4.0 Rotation (pre cata)

[daiceman]

Most of the other threads about this have very old information about possible skills we will use in the expansion, I was hoping we could get a thread going talking about our rotation in 4.0 (pre cataclsym) and what we will do for the month or so we are 80 with cataclysm skills

Some preliminary observations from the PTR:

Hammer of the Righteous hits harder then Crusader Strike single Target
Hammer of the Righteous does not proc seal of truth, crusader strike does, even with that, hammer hits harder
Hammer and Crusader share a 4.5 second cooldown and are our only holy power generators
Consecration is 2416 mana untalented (483 talented) and has a 30 second cool down
Holy Wrath hits all targets, but only stuns demons and undead (dragonkin and elementals may be glyphed in)
Avengers Shield hits for 4.8-5k and has a 15 second cooldown

[Meloree]

From my limited testing on PTR, it seems like:

CS-J-X-CS-X-X is the base rotation, and X gets filled with ShoR > AS > Holy Wrath > Consecration.

If Grand Crusader procs, or AS comes off cooldown during a phase that would otherwise have been CS-J-ShoR, and you recognize it fast enough (on PTR, GC procs seemed to happen at least 1s after the CS was hit, so it was pretty tough to recognize the proc and react in time), then CS-AS-ShoR-CS-J-X seems to work. We have enough mana to push back Judgement occassionally.

I am wondering (just now, haven't tested) if it's worth hard-casting Exorcism during any of the otherwise empty X spots. It's a ton of mana, but it hits pretty hard, especially with Vengeance up. It might not be possible to do often, but it may be a DPS gain to do once in a while.

[Alandrek]

Hammer of the Righteous hits harder then Crusader Strike single Target

This is unintended.

Crusader Strike should win handily against single targets. If it's not doing that yet it's just a numbers adjustment, so you'd be better off testing as if it were the case rather than trying to build a rotation based around something that you know we're likely to change.

http://forums.worldofwarcraft.com/threa ... 1&sid=1#17

[inthedrops]

I am wondering (just now, haven't tested) if it's worth hard-casting Exorcism during any of the otherwise empty X spots.

I'm not sure what hard-casting means. But just in case it means something other than what I'm thinking, the instant exorcism for prot spec was removed.

I went in over the weekend and toyed around. It was definetely foreign but I was kind of getting the hang of things after a few trash pulls in a dungeon with myself and a healer.

The thing I am currently disliking, A LOT, is how long it takes to build up 3 holy power. It feels like an eternity. I got to the point where I'd just save 3 HP from the prior pack so that I could ShoR early on in the next pack. I really don't like this.

[PsiVen]

From my post yesterday in a Cataclysm thread:

AS > Judgement (no SD before ShoR) > ShoR/Inq (3 HP) > CS/HotR > Judgement/Holy Wrath/Consecration

Always put AS on cooldown so that you can benefit from Grand Crusader procs.
Always delay ShoR for a judgement if Sacred Duty hasn't procced.
Always use your finishing move before your CP generator (obviously).
After that, fill in with the relevant spells. Judgement has some single-target priority, but HW/Cons still have chances to be cast. HW is great for snap AoE and lacks the melee requirement of HotR, so it's often a good opener if AS is on cooldown or there's more than 3 targets.

Holy Shield lasts 20 seconds, so you can delay CS/HotR 4 GCDs per finisher without it falling off. Holy Power generates and spends even if you miss, so nothing changes based on hit.

It's not perfect, but you get the general idea: work around the assumption that our procs are worth maximizing, and this pattern emerges.

[daiceman]

I'm going to create a program that will run for a simulated 5 hours of combat looking at different priority queues. So far I have 9 events, anyone have anymore they can think of? The current 8 I have are:

Grand Crusader AS
Judgment without Sacred Duty
Shield of the Righteous with 3 HP
HotR/CS (depending on situation, regardless, either generates HP)
AS
Judgment
Consecration
Holy Wrath

Can anyone else think of something to add?

Since we have such a small number of items, I'm just going to have my program run through every possible organization and record the number of each time every ability is used per rotation.

[theckhd]

I've already written the simulation, take a look at the MATLAB 4.0 thread. If I get time, I'll post a few examples tomorrow.

[theckhd]

You may want to check out the entire matlabadin project if you have an interest in coding simulations, we can always use more manpower.

In specific though, the prio_model and prio_sim modules simulate a priority queue structure over as long a time period as you want. There's also a rotation_drawer model that makes pretty graphs, though that's only useful for small time windows.

For example, here's a snippet from prio_model that shows how the condition/action framework works:

Code: Select all

prio(1).castname={'ShieldoftheRighteous'; ...
                  'CrusaderStrike'; ...
                  'Judgement';...
                  'AvengersShield';...
                  'HolyWrath';...
                  'Consecration'};
                       
%action is the structure that contains post-cast actions, including
%resetting spell cooldowns
prio(1).action={'hopo=0;sdflag=0;dur.SD=0;'...
                'ccd.CS=cd.CS;hopo=min([3 hopo+1]);if rand<mdf.GC ccd.AS=0; end;';...
                'ccd.Jud=cd.Jud;if rand<mdf.SacDut dur.SD=15; end;';...
                'ccd.AS=cd.AS;';...
                'ccd.HoWr=cd.HoWr;';...
                'ccd.Cons=cd.Cons;'};
         
%setup contains commands to be evaluated at the beginning of the
%simulation.  For example, setting ccd.AS=13.5 would simulate pulling with
%AS, making it unavailable early on in the queue.
prio(1).setup={'ccd.AS=13.5;'};

%cond is the structure that contains the conditionals which will be
%evaluated.  Each string should give a logical 1 or 0 when evaluated.
prio(1).cond={'hopo>=3 && ccd.CS>1.5 && ccd.Jud<=0 && dur.SD<=0';...
              'hopo>=3';...
              'ccd.CS<=0';...
              'ccd.Jud<=0';...
              'ccd.AS<=0';...
              'ccd.HoWr<=0'; ...
              'ccd.Cons<=0'};

%cast is the index of the spell that will be cast if the conditional is true
prio(1).cast=[3;1;2;3;4;5;6];

The "action" section defines the action to be taken when any of the spells given in "castname" are cast. So when ShoR is cast, we reset the holy power counter (hopo) to zero, turn off a flag for Sacred Duty, and set the duration of Sacred Duty to zero. When we cast CS, we reset the cooldown of CS to it's base (4.5s), increase holy power by 1 (enforcing a max of 3), and make a random roll for a Grand Crusader proc (which if successful resets the cooldown on AS).

That's all fairly static though, and won't need to be modified from the defaults. The real meat is in the condition ("cond") block. This is where we can set up any priority queue we like. The default I've put in there is:

Judgement if HP=3 and CS cooldown is > 1.5s and Sacred Duty isn't up
ShoR if HP=3
Crusader Strike
Judgement
Avenger's Shield
Holy Wrath
Cons

The first line handles the situation where you cast CS to get your third HP and want to cast Judge before ShoR to try and trigger a SD proc.

You can re-arrange these and/or write your own to try different queue structures. I'm still working on the code a bit - I want to clean it up some and implement proc tracking a little differently so I can include SD procs on the plots. The plan is to code several different priority structures so that we can run a long simulation of each and compare the results.

Note that the simulation does currently track damage, but I haven't updated the ability_model module to the latest beta or PTR build yet. By default, we were going to code it for level 85, but I could imagine adding a "ability_model_PTR" module with spell coefficients set to their lvl 80 values.

Alternatively, you could cheat and just define dmg.X for each skill individually before running prio_model and prio_sim. That probably won't be as accurate though.

Here's an example plot of a quick simulation of 30 seconds of combat in 0.1s increments:


The blue ShoR indicates a SD-induced critical strike.

[daiceman]

Hah, I knew there was a reason I kept my mat lab books from college, I'll have to look at the code and see if i can make any sense of it.

[Darielle]

Wouldn't it be better to AS > CS, except in the situation where you need the CS earlier?

For example, there are times when AS comes off cooldown at the same time as CS (e.g on that 30-sec simuation, it happens at 13.5 seconds if you assume the AS was done before the first CS as a "pull"). Pushing the AS first means that your following CS will have a chance to proc AS, whereas if you CS'd first, you're going to AS anyway, so if GC procs, that's a waste of what could have been a high damage move. If GC didn't proc, your overall is a wash, but if GC does, it's free extra damage.

[theckhd]

It really depends on the final tuning. Keep in mind that by pushing CS back, you're also pushing back your next ShoR. So the comparison is really something like 1/9 of a ShoR vs. 20% of an AS, plus any cooldown clashes you create or destroy in the process.

The whole point of writing the code was to be able to compare these sorts of things so we can determine what priority queue structure yields the best TPS. The queue I put up is just a rough guess based on the assumption that ShoR hits like a truck.

[Darielle]

It really depends on the final tuning. Keep in mind that by pushing CS back, you're also pushing back your next ShoR. So the comparison is really something like 1/9 of a ShoR vs. 20% of an AS, plus any cooldown clashes you create or destroy in the process.

It seems to more or less even out if it does get pushed back and GC doesn't proc, since anything that does get pushed back basically eats into the occasional gcd pause anyway, or at least that's what it feels like when trying it out both on the dummy and in instances. Numbers is harder to say.

[theckhd]

It seems to more or less even out if it does get pushed back and GC doesn't proc, since anything that does get pushed back basically eats into the occasional gcd pause anyway, or at least that's what it feels like when trying it out both on the dummy and in instances. Numbers is harder to say.

That argument doesn't make sense. If you push back CS to cast AS and GC doesn't proc, you now have two GCDs to fill before ShoR. If you cast CS first, AS fills one of the GCDs, leaving only one empty GCD. We don't have enough spells to fill all of our GCDs so as it is, which tends to favor prioritizing short-cooldown spells to minimize the number of gaps we need to fill.

Again, it will really depend on how hard AS hits compared to ShoR. If AS hits hard enough, it could still be a DPS to push CS (and subsequently ShoR) back. If ShoR hits hard enough, it won't be.

[Navan]

It seems to me that if they just add some way for Prot to work in an Exorcism every 8-9 seconds we would have a full rotation. Perhaps something like Glyph of Turn Evil for Exo or perhaps a Talent like Early Frost where the first cast of Exo in 9 seconds is instant but every one after that retains 1.5s cast time.

[sculder]

It seems to me that if they just add some way for Prot to work in an Exorcism every 8-9 seconds we would have a full rotation. Perhaps something like Glyph of Turn Evil for Exo or perhaps a Talent like Early Frost where the first cast of Exo in 9 seconds is instant but every one after that retains 1.5s cast time.

I've been doing a lot of tanking on the ptr, and while I initially hated the large gaps in our rotation, sometimes it's not bad at all. Grand Crusader seems to proc a LOT, giving me some AS's to fill gaps, and as long as you stick to a logical priority system, I was able to stack to 3 HP for a nice big ShoR multiple times in a fight, with very few gaps.

All in all we'll definitely have to be watching spell cds very closely during a fight, to ensure the best rotation. At this build it seems quite dynamic, but as long as I payed attention to spell cd's it actually worked out pretty well. (mostly thanks to grand crusader procs...)