Hello

I often asked myself in the last time, how would this change with Lava Burst end up with haste and my rotation. Knowing the fact, that the global cooldown scales with haste, but not the spell cooldowns, I got the feeling that the ele rotation would become a pitty.
My little study shows why:


-=Basics=-
Lava Burst, Chain Lightning and Flame Shock have all given cooldowns which has to be "filled with spells". E.G: On 0 spellhaste there would fit 4 Ligthning Bolts in a Lava Burst cooldown - with more haste there would be a gap between casttime and end of cooldown. So what we do? Wait for cooldown, what would make haste nearly senseless, or do we put in another LB? What about when there are more cooldowns working? As during a Lava Burst and a Chainlighting cooldown?
This is what i adress here

-=Calculation Basics=-
As i consider only the Haste and the Rotation, i took average spell data from simucrafts latest calculation.
http://code.google.com/p/simulationcraft/wiki/SampleOutput

I put this into a Excel Spreadsheet which can be found here:
http://code.google.com/p/dpsonhasteandrotation/downloads/list

The green marked cells are the inputs for the fully calculated (inlcuding, talent, crit, gear) average damage per spell

In the dps calculation of a rotation are cooldowns considered. Meaning that a rotation between two Lava Burst are caped in the minimum length of 8sec.
No lags considered.
Flame shock Glyphe assumed
No gimpness considered :)


-=Result=-
The following picture shows DPS graphs for several rotations and how they scale with spellhasterating

http://img201.imageshack.us/my.php?image=dpshasterota3zz1.gif
http://img179.imageshack.us/my.php?image=dpshasterota4dd2.gif

As you can see, several rotation scale partially weaker than others or are earlier caped than others. This is really interesting to see that nearly all rotations have at certain spellhaste level a big leak of hastescaling. Changeing the rotation at this point is necessarly and unavoidable.
Changing the rotation is also resulating in a leak - how big this leak is, is deping on how smooth you change your rotation

2) CL 2xLB
This graphs shows the problem the best: there is a good scaling with haste at the beginning, but as soon as the 2 LB are faster than the cooldown of the CL (at ~15% spellhaste), the LB scaling is finished. After that, only the CL castingspeed scales to a given level where the casttime hits its natural border (at ~51% spellhaste) of min global cooldown of 1sec.

3) CL 3xLB
In principle the same, but the first kink of the scaling is above 50% spellhaste - which is in my opinion unreached anyway

Looking at this two rotation we see, that until a spellhaste of 17%, rotation 1 is the better, and with more spellhaste rotation 2.

When you now follow the max dps, you should change yout rotation every couple spellhaste percents


BUT THIS I PURE THEORY!

-=Recommendation and Usability=-
Looking at the fact that the difference from one to the other rotation is fairly small - like less than 5% i think you should give a *@@# on superduper complicated rotations like rotation 13 or 14.
I found myseld in the last days really focusing on the proper rotation than other import stuff at a boss fight. So for me its like: the easier the better and at the end: better dps because of no strugling in rotation.

Therefore i recommend to go for rotatins like nr 6-10 which has the form of

FS LvB CL x*LB LvB CL y*LB - where x and y has to be choosen according your hasterating

With this rotation changes ou nearly have constant haste rating and the leak between rotationchanges are pretty small.x and y can be the same but also x can be y+1.


Here the pictue from above but only with the recommendated rotations in. You can clearly see on which spellhaste level the change from one rotation to the other should be

http://img523.imageshack.us/my.php?image=dpshasterota5jx2.gif


WARNING: The exact point might move a little depending on YOUR real gearstats and probably also on lag - but this gives a good picture how it works.

If you don't wanna make a superduper rotation analysis considering lag und equip in every detail I would say: As soon as you start feeling a stuck in the spellflow because cooldown hasn't finished yet, the next time you get more haste on your equip you should also change rotation :) - i know strange descpription :P

Doing rotation on the basis that CL is also allways on its cooldown seems to be the best, because obviously CL is a better DPS spell than LB, but as the graphs shown, this is only valid at very low spellhaste level. after a certain spellhaste level, you really get into trouble to organize the LvB and CL cooldowns without any gabs. filling that gabs up with something like a glyphed Earthshock can maybe fix taht problem, but honestly: noone is able to keep up such a superduper timing rotatio and at the end he will struggle with it and endup with a lower DPS than a guys who does the simple way

During bloodlust/heroism phase: pure LvB and LB spam seems to be the best, because even there, the changing the key on keyboard for switching the spell every 1-1.5 seconds for 3 different keys is not that easy :p

-=Wanna do your own Rotation Calc?=-
- Download the spreadsheet from the link above.
- Calculate your own average spelldmg data for the spells with a theorycraftmodel of your joice or your experience
- fill in this data

If you wanna changes rotations: try to find out what i did. to complicated to explain, expecially when CL cooldown goes longer than the LvB cooldown :)

Please discuss

Greetings
Mitrandir

P.S.: Please don't blame my english - i'm not a native speaker ;)

This r good thread.

Realy found this usefull as i altso had a bit of trubble squeezing in fs and lvb in my rotation. And the graps are realy nice for gettin the best rotation depending on gear.'

will test it out tonight 10/10




Eloxy

OP: In essence, what you're dealing with here is the haste drop problem discussed about a month ago on EJ (http://elitistjerks.com/f79/t27016-elemental_wotlk_talents_discussion/p59/#post949675). Unlike what you suggest, though, there is no perfect solution to this problem (the reason is that you use very raw data, which leads you to inaccurate results). In essence, you need to keep Flame Shock up, cast LvB as soon as possible, followed by CL and LB as fillers. The problem is, as you stack haste, you have to delay LvB cast due to its cooldown nature, which effectively drops the dps. An alternative is, as you said, waiting, but this again has no effect on dps.

Specifically, the areas where dps drop will occur are as stated here:
http://elitistjerks.com/f79/t27016-elemental_wotlk_talents_discussion/p60/#post952700

You'll simply want to avoid these haste areas (+ latency) and you should be in the clear. Downside: T7 gear leaves you in the middle of the first drop area (slightly above 19%). Since you'll need to fill up most gem slots with hit gems, this leaves nearly no maneuvering space to solve the problem.

Basically - we're dealing with another problem of a heavily broken spec, simply because we are a class that needs to cast a short cooldown ability in its regular rotation. There's been like a billion solutions proposed, but tbh, I don't consider it that massive an issue as long as we're facing far greater problems. So, I'm expecting this to be the first problem solved by Blizzard sometime in 2010 ;)

There's something I don't get. Let's say, rotation 8:

You're doing 5 LBs after a CL, and doing a LvB after that. Wouldn't it be more DPS if you did 3 LBs, a CL, another LB and then LvB?

This makes me got damn mad. Gahhh"#!"#%/(! is the devs some kind of complete tards or what?! All news that comes on mmo-champion etc from blue post is always about Druid, Paladins and Warlocks but like never on Shamans! Will our class ever be fixed!? The ~1k DPS that Shamans is behind the least worst caster dps with the premade 80 chars dont mean anything or what? Also we got like 0 coeffients wich will make us only fall more and more after other casters. Getting so got damn frustrated of the thought that us Shamans gets completely ignored...

Sorry to steal your thread but just had to write me off something

@Trivial
This rotation style in considered in rotation 9-12 and at the end 11 and 12 (which should be pretty cooldown optimised) are the best rotations for certain haste ratings, but not for all levels of it.
as you see with rotatation 11 you hit pretty fast the scaling cap and other rotations are better. in rotation 12 where we have one LB more after each LB, it scales much better, but with a lower dps level on 0 haste. above 650 haste is the best.
[b]BUT[/but]
in the realistic hasterating areas, the more simple rotions with just one CL after a LvB are better because of avoidance of these cooldownproblems
AND
not to forget: its much simpler to go for that instead of this pretty complicated rotation with 2 CL in each LvB cooldown

@Nathaira
Thanks for the links, i read to less the boards to follow all up ;)
But it seems that the final conclusions are pretty the same: Fs Lvb CL x*Lb LvBCL y*LB are the best rotations to go and depending on the haste you have y and y are more or less.
and there are haste leaks at the points where you should switch the rotation - somewhat like: you can spend 40haste for nearly nothing

updated the inital posting

changes:
- some new rotations
- rotations renumbered
- some recommendations
- changed input data from hasterating to haste%
- corrected calc error in rotation 13,14

sorry that some of the post above doens't make much sense now

What is really annoying me about this is just how hard it makes the theorycrafting. Actually, what is really annoying me is how bad it might make our DPS, but anyway.

I am at the moment using a rotation of

FlS, LvB, CL, LBx2, CL, LvB, LBx2, CL, LB

Effectively I am sneaking in 2 Chain Lightnings between the 1st and 2nd LvB, just to max DPS a bit more. Otherwise, its pretty much like the advised rotation the OP gave.

The problem is having read the whole thread on EJ, I am nowhere nearer knowing the ideal way forward. Is it better to use this rotation, or drop a cast, and have a 0.3sec wait until LvB? Right now I am just packing in as many casts as I can, as it feels wrong to stand and wait.

Its not beyond me to do the maths of it all myself, but I am finding it unlikely hard and fast rules will appear.

I guess what I am missing most right now is some way to compare Haste to other stats. I have always calculated my own stat weights (which are very close to those given in the old "thinktank" post on EJ), and used them to decide on gear choices. With all the problems haste gives us though, I am truly puzzled as to how to value it.

@Mmootimus
i updated the recommendations sections up to your questions.

i suggest you to really give up your trys to bring 2 CLs in a LvB cooldown.
your rotatio nreflects my rotation 13. as the graphs shows its the best at 0 haste, but scaling badly because of a to complex cooldown puzzle and you find yourself fast enough only waiting on cooldowns

keeping up this kind of rotation is also very hard focusing work.

i hope you don't come up with the idea of using castsequenz, because this is a no-go :P
using castseuquenz is like shooting in your feet and try running a marathon. it just produces to much additional lag becuase of additional client/server communication between spells

---

Q u o t e:
@Mmootimus
i updated the recommendations sections up to your questions.

i suggest you to really give up your trys to bring 2 CLs in a LvB cooldown.
your rotatio nreflects my rotation 13. as the graphs shows its the best at 0 haste, but scaling badly because of a to complex cooldown puzzle and you find yourself fast enough only waiting on cooldowns

keeping up this kind of rotation is also very hard focusing work.

i hope you don't come up with the idea of using castsequenz, because this is a no-go :P
using castseuquenz is like shooting in your feet and try running a marathon. it just produces to much additional lag becuase of additional client/server communication between spells

---

Yes, I have been using castsequence

I was worried castsequence was giving me heavy lag, but with no working version of laginfo since 3.0, and target dummies which are messed up by CL jumps, I haven't done any testing specifically on lag.

(Btw, seriously Blizz, please give us some spread out practice dummies, just in one city or something).

Anyway if its true that /castsequence blows, then I doubt I will be doing the rotation I gave above. I may just have to drop CL entirely in raids regardless, as my new guild doesn't seem too keen on S Priests, so replenishment may be limited... however I would still like to work out some answers to the questions here.

As I said before the main one is simply how to turn haste into a spellpower equivalent value now there are so many rotation isssues.

@ castsequenz:
try the follwoing
try this macro
/castsequenz reset=5 chain lightning, lightning bolt, lightning bolt
for 1 minute
and try to make the same rotation manually with standard spell macros for 1minute and count the effectivly castet spells at the end
should maby also work with pure Lb spam

@ spellpower equivalent
you can't put haste into a spellpower equivalent.
because haste works as a FACTOR to the spells dmg and spellpower is a constant addition to the spells dmg.
roughly and simply said:
on a 2k LB 1%haste gives 20 spellpower - on a 1.5k LB 1%haste is only 15 spellpower

---

Q u o t e:
(Btw, seriously Blizz, please give us some spread out practice dummies, just in one city or something).

---

/SIGN
of blizz test on that dummys in OG - noone should wonder why they think shamy dps is ok - because CL hits everytimes 3 targets ;)

---

Q u o t e:


@ spellpower equivalent
you can't put haste into a spellpower equivalent.
because haste works as a FACTOR to the spells dmg and spellpower is a constant addition to the spells dmg.
roughly and simply said:
on a 2k LB 1%haste gives 20 spellpower - on a 1.5k LB 1%haste is only 15 spellpower

---

Errr... you can, although only for a given gear level. Obviously as your gear changes, so will the value of haste.

Haste isn't something special in this regard. 1% additional crit will be worth more additional DPS if you have 2,000 spellpower than if you have 1,000 spellpower (as the crits will be bigger). Just like Haste, the spellpower equivalent value of crit changes as your gear does.

This is the problem. Its quite easy to see what effect an extra 1% haste has for pure LB spam. But put it into a rotation though - a rotation that changes dependent on haste - and the theorycraft just got a lot harder.


Was already planning a similar castsequence test... will advise how it goes.

---

Q u o t e:
Q u o t e:
(Btw, seriously Blizz, please give us some spread out practice dummies, just in one city or something).

---

---

Q u o t e:
/SIGN
of blizz test on that dummys in OG - noone should wonder why they think shamy dps is ok - because CL hits everytimes 3 targets ;)

---

Some boss-level dummies would be nice too.

/target Half-assed implementation
/sigh

Extremely unscientific test results:

LB spam with /castsequence macro - 78 casts
LB spam without macro (ie. button spamming) - 91 casts

Test time was 135 secs. No, I haven't pulled out that LO procs (thats why its unscientific).

Anyway, I have at least proved to myself what I suspect most people knew - castsequence sucks.