Moonkin PvE Discussion, 3.3.3 - Page 5

Myllo · 03/12/10, 2:41 PM

Are these trinkets Calcs assuming Reign procs every 2 seconds (the ICD).

conghaile · 03/12/10, 3:21 PM

Now that it looks like it has been determined that the best time to cast IS is at the end of Lunar Eclipse, I was thinking about whether it should be cast just before the eclipse ended or if a last starfire should. I trotted over to Hamlet's TTT to see the current recommendation and saw this:

For example, if Lunar is just ending and Insect Swarm is down, you can cast one extra Starfire and then Insect Swarm. During the GCD from the Insect Swarm, watch to see whether the Starfire crit (it's actually useful to watch for Nature's Grace, as that buff will appear before the Solar Eclipse will). If it did, you can immediately switch to Wrath for your next spell. This should become second nature eventually.

I thought I had a counterpoint to this strategy: because Blizzard wanted to eliminate the golfing from patches past, the solar eclipse cooldown sometimes lasts longer than the lunar eclipse itself. If you make the decision with [eclipse time remaining < Starfire casttime] to cast another starfire and check during the Insect Swarm GCD for a crit, it may crit and not actually proc solar eclipse. When this happens it is extremely annoying, and it might be a dps loss. I was wondering if there was any validity to the theory, so I did a thought experiment to try and bear it out (Spoiler alert: I lost).

Assume eclipse is just ending, and the last starfire cast was a crit as it most likely should have been. This gives 3 seconds until nature's grace falls off. Assume that the first starfire cast doesn't crit. If cast starfire-IS-starfire is cast, there are two GCD lags, and if the caster has enough haste to overcome the GCD lags that second starfire still falls under Nature's Grace. If IS-starfire-starfire is cast, there are still two GCD lags and haste might push that second starfire under Nature's Grace. The two scenarios are a wash.

However, if starfire is cast first and it crits, there is a chance it won't proc Eclipse. The caster will see this and not cast a misplaced spell, but Eclipse won't be activated yet, giving rise to possibility of an extended Eclipse downtime.
If Insect Swarm is cast first and the next Starfire cast crits, it is certain a misplaced Starfire wll be cast, although Solar Eclipse is up regardless.

I don't have an exact grasp on the typical cooldown overlap, but it feels to me to average about a second (I would say that's probably generously long in favor of my theory). Cast time of starfire under eclipse for my current gear is 1.9sec, and crit chance is about 65%. So the probability of Eclipse downtime going beyond the second starfire cast in the recommended scenario is 1.0/1.9*(0.65) = 34.2%, at which point the DPET difference between a regular Starfire and an eclipsed IS'd Wrath is lost, which is about 3000 DPET in WrathCalcs for my current gear. The chance of going beyond the third (multiply by .35 again) is 11.9% at which point 6000 DPET has been lost (Nature's Grace is gone at this point). 4.2% chance to lose 9000DPET. Going no further than that an average of (.342*3000+.119*3000+.041*3000)/(2.3+2.3+1.9) = 232 dps is lost. In my counter-scenario, 3000/(2.3+2.3+1.9) = 461 dps lost over that same time. Big difference. My theory does not hold up.

NOTE: After I wrote this, I realized if starfire crits as the GCD from IS is ticking away and the caster is using Nature's Grace instead of the overhead moon to determine whether to move on to Wrath, then there WILL be a miscast Wrath. This changes the calculation a bit (note the Nature's Grace dropoff moves up a cast). DPET difference of a Wrath under IS only for me is 2500DPET, so the calc becomes:
(.342*2500+.342*3000+.119*3000+.041*3000)/(1.0+2.3+2.3+2.3) = 299 dps versus 3000/7.9 = 380 dps. My theory is still disproven.

EDIT: Had to fix my math after an epiphany in the shower this morning. Fixed mistake in bold (used to be 1-0.65), the rest of the numbers follow after that.

• Hamlet · 03/13/10, 12:54 AM

New WC up:
http://elitistjerks.com/attachments/...0312_3.3.3.xls

You can select DoT refreshing pattern independently for each DoT now. You do so on the front page. Also, you can simply select "unused" from that menu, rather than having 4 separate rotations fully laid out in the sheet for various combinations of DoT usage. This makes the sheet somewhat smaller and faster.

Handling Glyph of Starfire:
To avoid artificially inflating the value of always refreshing Moonfire quite so much, it now assumes that only half of all Moonfires get extended if you use that refreshing pattern. A guess, but better than nothing.

aceofsween · 03/13/10, 8:20 AM

The results from the new WrathCalcs are very interesting. You can come up with several different combinations and still come out with relatively similar DPS results. The best still seems to be MF/SF/SF and After One Eclipse, but using IS/SF/SF and setting IS to refresh After Every Eclipse is a loss of about 6 to 8 DPS, well within a margin of error (btw I'm not sure what WrathCalcs margin of error is to be honest).

Considering that the results are in fact lower and that you provide a unique buff with MF/SF/SF, I definitely am in favor of this particular rotation.

Videl · 03/15/10, 4:09 AM

Additionally, that rotation has overall lower dot uptime which has some non-negligible dps value in actual play, though it's very difficult to quantify.

aceofsween · 03/15/10, 2:54 PM

Since most people these days are using their left over talents to pick up Owlkin Frenzy, would it be feasible to include it in WrathCalcs with an uptime percentage? Certain bosses (particularly Blood Queen) give us a respectable uptime. It also seems to proc oddly for abilities that randomly target raid members, like Putricide for Malleable Goo and Frozen Tomb, even if they don't actually target you yourself.

Sadly, the pulsing frost aura by Sindragosa doesn't share the same quality as Blood Queen which I really cannot explain. Neither does Mystic Buffet. I'm basing this on memory, so it's possible I could be incorrect, but as far as I can recall neither of these abilities do.

Has anyone taken the time to collect and figure a decent uptime percentage for the various bosses?

Voek1 · 03/16/10, 12:11 PM

I spend some time looking at WrathCalcs' DPETs during Bloodlust/Heroism and I noticed that during a Solar Eclipse Starfire's DPET is slightly below that of Wrath. However, if MF is not up it can be better to cast MF followed by 3 Starfires and then switch back to Wrath, since the extra damage from MF more than makes up for the small loss by casting Starfire during Solar. Even if you don't have enough time left in Solar to cast those 3 Starfires, then you are likely to extend the MF in the upcoming Lunar before the MF drops off. There is however an (undesired?) side-effect to consider, i.e. casting MF early during a Solar has an effect on future MF refreshing. Moreover, if you know in advance that during a Bloodlusted Solar you will refresh MF, then this also affects the optimal pre-Solar refreshing of Dots.

It seems to me that using the following (partial) set of rules gives the maximum benefit of MF during Bloodlust:
A. If Solar Eclipse is up AND Bloodlust is up AND MF is not up, then:
1) if time in Solar left < time to cast MF + 3 Starfires, then cast MF and as much Starfires as possible (without losing potential Wraths that can proc Lunar)
2) if time in Solar left > time to cast MF + 3 Starfires, then wait till time in Solar left = time to cast MF + 3 Starfires and then cast MF and 3 Starfires.

The logic behind rule A2 is that you only want to fit in 1 MF and 3 Starfires and that delaying your MF helps to increase its up-time in the upcoming Lunar, which is good since you dislike refreshing MF during a Bloodlusted Lunar. If the up-coming Lunar won't be during Bloodlust, then I am not sure what is optimal.

B. If Solar Eclipse is up AND Bloodlust is up AND MF is up, then:
1) cast as many Starfires needed to fully extend the current MF, then see A.

Is there a flaw in my reasoning or are there better ways to deal with Bloodlusted Solars?

Yijiao · 03/17/10, 4:19 AM

I was curious if anyone has done the math over when it becomes appropriate to forego any casting of wrath during heroism(with the exception to reproc lunar) because starfire becomes better dps at higher levels of haste...

In raids, during hero, my starfire gets down to 1.33 and I have been experimenting with continuing starfire the whole time with good results in terms of output.

Drui · 03/17/10, 5:36 AM

Good question. I often got better results in just staying with Starfire while Bloodlust ist up or switch to a 1 SF 1 Wr Rotation to get less clipped Wraths. Could someone do the correct Maths for us? :-)

aceofsween · 03/17/10, 9:55 AM

The answer to these questions is already solved... Set Bloodlust to TRUE under the first page of Wrathcalcs and then go to the rotations page and then just look at the DPET.

Wrath's is mostly unchanged (it does go up a little bit), but Starfire's jumps up significantly. For me, Eclipsed Wraths still do more damage than a Starfire with Bloodlust, but the difference is minor (300 DPET). For someone with more haste, it's definitely possible for Starfire to outstrip Wrath.

Voek1 · 03/17/10, 10:47 AM

Originally Posted by aceofsween

The answer to these questions is already solved... Set Bloodlust to TRUE under the first page of Wrathcalcs and then go to the rotations page and then just look at the DPET.

Wrath's is mostly unchanged (it does go up a little bit), but Starfire's jumps up significantly. For me, Eclipsed Wraths still do more damage than a Starfire with Bloodlust, but the difference is minor (300 DPET). For someone with more haste, it's definitely possible for Starfire to outstrip Wrath.

If I am correct then the DPETs are based on spamming the relevant spell. During Bloodlust the time to cast a Wrath is 1.1 seconds including the proposed lag/reaction values in WrathCalcs. However, if a Wrath is preceded by a Starfire, then you can queue the Wrath and reduce the cast time of that Wrath by 0.1 seconds to 1 second.
Suppose the DPET of a Starfire is slightly above that of Wrath during a Solar Bloodlust, then (ignoring Dots) rotating Starfires and Wraths can be better than spamming the highest DPET spell, Starfire.
Basically, a Starfire increases the DPET of the next Wrath by roughly 10%. So, simply comparing DPETs during Bloodlust is incorrect.

Please correct me if I am wrong.

Edit: On a second thought, by the same logic every Wrath reduces the DPET of the next Starfire, so this idea of using Starfires to queue Wraths might not help at all.

aceofsween · 03/17/10, 1:03 PM

That's only half correct though. You can't queue your next Starfire during Wrath, which essentially increases the cast time of your Starfire by .1 second. It would work out like this:

Starfire (1.3) + Wrath (1.0) + Instant Delay (0.1) + Starfire (1.3) + Wrath (1.0) + Instant Delay (0.1) + ...

You're still getting the instant delay, it's just being applied to Starfires instead of Wraths (with the exception of the first Starfire). Whether that actually works out to be better than just spamming Wrath or Starfire, I'm not really sure, but somehow I doubt it. What might work out better though would be to watch NG and only use Starfire with NG is up, and using Wrath while it's not since you're not gaining anything from NG while casting Wrath.

Yijiao · 03/17/10, 2:43 PM

Well, with starfire at about 1.3, you can get two non-crits in a row and still keep up NG.

Assuming your starfire crits 60% of the time, the chances of not critting 3 times in a row is only 6.4%, so you could probably safely spam.

nzoq · 03/17/10, 4:30 PM

How valuable is delaying a FoN cast to line it up with procs and other cooldowns?

I was unable to find information pertaining to FoN scaling in the guide. I've always assumed they scale with Spell Power, but I didn't know exactly how.

I performed a very quick test of one cast of FoN in three sets of gear: moonkin gear, bear gear, and no gear. Here were the results (all casts were in Moonkin form)...

Moonkin gear (1305 Stam, 3535 SP, 2208 AP)
Average "hit": 550
HP: 7430
Bear gear (3112 Stam, 76 SP, 5539 AP)
Average "hit": 383
HP: 9868
No gear (100 Stam, 67 SP, 170 AP)
Average "hit": 367
HP: 5805

In all cases, the trees made about 50 attacks on the Grandmaster dummy. I think the sample size is too small to make any conclusions about Crit or Hit scaling, and Haste is too hard to compare due to positioning issues...

I did some further experimenting and (as expected), the trees seem to snapshot your stats when they're summoned. Assuming linear scaling with spell power, popping FoN at the peak of a trinket like Dislodged Foreign Object (1210 SP heroic) would be worth about 60 damage per hit, a 15% damage increase.

Has anyone performed more extensive testing?

Yellowsix · 03/17/10, 6:27 PM

Info on pets is always tough to come by... the Team Robot simulator uses the following for treants, which was determined from some rather quick estimates done in-game by inspecting the pet info:

base attack speed: 2.0

- Treants get some kind of scaling from the caster's haste gear, but I couldn't quite figure it out... right now I just hacked it and estimate that treants get 50% of the caster's spell haste, but not nature's grace; they can also be buffed by heroism after they are summoned. If anyone has better information, please correct me.

The treant damage formula that we use is:

([base damage]
 + (
     (311
      + ((331 + [raid strength buffs])
         * [raid strength multiplier buffs])
      + [raid AP buffs]
      + (0.536 * [player spell power])
     )
     * [raid AP multiplier buffs]
     / 14
     * [attack speed]
   )
)
* [raid physical damage multipliers]
* [brambles]

Translating some of that out of code speak:

- 561.58 "base" weapon damage
- 311 "base" AP from somewhere
- gain 1 AP per strength, and have 331 base strength + raid modifiers
- gain AP from raid buffs
- gain AP multipliers from raid buffs
- gain 53.6% of the druid's spell power as extra AP
- damage is multiplied by brambles
- damage is multiplied by physical damage raid buffs and debuffs

Don't take any of that as gospel, and please correct me if any of it looks incorrect.