Wednesday, March 11, 2009

Patch Notes 3.1: Ok, We've had some time to think.

Update: Erdluf did some testing on the Spell Queuing system. You can find his results here. The short version of his results is that you can cast a spell upto 0.3 seconds before your prior spell is finished casting and have the next spell be queued. This gets a little more complicated when the Global Cooldown becomes involved. Apparently you can queue a spell during the global cooldown but the GCD, but you can't queue a spell during the first second of the GCD.

So, this means that spell queuing works well with Starfire and Wrath casts with little haste. However, once you start getting close to the 1 second GCD, haste starts to have real issues with the spell queuing system.


Like a lot of you I was excited by the possibilities of all the changes changes introduced to the PTR last Friday. Its the first time I've every seen a significant number of buffs introduced outside of an expansion and I liked the changes, but it also made me nervous. Moonkins have stuck by the Starfire for Raiding/Wrath for PvP mantra for years, and all of a sudden this core principle of our spec seemed to be threatened. However, the more reading and math I do the more I realize things are going to stay the same. That is not to say that these changes are insignificant, I just don't think they are going to change our core play style as much as it seemed at first.

Before I get into specifics lets get one thing clear. These changes are only on the PTR. It is very possible and somewhat likely that changes will occur. Don't get your heart set on anything until it goes live. Now, lets look at some specifics.

Insect Swarm:
In my last post I speculated that the Coefficent would be brought up to 0.8 from 0.76. Well, Dukes on the EJ forums did the testing (link), and saw the coefficent has been increased to 1.2. Obviously this is much more then I expected and it is now obvious that IS will be staying as a part of the Moonkin Rotation. Realistically this was probably necessary to keep the spell relevant. IS dropped out of our rotation in T5 content because it didn't scale as well with Spell Power, and it was in danger of having it happen again in T8 or T9 content.

Natures Grace:
I think Randomsmo said it best on the WoW forums:

NG is exactly the same for Starfire in 3.1, except it will effect multiple spells.
NG is exactly the same for Wrath in 3.1 (unless you had really horrible haste), except it will effect multiple spells and it *might* (this is unconfirmed, so don't get too excited) lower the GCD to prevent clipping.

There are two angles from which you need to look at this change. The first is a cast time perspective, the second is an uptime perspective.

CT Perspective:
From the cast time perspective this does almost nothing for Moonkin raiders. For Starfire, 20% haste brings an unbuffed starfire down to 2.5 seconds just like the live NG currently does. For Wrath, it brings the cast time down to 1.25 which is a nerf when compared to the live version of NG. However, the impact is minor once you start considering raid buffs and haste from gear. Assuming your specced correctly a moonkin is going to have an additional 11.41% haste from talents and Wrath of Air, and that brings our haste total to 33.67%. Now you need just12.21% haste from gear to reach a 1 second Wrath. That is just 400.45 Haste Rating. If your raiding 25mans consistently then your most likely already past this point. If your not I'm sure that raiding Ulduar will push you past it quickly.

Sure, you could gear differently so that you take more Crit and minimize your Haste rating, but I don't think that is very likely. I have a spreadsheet that I use to value gear items. I built it to eventually provide a comprehensive list of gear choices with a relative value attached to each. When I updated the model for these changes I was surprised at how little changed. The values changed but the ranks didn't for the most part. The fact of the matter is that haste is unavoidable, and anyone who gets to the higher levels of raiding will blow past 400 haste whether they like it or not.

UT Perspective:
From an Uptime Perspective this is huge. Lets remember one important thing. Cast time is determined at the beginning of the spell cast. So a 3 second buff will affect multiple spells even if it takes more than 3 seconds to cast those spells.

Without haste from gear, your Starfire will have a cast time of 2.24 seconds (3/(1.03*1.03*1.05*1.2) = 2.24). Therefore I think it is safe to assume that you will usually be able to start two Starfires in 3 seconds.

Wrath will be down to 1.12 seconds without any haste from gear (1.5/(1.03*1.03*1.05*1.2) = 1.12). Therefore I think it is safe to assume that you will usually be able to start three Wraths in 3 seconds.

So, we can calculate the theoretical Natures Grace uptime like this assuming a 40% crit rate for both spells:

SF NG Up Time = 1 - (1 - 0.40)2 = 64.0%
Wrath NG Up Time = 1 - (1 - 0.40)3 = 78.4%
Actual uptimes will be lower then this because of movement and casting spells that don't crit, but this is a significant increase over the live version of Natures Grace which would be a little less then 40% for both spells.

So to sum up the section, from a cast time perspective this change change has little impact, but it will affect a lot more spells because Natures Grace will have significantly more Uptime then the live version of the spell. This is a buff for PvE moonkin.

Spell Queuing:
I hesitate to talk about this because I don't have any formal understanding of the mechanic. I only know what I know from personal experience and random comments on various forums. So, this is a warning, I believe that what I'm about to say is correct, but I don't have a solid foundation to base it off of.

It is my understanding that WoW has a Spell Queuing system to help minimize the impact of latency and human error. The way I think it works is if you try and cast a spell a little bit before your current cast is completed, then the game will Queue the spell to start casting when your current cast is completed. It's kind of like a caster version of auto-attack, and it works great for longer cast time spells like Starfire.

However, for shorter cast time spells like Wrath the system starts to breakdown because the global cooldown get is the way. If your Wrath cast time goes below 1 second the queued spell will be unable to cast because the global cooldown blocks it. So the timing of your next spell cast is determined by when you press the button reintroducing the impact of human error and increasing the average time between Wrath casts. Even a small increase in cast time like a tenth of a second will increase Wrath's average cast time by 10% and have a significant impact.

I'll say it again that I don't completely understand the process, but this is one of the main reasons Starfire is currently considered to be a superior dps spell then Wrath.

There have been some interesting posts on the EJ forums discussing this subject. It is a little spread out over the thread but it starts here with post 485. Update: Erdluf did some testing on Spell Queuing. You can find his results here.

I don't want to rehash everything I did in my four part Eclipse series, but obviously with the changes to Natures Grace and Eclipse there is an impact. If these changes go live I will update the analysis, but I want to just give you a summery now. Murmurs also did some nice math regarding this change on the WoW forums. you can find it here.

Spell vs Spell:

I've run the numbers for the hypothetical moonkin I used in my Eclipse post who had 2000 Spell Power, 35% crit, and 6% haste from gear. Under these conditions the average Starfire with Eclipse will do 3,499 DPS. A Wrath cast using the same stats will do 3,791 DPS on average. Spell vs Spell a Wrath cast with eclipse will do 8.3% more damage on average than a Starfire cast with Eclipse.

Rotation vs Rotation:
While the Spell vs Spell view is all fine and dandy we don't cast spells by themselves. We cast them as a part of a rotation, and the rotation view of Eclipse is not kind to Wrath. Using the hypothetical moonkin, an Eclipse rotation that favored Starfire did on average 3297 DPS. An Eclipse rotation that favored Wrath did only 3301 DPS. I ran the numbers at higher levels of gear and Wrath did improve relative to Starfire, but at most Wrath's advantage was at most 2.5% to 3% and this is ignoring the impact of Spell Queuing. If you add a tenth of a second lag to Wrath its advantage is completely erased and then some.

I can't help but concluding that the Starfire rotation is still the superior rotation if these changes go into affect, but that does not mean that this change is meaningless.

This change helps us in three ways.
1. First, if your looking for a more stable DPS rotation then the Wrath rotation is for you. It has a 100% chance on a SF crit and the buff is a straight damage buff. Therefore it is not affected by RNG nearly as much as a Starfire rotation is.

2. Second, this change limits the impact of mistakes. We've all been in situations where we get to focused and accidentally proc the wrong Eclipse. Now that the Wrath buff is 30% this mistake isn't as costly as it is on Live.

3. Each boss has its own issues and the all favor one style of casting over another. Stand and Nuke bosses are great for long cast spells like Starfire, but in high movement fights like Thaddius, Starfire can be an issue because you have to move before you can complete the cast. Therefore Wrath may be better for high movement fights because you will be able to get more casts off.


Anonymous said...

Even after your stats indicate Wrath is supperior to Starfire in DPS, you still consider SF rotation the best?

Wrath beats SF hands-down(period)

I love getting into DPS races with other moonkins who use SF rotation in raids. They are usually 1000dps lower and then whisper me with, wow!, how'd you do that?, what am I doing wrong? etc.

Very simple, I don't use SF - ever!

Anonymous said...

^ and then you run into some decent geared moonkins and they blow you out of the water. Not only do they beat you on the damage meters, they have 70% mana while you've been OOM for the last minute of the fight, all the while they're doing more aoe damage/throwing up IFF for the raid (i.e., they don't have to waste talents on mana regen... not that you necessarily do either but you could).

Cdin said...


Apparently you did not read my entire post or choose to ignore the parts that don't agree with your view point.

In an ideal Mathimatical sense Wrath > Starfire spell for spell in terms of DPS. However, If you look in my post you will see that when you compare the rotations Wrath's advantage deminishes. On top of that, It ignores the addtional Latency that Wrath suffers from but Starfire does not.

In a practicle sense Wrath and SF are very close to each other in terms of DPS, but SF is clearly superior in terms of DPM. Therefore SF is the favored rotation in my opinion. That is not to say that there are not times when Wrath is superior, as I have indicated a couple of examples in my post, but SF is the main rotation in my opinion right now.

Now the second thing I want to talk about is your comment that you beat other moonkin by 1000 dps in races. I find this statement a little funny. Not because I think it is incorrect, but it makes some huge assumptons and ignores basic facts.

Fact of the matter is a lot of things going into determining someone's DPS for a particular fight, and equating a 1000 DPS difference to spell choice is unrealistic. Even under ideal circumstances wrath is not that good. You also have to consider gear, RNG, internet connection, role with in the raid, and you may just be a better player then they are.

There are plenty of things that can cause you to beat someone by 1000 dps that have nothing to do with spell choice.

Murmurs said...

A very solid assesment Graylo, as always.

It's not all that unexpected that, spell by spell, a Solar Eclipse outperforms a Lunar Eclipse; at least I wasn't surprised by it.

Most things you mentioned show why Lunar is still favored despite this, however a few things that you didn't mention were: Potions of Speed are better than Potions of Wild Magic. While this only effects one Eclipse proc, that is a major benefit to a Lunar Eclipse.

The Starfire Idol still trumps the Wrath Idol. This also makes us want to cast Starfre more than Wrath in ny given rotation.

The Meta Gem favors a Lunar Eclipse. While not a major increase, 9% more damage on critical strikes plays into the favor of getting more of them.

Genetic said...

Graylo you should address the Starfall Issue, with the lowered cooldown how big a roll will the new starfall play in moonkin DPS and will it be worth glyphing for with either/or Glyph of Focus and Glyph of Starfall.

Murmurs said...

I can answer that one Genetic:

Right now, from testing that I've done, I can say that the Glyph of Starfall is not worth losing any of the other three Glyphs over. As long as you can keep a 50% uptime on Insect Swarm, then the Insect Swarm Glyph will net a higher DPS return.

The Moonfire and Starfire Glyphs are still the highest.

I haven't done much testing on the Glyph of Focus. I can see some pros and some cons to it.

Overall, however, the choice isn't overly significant. We're speaking in terms of probably around 100 DPS if that. If you're raid is generally light on AoE damage, then either Focus or Starfall are perfectly viable without destroying your DPS nor the scaling of Insect Swarm.

Luminnous said...

I am still wondering one thing. You call it a Starfire rotation, but to proc the Starfire buffed eclipse you have to cast Wrath many times and when you begin with Starfire you only get 3-4 Starfires off during the proc.

Mabye you are casting Starfire through the hidden CD and then back to Wrath? Is this where the rotation becomes Starfire heavy?

What is the average proc casts for you when casting wrath to proc Starfire? If your lucky, all three casts will crit, but not usually.

Just curious.

Chad Tondut said...

@ Luminnous

Normally you only cast wrath during the time where eclipse is coming off internal cooldown, there is usually a 5-10 second window where you are casting wrath each rotation

Eilelia said...

I have been immersed in moonkin-theory research the past week or so, and I have to say that I love your blog and all the comment discussions. After integrating a lot of the things learned here, I am quickly and steadily climbing up the raid DPS meters and feeling like a much more competent and useful laser chicken. So, ty =)


A guildie recommended that I max DPS by alternating wrath/starfire - basically switching on each Eclipse proc and staying with that spell till it procs the other way. Thoughts?

Cdin said...


I wasn't surprised that Solar won in a spell vs spell comparison, but I was surprised at how close they were when I compared them rotation vs Rotation. I found that when you combine the change to NG and 30% crit from Eclipse, that NG has almost 100% uptime. So I guess it makes sense.


I should probably do a Moonkin Glyph post at some point but I've got other some stuff I want to do first.

The short answer is what Murmurs said. My napkin math and what I've read agrees with what he said.


When I say a "Starfire Rotation" I mean a rotation that focuses on SF and buffs its damage. This will of course include Wrath to proc Lunar Eclipse, but I consider the entire thing a a starfire rotation.


I've heard that suggestion before and I don't really like it, but it may become better with 3.1.

Basically Lunar(SF) eclipse is much better then Solar(w) eclipse. By switching back and forth your are using an inferior buff 50% of the time. It may be a little easier to manage but it is definately a DPS loss.

It may be a little more viable in 3.1 because Solar Eclipse is buffed, but we will have to see how everything shakes out.

Cyclo said...

With the NG change to a 20% haste buff over 3s and the fact that the gap between wrath and starfire eclipse become minimal, could a wrath x2 starfire x1 (to maximize the cast time reduction of each NG proc and reduce the impact of the eclipse spell switch) become actually the best way to try to proc eclipse?

Because there are 3 ways of using this NG proc, wrath x3, starfire x2, or wrath x2 starfirex1, assuming no haste at all, wrath x3 provides a 20% haste over 4.5s of cast with 3 spells to proc NG again, starfire x2 provides 20% haste over 6s but only 2 spells to proc NG and wrath x2 Starfire x1 provides 20% haste over 6s with 3 spells to proc NG again.

Since Wrath has travel time, if the first wrath would actually proc eclipse, chances are that the player would be ending his 2nd wrath or maybe already casting starfire, so the logic about losing 1 spell worth of cast time in each eclipse proc could be kinda sorted out.

Sorry for the long Comment