Tuesday, June 23, 2009

Patch 3.2: The New Eclipse

Update: Well don't I look like a psychic? There will be a new Idol with the release of 3.2. My first impression is that it looks good, but it really depends on the proc rate.

The patch notes have been out for a couple of days now and the Moonkin community has had a chance to digest the changes. You may remember my first reaction was shock. I didn't expect there to be any significant Moonkin changes even though I knew we are currently a little under powered.

Well, I'm still a little in shock and part of me can't wait for the changes. (The other part is worried that I won't be able to complete the hard modes and get the Drakes before they are dropped.) There have been a lot of numbers thrown around since the announcement and a couple of interesting ideas on how this will change our rotation. Let me try and provide a little more concrete info then what I provided last week.

Methodology: I am about to present quite a bit of numbers without showing my work. So, I would like to give you a little back ground on how I came up with these numbers. First, I assume a moonkin very similar to myself. I used my talent build, 3150 spell Power, 37% Crit rating (40% with Imp FF), 13% haste from gear, and fully hit capped. My model also assumes that the moonkin has 2T8 and 4T8 equipped, but excludes heroism.

I have modified my model in several ways. I have shortened it to 5 minutes worth of casting, but increased the number of trials to 40,000. Factored in a delay when choosing which spell to cast when Eclipse first procs. I have also made all the damage realtime, meaning that the DoT damage is accrued as it ticks instead of upfront. I have also added some lag into the model. I've tried to add lag that is consistant with SimCraft since those numbers or often quoted on the forums. Basically I used 0.075 seconds for Queue Lag and 0.150 seconds for GCD lag. Meaning a Queued spell has less lag then a spell with a cast time equal to the GCD.

I will put the new models up on file front when I get a chance.

DPS:
No one doubts that this is a DPS increase. The question is how much, and I've seen several numbers thrown around. Starfox, the druid Simcraft author, updated Simcraft with the changes and estimated that it would be a 8.15% DPS increase. Hamlet, another prominant moonkin on the EJ forums, made a comment here last week and estimated that it was a 6% DPS increase using Wrathcalcs. I even used an early form of my model to estimate the increase to be 10.7% though I new that to be bit high due to the model's imperfections.

I've had a chance to play with my model a little more, and I have some numbers I'm confident in.

The Lunar Rotation with zero lag had a average DPS of 5,474. When Lag was introduced the average DPS dropped to 5,100. Using a Twisting Rotation the zero lag model had an average DPS of 5,938. When you add lag the average DPS for a twisting rotation dropped to 5,404.

Avg DPS Increase (no lag) = (5938-5474)/5474 = 8.48%
Avg DPS Increase (lag) = (5404-5100)/5100 = 5.96%


Obviously my first number of 10.7% ways quite a bit high, but my new numbers are falling in line with what other people are finding. Now, I think it is reasonable to expect a DPS increase in the region of 6%.

While these purely mathimatical gains are great, I don't think it's going to stop there. I also think that this simplifies the Moonkin rotation a little bit and that should also result in slightly higher DPS. Currently we have to watch not only for the Proc, but we have to watch for cooldown expiration. Since we have fewer things to watch for we should impliment the rotation more easily. An easier rotation means fewer mistakes, and few mistakes mean increased DPS.

Note: I want to make a quick note on potential Nerfs. I have seen several people wondering if Blizzard will nerf the Eclipse buff or possibly reduce the proc rate, because having two cooldowns seems over powered. While this is possible I don't think it will happen. At least not in 3.2. Most reports I see have moonkin Lagging behind other DPS classes by quite a bit. Ghostcrawler him self said were lagging a little behind a month or so ago, and I don't think the change to Improved Moonkin Form corrected the issue.

While I do think it is strange that one talent will have some much impact on our DPS output I don't think it will make Moonkin over powered. If it does, I don't think it will be recognized until after the patch goes live.

Stats and Gear:
Obviously, when your rotation changes, the way you use stats changes as well. So, this has a big impact on how we value the different DPS stats. For the first time in my WoW life I can confidently say Crit > Haste on a point for point basis. There are the values I found.

StatNo LagWith Lag11% Haste With Lag
Hit Rating1.6081.5951.614
Spell Power111
Haste Rating0.6780.3890.941
Crit Rating0.7530.7410.742

So, this obviously changes the way we gear right? Not really.

The fact of the matter is that there are relatively few options for gearing, and what makes most items so good is not the quality of stats but the quantity of stats. I plugged these numbers into my gear list and I was surprised at how little changed. Most of my BIS choices I would still consider BiS because they are of a higher ilevel. Stepping down from there things shifted a bit more, but remember the first two columns presented above are post 400 haste numbers. The third column is a below 400 haste valuation. As you can see it is still important to try and get around 400 haste.

Remember, that Ulduar gear is not loaded with haste. When I and other people built our Lunar BIS lists, almost all of us had our haste rating dropping into the 400 range when I had over 600 in Naxx. There are a few exceptions but for the gear that is good now will be good after 3.2.

Idols:
This seems to be the biggest question on everyone's mind. Which idol should we use?

It was suggested on the EJ forums that Idol Swapping might be viable between eclipse procs when we refresh DoTs. The way it would work is that you would swap Idols while you cast your instant spells between Eclipse procs. It is possible because swapping the idol can't interrupt an instant cast and your Casting GCD would still overlap with your Swapping GCD just like in the old days.

While I think it is a very interesting idea, the process has several issues that make it impractical in my opinion. For one, the last time I checked the equipment swapping GCD was not affected by haste. Meaning you get a 1.5 second GCD instead of a 1 second GCD. On top of that the swap isn't instantaneous. Therefore the you will have a little more time added on for an even longer GCD.

This process would also force you to have a lower DoT up time. I'm not 100% sure of when the best time to refresh DoTs is, but this change would force you into the minimum DoT up time, so you may be losing DPS in this way.

In general I think Idol Swapping would significantly complicate the rotation for a relatively small gain in DPS. Therefore I suggest picking one Idol and sticking with it, but once again which Idol do we choose?

I am no longer confident in the Idol numbers I presented here. I seem to have over valued the IS idol and I am not sure why yet. I will say this though, of the three Idols currently in game the SF Idol is the best using a twisting rotation. The Wrath Idol is a distant 3rd. The IS idol is solidly behind the SF Idol, but there may be situations where It comes out ahead.

None of this matters however, because the new idol is likely to be the best. Reports from the PTR are saying it has a 50% proc rate, though I doubt that will last if true.


I've plugged the different Idols into my Twisting Model with Lag to see how they compare. here are the results:
Starfire Idol = 5,405 DPS
Wrath Idol = 5,372 DPS
Insect Swarm Idol =
5,419 DPS
Well that is a little bit of a shocker. Who would have thought the IS idol would come out ahead? I didn't. I ran the model twice to make sure it was correct.

In reality this isn't that big of a surprise. You may remember in my blog post on Idols I said that the IS does more for an individual IS cast then the other two idols do, but it falls behind because you cast IS only once every 14 seconds. As long as you could cast 3 Starfires or 5 Wraths within 14 seconds then one of the Nuking idols would be better. Using a Lunar or Solar rotation meeting those thresholds was a near certainty. With a Twisting rotation it is a little more in doubt.

We also need to realize that my model assumes a very high IS uptime. It refreshes DoTs during Eclipse. So, my number is probably a little inflated relative to some of your play styles.

When looking at the two nuking Idols the SF Idol comes out ahead. This is largely due to it being of a higher iLevel, and having a greater impact on an individual Starfire cast then the Wrath Idol does to a Wrath cast.

At this point I am not ready to declare the IS idol to the best Moonkin idol. I need to do some rotation testing to find out what is best, but it is clear that the IS idol improved relative to the nuking idols. When 3.2 is dropped I assume that many people will still stick with the SF Idol, but there will be some situations where the IS idol is better.


Of course, all of this is assuming that Blizzard doesn't add a new Idol in 3.2. So far, there is no official indication that this will happen, but I am hopeful for a couple of reasons. First, they've added new Idols with each Tier of WotLK gear. Second, they are Making Emblem's of Conquest and Emblem's of Triumph much easier to obtain in 3.2. Blizzard has said many times that they don't like them to drop from bosses, and I suspect that Blizzard likes Idols to be accessible by players of all levels. Therefore, I think 3.2 is a perfect time for Blizzard to introduce an Idol that impacts multiple Moonkin spells.


TL:DR Version:

In terms of DPS the change to Eclipse is pretty big. It will buff Moonkin DPS by around 6% - 8%.

It also has a big impact on how we will value stats after 400 Haste rating, but this will have little impact on the gear we pick up for Ulduar. Most of the gear doesn't have a lot of haste on it, and it is still important to get to the 400 soft haste cap.

The down side to this change is that it has significantly nerfed the nuking idols indirectly, and the IS idol may actually be viable when 3.2 comes out. It will be especially useful on fights with multiple targets that can be DoTed and if you have the 4T8 Set bonus.

20 comments:

Negathle said...

I like you, am still in shock about 3.2. It left me disgruntled for several days - my fiance commented on it, it was so pronounced.

It's unbelievable that they are pushing new content on us so quickly.

Tsunami said...

Looking forward to your comments on the new Idol =)

Druid T9 Balance Relic (Moonfire) (Class: Druid) -- Each time your Moonfire spell deals periodic damage, you have a chance to gain 200 critical strike rating for 12 sec.

It will entirely depend on its uptime, but seems like it will solve the problem of Idol switching \o/


Also looking forward to seeing what the new tier pieces look like on itemisation, 4 set is certainly a lot weaker than the 4 set from tier 7 :/

Hotball said...

I'm a bit skeptical. If using 1 spell = 0.75 crit (an overestimation already), 200 crit is roughly about 150 spell. That's already inferior to current shooting star idol. Of course, shooting star idol works only for star fire and that's probably account for 50% ~ 75% total damage for a Lunar rotation. That means the new idol better have at least 50% ~ 75% up time to compensate. If there's no internal cd, then the proc rate should be at least 30% for 75% up time, and 16% for 50% up time.

Raysa said...

"Who would have thought the IS idol would come out ahead?"

I did! :D

Tradyk said...

Graylo, just thought you should know, I was just on the PTR, testing the new eclipse out, and is currently not working.

Easy enough to test, proc eclipse using wrath, keep spamming SF, once the buff wears off, you won't get the buff when you crit with SF.

Also, the new MF idol does work, and is surprisingly useful. Gives about 4% crit, haven't had time to work out it's % proc chance or internal cooldown yet, still working on that. From the looks of it though, the % chance is fairly high.

Anonymous said...

Are the numbers on the idols considering Starfire spam during heroism and tier-8 4pc Starfire procs? I would assume the Starfire idol will still win out if you factor in those two cases.

Tradyk said...

Just to update, the idol has no internal cooldown, currently, on the PTR. Expecting this to change though.

Johnny said...

i just read that you are expecting roundabout a 6 % damage increase for the owls. Does that include the t8 2piece set bonus ? or do balance druids get additional benefit of it?

nin said...

I expect a serious nerf to Mimirons hardmode before 3.2. Otherwise there will be an exceptionally low amount of those proto-drakes around.

I'm surprised that we are considered as "a little low" on damage in current content. Compared to melee, sure. With little to no buffs for elemental shaman and shadowpriests, I don't see how we wont leave them in the dust come 3.2 on single target content.

Anonymous said...

"With little to no buffs for elemental shaman and shadowpriests, I don't see how we wont leave them in the dust come 3.2 on single target content."

Apparently you haven't raided with very many good elemental shamans or shadow priests :-(

nin said...

I have actually, both our elemental shaman and shadow priests are well represented on the top list of reported results. It does not take a genius to see how reliant shadow priests are at using mind sear to achieve good results, which is why I specifically said single target.

If you look at specific results in hard modes where single target damage is relevant, ie damage to Steelbreaker, XT heart or XT himself you will see how moonkins score way better than priests, and mostly on par with elementals for the same encounters. This is fairly consistent through many, many of the top parses.

This is not extremely relevant to the topic however.

Graylo said...

@Tsunami

I just saw those changes this morning. My first impresion is that the new Idol will be the best, but it does depend on the proc rate. The set bonuses don't wow me. I'll have to do the math on 2T9, but I can tell that 4T9 is underpowered just by looking at it. It is definately not worth dropping 2T8 for and I doubt the itemization can make up the difference.

@Hotball

First why do you doubt my valuation of crit? Second, we won't be using a lunar rotation when this idol is available. According to my model we will only be using SF around 48% of the time in a twisting rotation. And this new idol will affect about 86% of our rotation. Even more with 2T9. I think the new idol is very promising.

@Tradyk

Intersting, hopefully they can fix it.

@Anon1

My numbers assume 4T8, but exclude Heroism.

@Johnny

My numbers do include 2T8.

@Nin

I wouldn't be surprised if they nerf Mimiron's hard mode, but I'm not holding my breath.

nin said...

Re: the 4T9 I think Adorielle has a very good point, it does seem like a blanket placeholder for all the classes 4T9 sets. If that is the final design of the set bonus then I'll be very, very surprised. Blizzard has a pretty good trackrecord of designing set bonuses, and we players have a fairly mixed record of judging them prematurely :)

Hamlet said...

Yeah, I'm seeing around 7% against Patchwerk. That's using a slightly low DoT-uptime rotation with refreshes only after Eclipses end. The difference between different DoT rotations might be negligible though; the opportunity cost of using DoT's more has increased now that there's no post-Eclipse phase.

Another important point is that the 3.2 rotation is not only "easier" (it's not really hard currently to switch spells when SquawkAndAwe tells you to), but less susceptible to interruptions. You're more likely to have an Eclipse ready whenever you get back on target, which helps a lot and should amplify the DPS increase in practice on complex fights.

Idol swapping isn't looking useful anyway, too small a gain for the lost DPS time.

I'm not able to confirm your result on the IS Idol. It still looks quite a bit weaker than the Starfire one. What kind of IS uptime are you assuming? And to confirm, you should be seeing an increase of 121.2 damage per tick from the Idol with full buffs (374/5*(1.03)*(1.03)*(1.04)*(1.13)*(1.3)). That's a 60 DPS increase even with full uptime; it's odd if you're finding the Starfire Idol to be weaker than that.

Villainus said...

MMO is reporting that Balance of Power has been reworded since the patch notes first hit. It now reads that both OUR chance to hit and our chance to BE hit is 3/6%. Suddenly we're getting an extra 2% hit from talents.
I'd think this is a typo amidst the confusion of patch news but if not that's 52.6 hit we no longer need to gear for...

hamlet said...

It's a typo; already been tested on PTR.

Graylo said...

Just to let you guys know, I have post coming out soon with my reaction to the new information. I'm just waiting on some numbers to make me more confident in my thoughts.

@Nin

That was my second thought. It has to be a Placeholder.

@Hamlet

I calculate my IS number a little differently than you have listed. I applied the spell coeffecent as you indicated I need to a while back, and I didn't apply as many of the universal buffs. However, I will take another look at my numbers to make sure they are correct.

rinku said...

the idol has like 30-40% procrate without internal cooldown. did a 6 minute test on heroic dummy and the effect never dropped down.

eclipse doesnt seem to work yet.

ill try to do a longer (15-20min) test with combatlog to determine the actual procrate.

rinku said...

fight length 14 min 30 seconds.
Blessing of the Moon Goddess uptime 14min 25 seconds.

270 dot ticks
BotMG refreshed 206 times.

procrate would seem to be REALLY high so hopefully their goal is that this is passive buff if one just bothers to keep moonfire up.

nin said...

Lo and behold, Mimiron is nerfed :D