3. The point where Lightning Bolt hits 1 second cast under normal conditions (100% haste)
  4. The point where you can fit precisely X number of Lightning Bolts between Lava Burst cooldowns, where X is an arbitrary number between 4 and 6

Let’s be clear about something: the last one there is utter nonsense.  I know I badly need to update the article in question, but I’ve talked about it before in the Waiting for Lava Burst post and since seen it proven over and over again: trying to fit a specific number of Lightning Bolts between Lava Burst casts by gearing to a specific haste level is always always always a BAD idea, and likewise fitting an “extra” LB between LvBs compared to what you were doing before is actually a pretty smooth DPS increase.

What about the second definition?  The trouble with this is that Bloodlust only lasts 40 seconds (and if were the one casting it, 39 seconds!) and you will spend the vast majority of your time not Bloodlusted.  This is obviously true if you consider trash fights (which are not irrelevant), but it’s also true if you consider boss encounters.  The following chart is a simple visual representation of the portion of time spent under haste conditions in a 5 minute boss fight:

Chart showing the portion of a 5-minute fight for which haste effects are active.

It’s easy to see just how much of a fight is not affected by any extra haste, so I think we can safely discard definition number 2.

Definition numbers 1 and 3 are actually basically the same idea but put differently.  The reason that the cast times of spells is “capped” is not because the spells stop casting any faster – they don’t – but because the Global Cooldown is limited to a minimum of 1 second.  That means that any spell you cast, whether it’s instant or has a cast time, will prevent you from casting any more spells for 1 second at least.

When you reach 50% haste, both Chain Lightning and Lava Burst take 1.00 seconds to cast:

1.5 [base cast time in seconds] / 1.5 [150% cast speed] = 1.00 [final cast time]

Thus beyond 50% haste, while the spells will continue to cast quicker, you still won’t actually be able to cast any more spells until the Global Cooldown (GCD) finishes!  Their effective cast time beyond 50% haste is 1.00 seconds.

This chart, based on Binkenstein’s old TTT article, shows how much of our time is spent casting Lava Burst or Flame Shock (which always has an effective cast time equal to the GCD) in a normal “rotation” compared to Lightning Bolt (which does not hit the GCD cap until 100% haste):

The amount of time spent on "GCD" spells versus Lightning Bolt for total haste values from 0% through to 100%.

Again, we can clearly see that we spend the vast majority of our time casting a spell which isn’t limited by the Global Cooldown hard cap.  Additionally it’s worth pointing out that even when they’re limited by the GCD in terms of effective cast time, Chain Lightning and Lava Burst continue to benefit slightly from haste because their cooldowns start sooner if they cast quicker!

Some people consider the point at which Lightning Bolt hits this GCD cap – which is 100% total haste – a “hard cap“.

2.0 [base cast time of LB] / 1.0 [desired cast time - the GCD cap] = 2.00 [200% cast speed, which is equal to 100% haste from rating, buffs etc.]

Though Flame Shock continues scaling with haste beyond this point I’m prepared to accept this definition, and thus I’ll adopt the 50% haste value as the “soft cap“.

Haste ratings for caps

There’s a great table in the Elemental TTT listing what haste ratings correspond to the haste soft cap of 50% under various conditions.  ZAP! also tells you exactly how much haste rating will give you a desired haste percentage on its Calculators tab – you can set it to 50% and then look along the table to find what combination of effects you’re interested in.  But here are some quick figures for reference.

50% with full raid buffs: 1269 haste rating

50% with only Wrath of Air: 1406 haste rating

Remember that these are just for information – there isn’t any real consequence for going over the soft cap, as we’ll see below.  Additionally the hard cap is really out of reach for normal players, even with full raid buffs:

100% with full raid buffs: 2785 haste rating

But you might scrape against it with Heroism if you’ve got high haste gear:

100% with Heroism/Bloodlust and full buffs: 1386 haste rating

If you twin Heroism with Elemental Mastery you’ll hit 100% haste even at very low gear levels:

100% with Heroism, Elemental Mastery and full buffs: 778 haste rating

The “soft cap” and stat values

Using entry level tier 10 stats in ZAP!, stat equivalency points (EP) look like this using a relatively low haste value:

Spellpower: 2.0

Crit rating: 1.4

Haste rating: 2.0

Now let’s push the haste up to 50.01% and check again:

Spellpower: 2.1

Crit rating: 1.4

Haste rating: 1.8

Haste’s value has decreased by about 10%, but it’s still better than critical strike rating and worse than spellpower.

What about the effect on our overall DPS of exceeding this “soft cap”?  Here’s another graph, also used in the TTT, showing how even with entry-level stats our DPS continues to scale with haste way above the 50% “soft cap”:

Graph illustrating the very linear scaling of haste and DPS. Note that this does account for the effects of EM and Heroism use with 2t10 uptime levels (see the previous chart above).

Bottom line: The so-called 50% “soft cap” has no immediate effect on our stat priority and is not something to be avoided.

The 100% “hard cap” is undesirable but it’s also unreachable under normal conditions! Making sure to use Elemental Mastery and Heroism/Bloodlust separately will mean that most shamans only scrape against the hard cap for 40 seconds of any given fight, and thus maximise the value of both temporary haste effects.

Stat Interdependence

(and exceptions to “The Rule”)

As you can clearly see from the EP values above, the value of our stats depends on how much of the others we have.  Think of it like this: the harder our spells are hitting – which is determined by our critical strike chance, our spellpower and raid buffs/debuffs – the more valuable it will be to cast them faster (which is caused by haste).  And the more spells we are casting in any given timeframe, the more valuable it will be if they hit harder (which is caused by spellpower/crit).

This means that the basic rule – while generally reliable – is not hard and fast and does have minor exceptions.

If you have a lot of haste but not much spellpower or crit, you may well find that haste works out as less valuable for you than critical strike rating for precisely this reason – haste just isn’t very valuable if it’s all you have.  Likewise if you have a lot of spellpower and crit but not much haste, the value of haste can skyrocket even above that of spellpower.

We generally don’t see these sorts of situations for people who are “normally” geared, and gemming Runed/Reckless, because the itemisation team does a reasonably good job of balancing the stats that are available to us.  However some players who manage to, say, get a pair of phat haste trinkets, a haste/mp5 weapon, a haste/mp5 shield, haste/spirit rings et cetera end up finding themselves with masses of haste and not much “punch” to their hasted spells.  Likewise others who’ve had to settle for crit/hit gear (for example) and can’t get their hands on any haste trinkets may find that their spells are simply casting too slowly to get the full benefit from their overall power.

This was generally more of a risk during Trial of the Crusader than it is now, but I hope it explains why sometimes you will see genuine exceptions to the Hit -> Spellpower -> Haste -> Crit rule.  Most times you find yourself in an exception situation it’s a very temporary confluence of gear and circumstances which will change as you get upgrades which balance your stats out.

This is also the reason why, as haste values get higher and higher, the value for haste rating and crit rating converge and eventually – and I really do mean eventually – crit can become a superior stat.  This doesn’t tend to happen until above 70% total haste, which is at least 1876 haste rating.  Considering that our BiS 25-man heroic gear list has only 1446 haste, it doesn’t seem like this should really be a concern!

Conclusion

So let’s review all the stuff above:

  • At 50% total haste, Chain Lightning, Lava Burst and our instant cast spells reach their minimum effective cast time.  This doesn’t have any serious effect on either our DPS or our stat priorities.
  • At 100% total haste, the only spell which keeps effectively scaling with haste is Flame Shock’s damage over time component – but this is unreachable under normal conditions and isn’t really a problem if we encounter it for 40 seconds of Bloodlust.
  • More haste always equals more DPS.  Always.  However…
  • Gearing for specific haste values equals less DPS than following normal gearing priorities.  This is true for most situations, and those for which it isn’t true are generally so narrow and specific as to not actually exist in reality.
  • Point-for-point, haste is always worse than spellpower but better than crit rating.  However…
  • As with every rule, there are exceptions to the above – which can only be determined through theorycraft and are so temporary and specific that they should generally not result in regemming (but may be an indicator that you should go for a more balanced set of gear).

So I hope that clarifies the whole issue of “haste caps” and how the value of haste both changes and stays the same.  The comments would be an appropriate place for questions, answers, disagreements, observations, corrections and photoshopped pictures of Will Smith.

About these ads

Discussion

15 thoughts on “Elemental stats: Haste caps and relativity

  1. Fabulous as always Charles! A good illustration of stat balancing might be that a lot of better geared shaman find FTW glyph a DPS increase precisely because they tend to gear for haste where possible, and that 2% crit makes up a big deficit , without needing to regear and lose haste to make up for it.

    Posted by pewter | April 26, 2010, 10:34 am
  2. What impact does our choice of racial and professions have on the stat priority? I seem to recall and older article which mentioned that a Troll Engineer was the sole exception to the stat priority. Their priority would have been Hit>SP>Crit>Haste. Would you mind clarifying?

    Thanks!

    Posted by Maker | April 26, 2010, 2:02 pm
    • Well remembered and a good point! I do recall making a comment like that offhand a while ago. I think it was based on some sim results someone had posted on EJ. Since then SimulationCraft has been extensively updated and ZAP! itself also includes Berserking and Hyperspeed options, and it seems they don’t affect DEP values very much – not enough to change what’s written above anyway.

      e.g. at 1269 haste, DEP for a draenei with my stats:
      SP 2.111 – CR 1.446 – HR 1.811
      And for a troll engineer with the same stats:
      SP 2.160 – CR 1.477 – HR 1.780

      Berserking would add another half-heroism-sized chunk to the uptime graph (20s) and Hyperspeed would add a slightly bigger chunk (12*5 = 60s) – but Hyperspeed is just an extra 340 haste rating (as opposed to haste %) so doesn’t have an extreme effect.

      Posted by Charles | April 26, 2010, 5:08 pm
  3. My biggest gripe with high haste is the suboptimal spell queuing due to the client-side GCD lock (as described in your “Lag, WoW and You” guide, last part). Especially during bloodlust. I agree of course that more haste is always good, but elemental is seriously held-back by the GCD, more so than any other class/spec.

    Posted by Spasi | April 26, 2010, 2:05 pm
    • This is one reason why I’m a bit ambivalent about haste as a stat in the first place. As well as the extra “penalty” from lag/reaction time etc., there’s the stress on your keyboard, fingers and even brain of having to manage all the spell casting baggage (queueing, spell selection, cooldowns, procs, etc) with such fast cast times. Elemental shamans have to perform an average close to one action per second on a standstill fight.

      I don’t know at what point “more spells = more fun” stops being true. I think I’m getting used to playing with higher haste, but I’m not sure I like it. And of course I’ve got rather lower haste values than many shamans because of using 10-man gear.

      Anyway. While this extra delay doesn’t affect the (relative) value of haste in theorycraft because so little of our “rotation” is GCD-limited, it can play a really big part in the Chain Lightning equation especially at higher gear levels. Plus it’s annoying.

      Posted by Charles | April 26, 2010, 5:20 pm
      • This is my major gripe as well, considering we also have better EM uptime now with 2pcT10.

        I also run without 2pcT9 or 4pcT10, so the duration on FS is close to the cd on lava burst, meaning I’m forced to clip FS alot more to make sure I get the crit which is pretty annoying.

        Posted by Ateve | April 26, 2010, 5:58 pm
  4. Those graphs seem familiar somehow…

    Excellent post as always. Some of this stuff I missed, and it should probably make it into the TTT (although the haste entry is getting rather huge).

    Posted by masanbol | April 26, 2010, 4:15 pm
    • Well, yes, having all those graphs done beforehand did make it a lot easier to get the post written quickly and easily when I finally got around to it :) – which for some reason was midnight last night.

      Posted by Charles | April 26, 2010, 5:27 pm
  5. Bravo. Great read my friend.

    Posted by Meemuu | April 27, 2010, 6:42 pm
  6. How much am I gimping myself by using the same set (more or less) for my Resto as for my Ele?

    What I mean by this is I haven’t saved up enough emblems yet to buy any tier 10 Elemental pieces – all of my Tier 10 is Resto (4 pc – head, shoulders, chest, gloves). I have two pieces of 264 (Chest, Shoulders).

    Now – obviously, the MP/5 gear won’t be of any particular interest. But every single socket (with two or three exceptions) is gemmed for haste.

    How much am I hurting my elemental DPS in trying to hit the Haste soft-cap?

    Posted by niceas | May 6, 2010, 5:32 pm
    • I should note that by soft cap I’m talking about the Resto soft cap – dropping the cast time on a 1.5 second cast to 1 second.

      Posted by niceas | May 6, 2010, 5:35 pm
    • You’ll be losing more DPS from not having proper elemental pieces than from stacking haste. The 4t10 set bonus for elemental is particularly powerful and not having that will hurt you a lot. However, even a haste-socketed resto tier 10 will probably be worth more DPS than a proper set of ilevel 232 elemental tier 9.

      If you intend to do any “serious” offspec DPS then you really need to aim for 4t10 in elemental. So long as your guild isn’t struggling for DPS when you’re in your offspec I wouldn’t worry too much about how you’ve gemmed – it’s better to excel in your resto spec and be a bit gimped in your elemental offspec than to be below average in both specs!

      Posted by Charles | May 7, 2010, 12:02 am

Trackbacks/Pingbacks

  1. Pingback: Tweets that mention Elemental stats: Haste caps and relativity « Planet of the Hats -- Topsy.com - April 29, 2010

  2. Pingback: Elemental stats: Haste and spell rotation « Planet of the Hats - May 28, 2010

  3. Pingback: Elemental stats: Crit « Planet of the Hats - July 24, 2010

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Follow

Get every new post delivered to your Inbox.

Join 27 other followers

%d bloggers like this: