Dorlath

I've seen a lot of suggestions for aoe builds for leveling to 80, and I certainly think they're a good idea in general. However, I remember back to how crowded it was everywhere with so much competition for quest mobs right after the release of TBC. I suspect many people will face similar issues if they level their druid right after the expansion comes out and that would probably seriously limit the ability to get a group of mobs in one spot to aoe.

So I'm really thinking it may be best to stick to a build that focuses on single target dps for leveling.

 

Khalanis

A nice change to eclipse would make it proc a buff to start the duration.

Make eclipse an undispellable buff that lasts for a few seconds that applies the benefit to your next wrath/starfire(depending on which procced it) and puts a buff on you(once you've cast the triggering spell) that lasts x seconds after that. This way, even though it is a proc it gives the player a bit more control on how it's used so they can gain full benefit from it. This also means you won't be wasting any proc time on a different spell, and allows you to apply IS/MF per the appropriate spell to take advantage of imp IS.
Similar to the 4 piece pvp set bonus, procs with flexible activation time.

 

Erdluf

If IS falls off during Eclipse, you can refresh it immediately, or wait until Eclipse fades and then do the refresh.

I took Adoriele's spreadsheet, changed Eclipse wrath to 20%, and added IIS and the IS glyph. It shows

IS tick = 629, so every 2s you delay casting Wrath costs you 629 damage.
IS cast time = 1.49s. (after haste and latency)

Wrath cast time = 1.36s
Wrath damage (Eclipse, no IS) = 3415
Wrath damage (Eclipse, IS) = 3518

SF damage (MF up) = 5347
SF cast time = 2.52

Comparing: ('n' goes from 0 ... 4)
IS Wr*n SF ...
Wr*(n+1) IS SF ...

Bonus damage from extra Wrath cast is 3415

Penalty from delayed IS is (n+1)*1.36/2*629 = 428, 856, ...
Penalty from casting wrath without IIS is n*(3518-3415) = 0, 103, ...
Penalty from delaying the SF cast is 5347*1.36/2.52 = 2886

If n is zero (IS expired near the end of Eclipse), the penalties add up to 3313, and there is a very slight advantage to casting the extra Wrath.

If n is one or higher (meaning you have 3 or more seconds left on eclipse) you are better off refreshing IS immediately.

Also, casting IS during eclipse means casting SF sooner, meaning your MF DoT is a little bit safer against running out.

 

slazareth

Imo a better change for eclipse would be to make it a stack that lasts longer. So lets say the stack lasts one minute and has 10 stacked. Your next 10 wraths/SFs will have the eclipse buff. If you do not use them up before the minute is up, then they are gone. Now I don't know whether one minute or 10 stack is the right amount, but I think that sort of implementation would be a lot better than the current and would allow them to tweak the balance a lot easier.

 

Nilaus

Awesome idea, it is definately possible to implement since it is basically the same functionality as Combustion.

 

Polishedhead

They do indeed need to do something to make us not lose 1 second/1 cast on reaction time.

Take Fury Warrior's Bloodsurge for example:
Your Bloodthirst critical hits have a 100% chance of making your next Slam instant for 5 sec.

It gives us 5 seconds after a BT crit to react and use the instant slam, we aren't wasting any time. We can even wait 2 GCDs before taking advantage of the BT crit. They should do the same with eclipse, making the first starfire/wrath cast within 5 seconds trigger the 10 second buff like someone suggested above.

 

ranma

How did you get 35 seconds average proc rate. From my understanding, eclipse cooldown is 40 seconds. I think you might count the time w/o proc is active. However, if you want to calculate average damage boost, you should count the total time, not the time w/o proc only.

So, made it 20%*6.8/45= 3.02%, count filler only, it is 2.26% boost for 3 talent points.

 

Erdluf

Good catch. 2.26% for three talent points is not too good. However with 2 or 1 talent points that 45 goes to 51 or 57
I'll also drop that "filler" number to 70% and get

3/3 gives 70%*20%*6.8/45 = 2.1%
2/3 gives 70%*20%*6.8/51 = 1.9%
1/3 gives 70%*20%*6.8/57 = 1.7%

It looks like a pretty good 1-point talent (with these assumptions). The other points should probably go somewhere else/

 

ranma

Do you mind tell me how you get 5 seconds as time between when proc cool down is end and when next proc is up?

I know w/o haste, it takes average 2 SF to get one crits to get the proc, make it 6 seconds. and your next numbers confirmed that by adding 6s for each 1/3 chance of proc of crits.

Furthermore, for 1/3 eclipse, on average, you get 2 unproced crits before proc. So, you got 2 nature's grace before proc. It is -1s total on your average time, make it 56s total.

 

skeldi

Too difficult to play with as a 1 point talent. (read: unreliable)

 

Neshalin

It's not so hard if you're following a Moonfire/Starfire rotation. At some point Eclipse procs, you finish your 2.5-ish sec Starfire, then have at least 7.5 seconds to cast a stronger Wrath (5 to 8 casts depending on haste). You have to watch the proc anyway; it doesn't really matter what the chance is.

 

Blacknwhite

Some addons like Power Aura will help you to don't miss eclipse proc.

 

bavelb

I find it quite hard to miss the Eclipse proc on other moonkins (somethign about that omnious black flaming circle jsut screams "watch me), let alone myself. Regardless, I think I will invest one point in it in it's current incarnation. It certainly isn't a DPS loss for that one point unless you mess up your insect swarm uptime and it mixes things up.

I do worry about the times that misery isn''t available and you need to mix in faerie fire with a close, but not similar dropoff time compared with Eclipses cooldown.

 

Erdluf

When the proc is up you are probably in the middle of an SF cast, and that cast might crit. If that happens you could get the proc 41s after the previous one. On the other hand, you might be busy refreshing IS and even FF. Finally, at 80 you are going to have some haste.

This is napkin math. The point of the napkin math is to find out if the talent is worth taking. My assumptions are:

1) High SF crit rate (40% or more) and main rotation is SF or SF+IS.
2) You pay attention to the proc and switch to Wrath.
3) You can usually take advantage of the proc (you don't waste proc time on movement, battle-rez or other silly stuff).
4) Your Wrath-spam DPS (non-eclipse) is within a few percent of your SF-spam DPS
5) The 6.8s number assumes a certain level of latency, haste, and Nature's Grace interaction with the GC.

with those assumptions, the first point of Eclipse is a good talent point. It is certainly worth more than a 1% haste or 1% crit talent. It looks like the second and third points are bad (as in Genesis bad) at well less than 0.5% DPS per point.

Suppose filler damage is only 60%, and at 1/3 Eclipse the proc gives you 20% extra filler DPS for 6s every minute.

60%*20%*6/60 = 1.2%

even with these assumptions, the first point is a "good buy" compared to other DPS talents you may be taking.

Precision
Someone needs to do a more detailed analysis (or run a simulator) to get precise numbers (and even precise numbers aren't necessarily an accurate model of an actual raid).

It is possible that you will need to refresh IS during the Eclipse buff. That would probably mean you only cast four Wraths instead of five. It is possible that you get the buff when you were going to cast IS anyway (rather than another SF). In that case you will have time for six (or possibly even seven) wraths.

I was using 6.8/(time_between_procs), but not all of the time between procs was spent on SF. The more precise formulation would be:

(filler_dps_increase)*(dps_increase_time)/(total_filler_time)*(total_filler_damage)/(total_damage)

If filler_time/total time is only 90%, my answers were 11% too low (so 1.7% becomes 1.9%).

 

thedopefishlives

I'm actually going to attempt to model Eclipse now in the current version of Rawr.Moonkin. It's the only talent I haven't really modeled (well, aside from Owlkin Frenzy and anything having to do with AoE). However, I believe this is one of those cases where a simulator might tell us much more than straight theorycraft.

 

• Adoriele

I'm also putting inital calcs into the sheet, since I have a little free time at work (yay for finally getting sticky situations solved). Note that all calcs based on anything but a true spam rotation will be VERY ROUGH. Yes, this even include SFxinf, MF rider, as until you get to a certain crit/haste point, there's a chance your MF will drop off during Eclipse (obviously after that point it will be functionally identical to SF spam itself, and I'll treat it as such). I'll be modeling based on the following assumptions:

Eclipse lasts 10s. You will cast your spam spell once before switching to take advantage of it. You have a 10/duration chance of needing to refresh each DoT you have up, and both are going to be worth refreshing during eclipse. There are three states that you can be in, two of which are functionally identical, but of differing length: Normal rotation under Eclipse cooldown, lasts 30s; Normal rotation not under eclipse cooldown, lasts 1/Eclipseprocchance casts + spamspellcasttime; Eclipse rotation, lasts 10-spamspellcasttime.

 

Olddrippy

This would probably be too good, but maybe instead of two separate buffs......receive a stack like inner fire that gets consumed as you cast. The stack would have a short time limit so if you don't use the stack soon it expires. This way you would get your certain number of improved wraths if you use them in a reasonable time, but has just enough room to refresh moonfire or insect swarm at any time during the wrath spam because it has a fixed number of wraths that benefit.

 

Spiry

With Rawr being the feral sim, shouldn't the Moonkin one be called Squawk?

 

• Adoriele

Also, interesting to note: When the SF-boosting Eclipse is up, SF spam generates more mana than it costs. News at 5.

 

Arazan

Now *that* is interesting. Anyone done the math on whether a 71/0/0 type build would be sustainable because of stuff like this?

 

thedopefishlives

You'd think that, wouldn't you?
This doesn't really surprise me. In the current source version of Rawr, I've implemented some new charts, two of which show various aspects of mana regen. One shows your current talents as mp5; the other is a SimulationCraft-type regen chart showing your mana gain from various internal and external sources. One thing I've noticed is that haste is actually a mana-positive stat at certain gear points (don't ask me which, yet!). That is to say, in at least the test cases I've looked at, your haste talents actually have a (albeit small) positive mp5 value associated with them.

 

• Adoriele

Hmm. Only way I see that being true is in its interaction with JoW. You're going to get a proc per X casts, always, but those X casts take less time. Except when a point of haste moves you from X casts to X+1, which is going to be a huge hit to its benefit. Moonkin Form procs are more accurate as a reduction of the cost of a spell, and as long as you're not proccing enough to make the spell itself worth positive mana, haste would be a negative mana stat wrt Moonkin Form. All passive regen is haste-independent. OoC would get the same treatment as JoW, but the huge negative jumps would be much more frequent (it's much easier to go from X to X+1 casts per proc with OoC than with JoW).

[edit] Putting up first version with minor Eclipse support. Only the true-spam rotations are modeled, but preliminarily, for those rotations, Eclipse is only worth ~1 point. The first point, and this is no joke, adds 90 DPS to Starfire spam alone, 131 DPS to Wrath spam. The total after 3 points is 114 and 162 DPS respectively.

Attached Files

WrathCalcs.xls (113.5 KB, 91 views)

 

thedopefishlives

It might be my implementation of Moonkin Form, which, instead of doing a mana discount on the spell like your spreadsheet does, uses a mana-return mechanic like JoW but calculated in a different fashion. (As an aside, I might look into reimplementing this.) After a little experimentation, it is clear that JoW is the main determining factor in haste actually providing a mana benefit, and the plateaus of JoW are well-known.

 

dedmonwakeen

A question regarding the GCD reduction on Wrath crits.....
Does it also reduce in any way the MINIMUM GCD of 1.0sec?

http://code.google.com/p/simulationcraft

 

erragal

Nope. GCD is hard capped at 1.0.