Wednesday, October 28, 2009

25 (Heroic) Trial of the Grand Crusader: Anub'arak

The subject of Anub'arak hard mode is coming up more and more - spec, gear, rotation, etc.  I've killed it four times now which doesn't make me an expert, but this post will cover my experience and thoughts about the fight.

If you're not 100% familiar with Anub'arak, it has three phases:
P1 - 80 seconds long, with two add packs spawning.  You will see P1 three times.
P2 - 60 seconds long (I think - I'm actually not 100% sure).  Little adds spawn and you will have to manage frost according to your guild's strategy.  After this, back to P1.
P3 - Rest of the fight.  Phase 3 occurs when you get Anub'arak to 30%, which you need to do by the third P1.  This is Phase1 with no time limit (except the enrage), and Leeching Swarm.

The key to dps here is maximizing it in P1 and P3 by using AOE intelligently.

This is pretty standard - there are only two things I want to mention:
  1. Pre-potting : This is a very good fight to pre-pot on.  Talk to your MT if you need to, but you should get a countdown - our MT does 5-4-3-2-1-go, and I generally hit a Wild Magic Potion just after she says, "1".  This will be hard on your bank during progression, but the dps increase is very much worth it.  This will let you save your in-combat potion for p3.
  2. It's Halloween, and that means Candy Buckets.  And Candy Buckets mean Soothing Spearmint Candy.  This is 11 spell power and can stack up to four times for 44 total.  This is a very large dps increase, equivalent to 1-2 upgrades, so you should use it for as long as you can.
Anub'arak is a fight with a large AOE component, and you should spec accordingly.  I use this:

  1. I spec for Typhoon for reasons that I talk about here - it is a very large dps boost on AOE fights.  On most Anub'arak parses it accounts for 3-4% of my dps, and it also serves as an automatic Nature's Grace proc.
  2. Gale Winds should be pretty obvious.  A large increase to Hurricane and Typhoon damage.
  3. I ignore mana regeneration talents because I have found I don't need them - Innervate is sufficient regen and I can usually hit it during p2 when I need to be moving anyway.  If you do not have access to your own Innervate, I would definitely drop the rest of Imp IS and move those points into Intensity - otherwise you will have severe mana issues.
  4. Improved Insect Swarm - I drop a talent point from this because Imp IS offers the least dps increase for one talent point compared to the other options.
  5. Balance of Power - I considered increasing my hit on this fight and dropping one point of BoP in order to max out Imp IS.  What stopped me was Phase 3 - raid healing for this is really, really intense, and anything you can do to decrease incoming damage is a good thing.  6% reduction doesn't sound like much but when you're hovering at <500 life, it can be important.
After some experimentation, I ended up back with the standard glyphs for this fight: Starfire, Moonfire, and Insect Swarm.  There are two other Glyphs (Starfall and Focus) but I eventually decided against them.

Glyph of Starfall - Reduces the cooldown of Starfall by 30 sec.
The problem I had with this glyph is that it just didn't fit into the timing of the fight.  You want to use Starfall on the add packs - with 5 targets it does quite a bit of damage.  The problem is that add packs spawn every 45 seconds and take approximately 15-20 seconds to die.  With a 60 second cooldown, this means that your Starfall CD generally comes up right as the second set of adds dies - which means most of the benefit is wasted.  A 90-second CD fits in perfectly - you can use it right at the start of every other add pack.

Glyph of Focus - Increases the damage done by Starfall by 20%, but decreases its radius by 50%.
The problem with this Glyph is that the damage increase is just too small.   Even with the Starfall Glyph I never saw more then 8% dps from this spell, which means that a 20% increase would be 1.6% dps.  Both the Moonfire and Starfire glyphs are worth more damage then that.

I also want to mention two Minor Glyphs I would recommend for this fight:
Glyph of Typhoon - Reduces the cost of your Typhoon spell by 8% and increases its radius by 10 yards, but it no longer knocks enemies back.
Glyph of the Wild - Mana cost of your Mark of the Wild and Gift of the Wild spells reduced by 50%

Glyph of Typhoon should be obvious - if you are using Typhoon as a dps increase then you can't be knocking the enemies around and messing up everyone else.  The mana reduction is of minimal benefit, but it also doesn't hurt.
Glyph of the Wild is less obvious.  The reason I recommend this glyph is because if you find yourself nearly OOM and you need to AOE, then this glyph becomes a mana regeneration talent.   The key is that Gift of the Wild - when cast on 25 targets - will very nearly always proc Omen of Clarity.  This means that you only need roughly 1100 mana to cast a 2800 mana hurricane, which is a pretty significant savings.  You probably won't use this option a lot, but it's nice to have it if you need it.

Part of being successful at a fight is choosing gear that is optimal, whether due to mechanics or other factors.  I gear specifically for Anub'arak 25 simply because it's the only thing approaching progression for my guild right now.
  1. 4t9 - 4t9 is better on this fight then 2t8/2t9.  Not for the 4t9 set bonus, but for the extra stats you get from the upgrade - extra spell power, crit, and haste all benefit your AOE, while set bonuses do not.
  2. Value haste over crit - I already do this, but for Anub'arak you should take it a bit farther.  The reason behind this is that Hurricane doesn't scale very well with crit because it doesn't benefit from Vengeance.  On the other hand, Hurricane scales extremely well with haste.  Since Hurricane will most likely be 30% of your damage or more, you can see significant gains by favoring haste for this fight.  
Moonkins have two general cooldowns - Starfall and Force of Nature.  I use Starfall on every other AOE pack, but Force of Nature is a bit trickier.

Anub'arak has a 9:30 minute enrage.  That means with a 30 second duration and a 3 minute cooldown, you can theoretically fit in four FoNs.  In reality it's going to be better to aim for three - hopefully you won't be that close to the enrage, and using three allows you to time one with Heroism/Bloodlust.

My three FoNs are:
First p1: Just before Heroism.
Third p1: Right at the start.
Late p3: When the cooldown comes up.

As we are killing the boss faster and faster I may have to change to this in order to keep 3 cooldown uses:
First p1: At the start.
Second p1: In the middle with ~30 seconds to submerge.
Middle p3: When the cooldown comes up.

Special Note: Barkskin.  Usually in p1 and p2 you won't have a need for this, but it can be good to use it in P3. Every time it's up, I use it on the last 12 seconds of Penetrating Cold.  The reasoning here is that generally your healers should be able to heal that first tick without any help from you - if they can't then you have bigger issues.  So you want to use Barkskin for the last 12 seconds, where it will both 1) keep you alive as they are trying to reduce healing on you and go to their new targets, and 2) decrease the amount you heal the boss from Leeching Swarm.

The only special thing about Anub'arak is that you will be weaving in AOE at predictable intervals.  This isn't going to make a big change - Eclipse is severely RNG-dependent, so it's hard to tell what spell you would be casting.  Anub'arak is thankfully a stationary fight in p1 and p3, so we don't get hurt by a need to move around.

In general, this is what I do:

  • IFF->IS->MF->SFx3 to start.  I use starfire right off the bat because I want to extend my Moonfire, and I know I'll need to AOE shortly.
  • (First Add Pack) : Starfall -> Hurricane -> Typhoon -> back to single target.  My phase1 AOE is shorter because some healers help out - you may thing that wouldn't be much, but keep in mind that balance doesn't have a lot of talents that benefit Hurricane.  A resto druid can put out some pretty respectable numbers, as can priests and shamans.
  • (Second Add Pack): Ignore and continue single target.  Once Anub'arak submerges, Typhoon -> Hurricane.  At that point they are usually dead and on to p2.
  • Run around, DoT/nuke adds, and do whatever your p2 strategy is.
  • IFF->IS->MF->SFx3 to start, for the same reasons as p1.
  • Odd add packs (1/3/..):  Starfall->Hurricane->Typhoon->Hurricane->Single Target.  Since healers actually have to heal in phase 3, you will get to AOE a bit more.
  • Even add packs (2/4/..):  Typhoon->Hurricane x2 -> Single target.
Keep in mind that all the above are estimates - I always hurricane until the enemies are dead or extremely close (<5%).  If your guild's AOE is weaker then mine, or if some people are dead and there is less AOE, then you may need to Hurricane more then I do.

Monday, October 26, 2009

How To Choose Gear Intelligently

I see a lot of threads with the following two topics:
"My dps is bad, how do I fix it?"
"Is X an upgrade over Y"?

Many times these threads reveal a lack of understanding about how to pick gear - it's very often that someone does poor dps not because they are bad at being a moonkin, but because they choose gear badly.  Picking gear can be fun (it's one of the reasons I play) and this post will talk about how I do it.

Gearing Toward a List or Not
There are basically two options you can use when choosing gear:
1) Gear towards a specific, BIS (Best-In-Slot) list.  This is a pre-determined list of gear that you feel gives you the best dps.
2) Pick up upgrades as they drop.

If you use the first method then this post won't really help you.  I don't really like gearing toward a BIS list because I have found that I never get it.  There is always an item (or two, or three) that refuses to drop or that I just can't get because of DKP/loot council/etc.  As a result, I prefer to take a more fluid approach to gearing.

This by necessity involves a little more thought then just following a list, and this post details most of the logic/thought process that I use.  Keep in mind while reading this that these aren't in a specific order - you want to consider everything, not just look at one particular section.  Just because Hit is more valuable then Spirit doesn't mean an item with Hit is better dps then an item with Spirit, for example.

You have to know what stats are good, and approximately how good.  This can be as simple as:
Hit (to cap) > Spell Power > Haste (before 400) > Haste (after 400) >= Crit > Spirit > Hit (after cap)

The only problem with this is that it's a little inexact - you don't know how much any particular stat is better then other stats, after all.  What I generally do is use the following values:
Hit: 2 (before cap, 0 after)
Spell Power: 1
Haste (before 400): 1.1
Haste (after 400) and crit: .8
Spirit: .33

These are loosely based on Graylo's numbers here.  I don't use the exact values because I don't want to bother multiplying stuff by 0.816, and I feel that these are close enough for eyeball work.  If I compare two items and they are close (within ~5 dps) then I break out a spreadsheet and compare them using more precise values, but for the majority of cases the above works fine.  Most upgrades are on the order of 20+ dps or more, so it doesn't matter if you use exact numbers - the result will be close enough.

Know What Gear is Available To You
You must know what gear is available.  The keyword is the one in italics, though - if you are in a 10-man only guild, then you are going to choose gear differently then a person running 25-man content.  There are a lot of sources for this:
Graylo's BIS lists, here and here.

Keep in mind that I'm not advocating a BIS list here - but it can still be useful to use a best-in-slot list simply to see what gear is available. 

Know What Gear Others Will Want
It's a simple fact of life that in WoW, you are gearing in a raid (at least in endgame...).  And that generally means you do not get every single item you want.  This especially refers to shared slots: Trinkets, Rings, Amulet, Tier gear, and Weapon/Off-Hand.  These are slots that a LOT of poeple are going to want, so you need to pick you gear with that understanding - because maybe something that would otherwise be BIS is something that it will take you a long, long time to get.

On the other hand leather spellcaster items are probably something you can pick up pretty easily, because so few people will want them.  This especially refers to non-spirit leather - there's not a lot of it, but you can generally expect only 1-2 people in the entire raid might be interested.

Set Bonuses
I've talked about some set bonuses, but now I want to discuss them in more general terms.  The key to set bonuses is that they don't take up itemization points like stats - they are essentially free dps.  This means that except for special cases, you are almost always better off aiming to have at least four slots dedicated to tier gear.  That can either mean going for a 4-piece bonus or a 2/2 split, but it's a rare situation where you want to use more then one piece of off-set gear.

Gear Toward The Hit Cap
I'm not sure I can emphasize this enough.  I see a ton of posts where people talk about having low dps and it turns out they are massively above the hit cap.  Unlike crit or haste, the hit cap isn't a soft cap - you get zero benefit from it once you pass the 17% mark.  So it's very important to choose your gear with the hit cap in mind (if you don't know the hit cap, I recommend reading Wisp's raiding sticky here).

I'm also going to take a second to talk about Balance of Power here - there are some people who advocate dropping BoP.  I'm not one of them - the problem is that BoP also provides a secondary benefit of reducing all spell damage you take by 6%.  While this isn't a big number, keep in mind that almost all the damage you take will be magic, and dead = 0 dps.

The primary reason not to drop BoP though is that there's just no benefit from it.  Balance is a pretty limited tree - the other talents you can pick up if you drop BoP are stuff like Brambles or Genesis, which provide single-digit dps increases.

Minimize Spirit
This is probably the second source of "well geared" moonkins doing low dps (after having too much hit) - having too much spirit.

Wotlk itemization has been pretty consistent.  Every caster item looks like this:
Spell Power
Two out of the following: (Hit, Haste, Crit, Spirit)

Out of those Spirit is by far the worst.  Some spirit is unavoidable - but in general if you pick items with spirit over items where you could get crit/haste, you are losing 30+ dps per slot.  That's for low-itemization slots like rings or bracers - you lose a lot more if you take a helm, chest, or leg slot with spirit, so avoid it whenever possible.

Maximize Sockets
Sockets, sockets, sockets.

They key to sockets is that they are cheap, itemization wise.  What you lose from adding a socket to an item is much less then you gain - so if you had two items that were "identical" (both SP/crit/haste and the same ilvl), the one with a socket will beat the one without.  I always choose gear in order to maximize sockets while keeping the rest in mind.

The Complete Whole
The end result of this generally ends up being very close to a BIS list - but a more fluid one.  I picked my weapon not because it's BIS (I'd prefer the Heroic Anub mace) but because:
  • I had the DKP.
  • It dropped.
  • Weapons are a very contested item, so if I waited for another one I'd have to horde DKP.
  • It's well-itemized, with a socket and no spirit.
From that I built the rest of my gear around it (because it provides hit) based on the other criteria I listed here.  I ended up with a little more hit then I'd like, but overall it's turned out pretty solid.

Monday, October 19, 2009

When To Reapply DoTs

I see a lot of questions about when to apply DoTs and I wanted to weigh in with my 2c.  I'm not claiming that my choices are optimal, but I find them to be a 1) workable, and 2) decent dps.

This post will be by necessity non-mathematical.  Not because I think math is a bad idea for this, but because I haven't been able to figure out how to model the benefit of a DoT when compared to casting another spell during Eclipse.

(An interesting thought I just had on that is that maybe my basic premise is wrong.  I've been trying to work out a formula of the form of f(A,B) = X, where A is the benefit of the DoT, B is the opportunity cost, and X would be the number of seconds left in Eclipse where it is still a good idea to reapply DoTs.  But if the number of seconds left doesn't matter, then a formula like this is impossible - I'll have to think on that more).

The basic problem I have with modeling the benefit of DoTs is that there are two negative factors.  When you apply a DoT during Eclipse:
1) You lost the benefit of Eclipse to one Wrath or roughly 1/2 of a Starfire.
2) You have to reapply that same DoT slightly earlier then you would otherwise.

The first is very easy to model, and my math based on it led to a "6-second rule" - if Eclipse has >6 seconds, it's better to reapply DoTs.
The second seems to be almost impossible - how to determine the negative effects of having to refresh Insect Swarm, for example, in 15 seconds instead of 21.  This makes it extremely hard to compare to the damage of a DoT and determine when to refresh.

The end result of trying to come up with this is I've changed my philosophy on reapplying DoTs. I previously used the 6-second rule on most fights.  After thinking about it though, it seems like this was optimistic - it ignores the effects of using slightly more GCDs on DoT damage... I think.  Therefore:

Single Target Stand-And-Nuke/Light Movement
1) Reapply Moonfire if it drops off to ensure maximum idol uptime.
2) Reapply Insect Swarm at the end of every Eclipse if it's off.  This has an additional benefit of allowing smoother Lunar->Solar transitions with WiseEclipse.
3) Reapply Insect Swarm during Solar Eclipse only if it falls off, no matter when it falls off.

The main benefit of these changes is that they require slightly less thought (as I never take into account the amount of time Eclipse has), and they lead to a slightly more predictable rotation, both of which are beneficial if I'm distracted, tired, or anything goes wrong during a fight. 

Single Target with Medium to Heavy Movement
1) Reapply DoTs if they drop off and I know I won't have to move for 5+ seconds.  Otherwise save them and use them for when I have to move.
2) Ignore the above for Insect Swarm if I know there's predictable movement in the duration of the DoT (for example: Ignis, and there's going to be a flame Jet within 14 seconds).

This is pretty simple - basically if I know I have to move a lot (H Twins as a goalie is a good example of this) then my DoT strategy changes significantly.  It becomes much more important to optimize dps by using DoTs as instant cast spells that I can hit while moving, instead of worrying about more optimal uptimes.  In general I try to keep Moonfire up just for the Idol effect, but other then that it's more important to save DoTs for when you can't cast anything else.

Two/Three Target Fights
1) Apply DoTs on any target that A) provides useful dps, and B) will last long enough for the DoTs to fully tick on.

The important note here is that it must be useful dps.  I don't advocate DoTing up all the adds on Iron Council because it's not useful - they heal to full.  I don't advocate DoTing up both worms in Heroic NrB because killing Acidmaw fast is key - lowering your dps to make the overall fight shorter just isn't that important.  Faction Champions is about focused dps, so again - no DoTing up targets (at least until the end, or on pets to draw off heals).

Good fights for this are stuff like Flame Leviathan (if you go up top and kill sentries), Freya triplets (as long as you don't kill anything too early), Mimiron p4 (same as Freya triplets), and Yogg-Saron p2.

AOE Fights (4+)

1) Keep up Moonfire for the idol effect.

It isn't worth it to spam DoTs on 4+ targets, Hurricane is better dps.

Hope some people find this useful.  I'd also be interested if anyone else has good math on when to reapply DoTs in Eclipse, I haven't seen anything concrete on it.

Thursday, October 8, 2009

Patch 3.3: tier 10 bonuses

So we have our t10 moonkin bonuses:

2 piece bonus - When you gain Clearcasting from your Omen of Clarity talent, you deal 10% additional Nature and Arcane damage for 6 sec.
Update - increased to 15% for 6 seconds.  So the benefit increases to approximately 3%

Omen of Clarity is something like a 6% proc rate on spells.  If we cast approximately 35 spells per minute (roughly the average based on WoL) then we should see 35*6% = 2.1 procs per minute, which matches what shows up in WoL.  2.1 procs per minute at 6 seconds each is 12.6/60 = 21% uptime.  So for a 10% increase, that should be about 2.1% dps.

However, there are several issues with this:
1) The 10% only applies to spells that are finished within the 6 second duration.  So some portion will almost always be wasted.
2) Omen of Clarity doesn't have an internal cooldown, so if it refreshes itself then part of a bonus will be wasted.
3) It's very RNG based.  At 2 PPM the number of procs per fight could vary widely.

So overall for single-target fights, I see this being very weak, <2%.  However, one thing to bear in mind - on AOE fights you will see a much higher proc rate on OoC, which means the uptime could be significant - 10% AOE damage on certain fights could be very, very good.

4 piece bonus - Your critical strikes from Starfire and Wrath cause the target languish for an additional 5% of your spell's damage over 4 seconds.
Napkin math time.  Say dps on single-target looks roughly like this:
20% DoTS/starfall/treants
40% Wrath
40% Starfire

Wrath crit rate should be about 55%, Starfire crit rate is about 75%.  Crits are 209% of a hit.  Then:
2.09*.55 / (2.09*.55 + .45) = 71.9% of your wrath damage is from crits.
2.09*.75 / (2.09*.75 + .25) = 86.2% of your starfire damage is from crits.

.4 * .719 + .4*.862 = 63.24% of your damage from Wrath or Starfire crits.  4t10 should just be a flat 5% damage increase to crits, so:
63.24% * .05 = 3.2% dps increase

>3% is certainly respectable and is roughly as strong as 4t9.  This does assume it's not buggy though.

Could be better, could be worse.  Depending of course on itemization, I don't see any issue with upgrading from 4t9->2t10, or from 2t9 to 4t10.  The set bonus + stat increases should make it worth it.

Saturday, October 3, 2009

3.3 - Black Magic

In case you weren't aware, Black Magic is being changed in 3.3:
Enchant Weapon - Black Magic: now cause your harmful spells to sometimes increase haste rating by 250.

I'm going to use Graylo's numbers and put haste rating (past the soft cap) at approximately .8 spell power.  Then the proc is worth 200 SP over 10 seconds and would have to have 63/200 = 31.5% uptime to equal a spell power enchant (about 23% uptime if you were below 150 haste, a pretty unlikely scenario).

I managed to test this for about 15 minutes before connection issues forced me off the PTR.  I'll update this once I have better data, but some initial observations:

I never saw it proc twice within ~30 seconds (two Eclipse durations).  It also never refreshed itself.  So that strongly suggests there is still an internal cooldown, probably of 45 seconds.
I don't know the proc chance for sure, but it doesn't appear to be even as high as 50%.  I did four tests where I waited 1 minute between procs, and the number of casts to proc were: 2, 5, 11, 6.

Using the ICD of 45 seconds and estimating the proc rate at 10% (based on similar spell effects), you should expect roughly 1 proc per minute, or a 16.66% uptime.  That puts Black Magic at roughly 50% the dps of a 63 spell power enchant, so it's definitely not worth it.