Elitist Jerks Moonkin Raiding

 11/30/09, 8:46 PM #2506 • Hamlet     Hamlet Tauren Druid   Mal'Ganis I think Build-a-Talent might be venturing dangerously close to WoW forum territory. Also, I'm not sure why there's all this sudden talk about Eclipse, which isn't going see any big change until Cataclysm. The problem with Eclipse is more subtle than people are always making it out to be. Lots of classes have RNG all over the place in their mechanics. I can think of two significant things that make Eclipse different: 1) It only happens a few times a fight. Most heavily RNG-based mechanics (say Missile Barrage) are "attempted" every few seconds, so the total variance over the course of a fight is small. I know of no way to make this point without invoking the Central limit theorem - Wikipedia, the free encyclopedia , but we can validly expect to have a higher variance in expected DPS output due to less frequent trials of an important RNG roll. This effect will actually be lessened in 3.3 when we're back to trying to proc it twice as often. 2) Interruptions work heavily against us. Interruptions during Eclipse downtime delay the next Eclipse, leaving mean Eclipse uptime during actual casting constant. But interruptions during Eclipse uptime result fully in lost uptime. So it's a lose/lose situation--interruptions during Eclipse count against us, but interruptions outside Eclipse fail to counteract them, because they don't reduce the total time we have to spend proccing Eclipse.* I think (2) might be more significant. Eclipse in its current form admits little to no ability to use any of skill or judgment to ameliorate the effect of interruptions. Ordinarily a good player with some short-term buff mechanics at his disposal and a good knowledge of a boss fight that lets him predict interruptions to some degree should be able to plan his ability use to manage the situation in some fashion. But we have none, so not only are we hit unfairly by casting interruptions in the average case, but there's no way to improve the situation with attentive play. *To make this slightly more mathematical: say we ordinarily have average Eclipse uptime E, but are only able to cast with frequency x. The normal cycle had Eclipse uptime E and eclipse downtime (1-E). But with cast time deflated to x, the Eclipse uptime is still E, with only Ex of that spent casting, and the Eclipse downtime is (1-E)/x, with only (1-E) of that spent casting. The preceding sentence contains the important point--Eclipse uptime gross (including interruptions) is the same in both cases, but Eclipse downtime net (excluding interruptions) is the same--heavily favoring the downtime in this interaction. Our new average Eclipse uptime out of time spent casting will be new ratio of Eclipse time to total time (both excluding interruptions): $\frac{Ex}{(1-E)+Ex}$, which is less than the E is was without interruptions. For example, where E = .75 and x = .8 (75% nominal Eclipse uptime and 80% casting frequency), Eclipse uptime will be reduced to 70.6%.
 11/30/09, 11:13 PM #2507 aceofsween Don Flamenco     Jaboom Troll Druid   Lightning's Blade I only brought it up as a counter suggestion to what someone else proposed and the only reason I expanded upon it was because I realized that it's not a bad idea for a half baked thought. While I wish that Eclipse was a little bit more dynamic in its implementation, I think that all you've done is highlight a known factor of Moonkin raiding and I don't expect that these weaknesses will change no matter what they choose to do with the class. Although, considering the topic, we have strayed a little bit into the realm of suggestive formulating, away from actually discussing Moonkin mechanics. However, those are for the most part already heavily discussed and the lack of new discussion is probably what brought forth the slight derail.
 11/30/09, 11:21 PM #2508 Erdluf Great Tiger   Erdluf Night Elf Druid   Echo Isles Arawethion, That tells us that Moonkin now hate interrupts even more than we did in BC. It doesn't say very much about how compare to other classes in an interrupt-heavy environment. I ran the latest released simc (t9, optimal raid, 2.2). DPS Ranking: 422913 100.0% Raid 13th 10033 2.4% Druid_T9_58_00_13_2T8_2T9 16th 9922 2.3% Druid_T9_58_00_13 I then re-ran with a 4s stun, about every 30s: raid_events=stun,cooldown=30,duration=4 DPS Ranking: 370003 100.0% Raid 26th 8396 2.3% Druid_T9_58_00_13_2T8_2T9 28th 8315 2.2% Druid_T9_58_00_13 The average raid member lost about 12.5% of his DPS to the stuns. Moonkin lost about 16.3% of their DPS, taking roughly a 1/3 bigger penalty than the average raider. Eclipse downtime increased from about 17% to 22%. Not surprisingly, DK's, Ferals and Rogues weathered short stuns relatively well. They lose auto attacks, but they continue energy/power regeneration. Considering only mana users, the top Moonkin went from 9th to 14th, being passed by a Ret, 2 Ele specs, and a 53/0/18 DG Warlock. A 16/0/55 DK lost only about 5% of his DPS, and climbed from 21st place, to 1st.
12/01/09, 3:38 AM   #2509
Rannasha
Piston Honda

Tauren Druid

Draenor (EU)
 Originally Posted by Erdluf I then re-ran with a 4s stun, about every 30s: raid_events=stun,cooldown=30,duration=4
Unfortunately, this is one of the areas where simulator and actual combat diverge as a good player will anticipate the stun (especially if it's on a fixed timer) and may opt to refresh DoTs earlier or swap nukes to prevent Eclipse from proccing seconds before the stun. By working the stun-timer into your spell-selection, you can reduce the DPS-loss it causes.

12/01/09, 4:01 AM   #2510
Hamlet
<Druid Trainer>

Tauren Druid

Mal'Ganis
 Originally Posted by Rannasha Unfortunately, this is one of the areas where simulator and actual combat diverge as a good player will anticipate the stun (especially if it's on a fixed timer) and may opt to refresh DoTs earlier or swap nukes to prevent Eclipse from proccing seconds before the stun. By working the stun-timer into your spell-selection, you can reduce the DPS-loss it causes.
But part of the point I was making above is that we have very little ability to play around interruptions. Refreshing DoTs is a good one, since the most common type of interruption is movement. But only a small percentage of our DPS is from DoTs; the great majority is from nukes and the Eclipse cycle. And the most common kind of movement is relatively randomly scattered--dodging randomly targeted abilities of some kind. Most DoT refreshes during that movement are heavily clipped, so while you're making some use of that time, it's minimal. Only when the interruption is somewhat regular and timed such that it's close to an integral number of DoT durations is the benefit of this large.

This is does shade into something else which I do think helps separate good players--moving during DoT refreshes at a time when you have a choice of when to move. This actually happens pretty often, and after playing the spec for months I do it instinctively. But, due to the low percentage DoTs represent in our total DPS, the actual benefit of this kind of management is quite small.

Swapping nukes to avoid proccing an Eclipse is of limited use. It's better than in the 3.1 rotation, where there was no way to avoid proccing a nuke during the post-Eclipse phase, so at least we have the option. But 1) it only works on perfectly predictable interruptions, which is often not the case when it comes to movement, and 2) it recovers the lost Eclipse uptime in a pretty unappealing way--by sacrificing Eclipse uptime by avoiding proccing it. I won't do the math right now on how far out it might be worth it to wait, but it will still be the case a good player can't maintain average Eclipse uptime equal to what it is when uninterrupted. The key point I was making above will remain: we feel a double whammy from interruptions because we lose both total cast time and relative Eclipse uptime.

12/01/09, 9:39 AM   #2511
Erdluf
Great Tiger

Night Elf Druid

Echo Isles
 Originally Posted by Rannasha Unfortunately, this is one of the areas where simulator and actual combat diverge as a good player will anticipate the stun (especially if it's on a fixed timer) and may opt to refresh DoTs earlier or swap nukes to prevent Eclipse from proccing seconds before the stun. By working the stun-timer into your spell-selection, you can reduce the DPS-loss it causes.
Careful spell use can mitigate the problems, but our options really aren't that good. Relative DPS numbers might be roughly (napkin math, 87% NG uptime, 80% Eclipse uptime, ignoring FoN and Starfall):

Cast: DPS (time spent on a PW fight)

"Normal" nuke: 100 (2%)
NG nuke: 120 (15%)
Eclipse nuke: 130 (9%)
NG+Eclipse nuke: 160 (62%)
Full-duration DoT: 170 (11%)
Clipped DoT: 20 to 160 (0%)

Average DPS: 151

We only have two casting options that are above average: A full duration DoT, or a nuke with both NG and Eclipse. We have to maximize our use of those options.

Even a short interruption (or movement) gives us a much higher chance of losing NG, casting clipped DoT's, and wasting, or delaying Eclipse time. Any "special" preparation you take in advance of a stun will itself be significantly below PW average.

 12/01/09, 11:27 AM #2512 shibbytastic Glass Joe   shibbytastic Tauren Druid   Stonemaul 2T10 and glyph of the wild I have been wondering if it will be worth our time to occasionally use gift of the wild to proc omen of clarity when we have 2T10. It seems like it will be useful at least as a third instant to cast while moving, but does it have any other place in our rotation?
 12/01/09, 11:53 AM #2513 • Hamlet     Hamlet Tauren Druid   Mal'Ganis Using 1s to cast GotW seems a roughly balanced price to pay for 15% increase for 6 seconds. Unlikely to have any benefit during stationary DPS, though it's surprisingly close. Might have some use if you sync it with DoTs, since they'll carry the benefit for their whole duration. But casting another instant slows down the Eclipse rotation. Definitely will have a use while moving.
 12/01/09, 12:17 PM #2514 Erdluf Great Tiger   Erdluf Night Elf Druid   Echo Isles It might also give you an incentive to stand in melee range. A 2s mace or dagger should have an 11% or 12% chance to proc OOC. It will have a chance to connect anytime you cast +, and also during movement or some interrupts. It is plausible that melee range could be attractive for Owlkin Frenzy procs also. OF is huge when you can get it. The problem in current raid content is that you almost never get it.
12/01/09, 2:14 PM   #2515
Royalite
Von Kaiser

Night Elf Druid

Aggramar
 Originally Posted by Erdluf It might also give you an incentive to stand in melee range. A 2s mace or dagger should have an 11% or 12% chance to proc OOC. It will have a chance to connect anytime you cast +, and also during movement or some interrupts. It is plausible that melee range could be attractive for Owlkin Frenzy procs also. OF is huge when you can get it. The problem in current raid content is that you almost never get it.
I really hope isn't that benefical. As I see myself in a ToGC25 fight on Gormok within melee range, then the RL asks "What the hell are you doing taking extra damage being in melee range?" and hearing the raid laugh for trying to be the new melee class when I say "trying to proc OF and OCC". [/exaggeration]

Or are you suggesting while moving out of the fires/whatever melee the boss on the move type thing? Where such a benefit would be more a situational type thing and not standard practice? Just going back to the post about not wanting to wasting one GDC for trying to proc for 6 seconds but how during movement it would be great.

Not saying it isn't a good idea but trying to wrap my head around how such an idea would work in maximizing dps.

 12/01/09, 6:30 PM #2516 Lord BEEF Soda Popinski     Werebeef Tauren Druid   Mal'Ganis How decent is this trinket for us? Nevermelting Ice Crystal - Items - Sigrie I've been stuck with sundial forever and this should be pretty easy to get, but the use doesn't seem all that great with the 3 minute cooldown.
 12/01/09, 9:05 PM #2517 kerg99 Glass Joe   Trik Night Elf Druid   Hyjal I have been thinking about the wrath and haste issue. Currently heroism/bloodlust do nothing during a solar eclipse but it seems blizzard wants us to cycle wrath and starfire. One very easy change would be to change tier 1 tallent starlight wrath. Currently it reduces the cast time of wrath by .1 sec per point for .5 second total reduction wich works out to be what 25% haste? or is it 30% i was never a math person. If blizzard changed this tallent so it increased the damage done by wrath or the crit chance of wrath by 20-25% the tallent dps increase would be lower but in reality by increasing the haste cap by alot our total dps should increase. We would also be able to use a haste proc trinket any time we want rather then timing it with lunar cycle and heroism/bloodlust would give us a real increase for our entire rotation not just the lunar part.
12/01/09, 10:42 PM   #2518
sp101
Glass Joe

Night Elf Druid

Kil'Jaeden
 Originally Posted by kerg99 I have been thinking about the wrath and haste issue. Currently heroism/bloodlust do nothing during a solar eclipse but it seems blizzard wants us to cycle wrath and starfire. One very easy change would be to change tier 1 tallent starlight wrath. Currently it reduces the cast time of wrath by .1 sec per point for .5 second total reduction wich works out to be what 25% haste? or is it 30% i was never a math person. If blizzard changed this tallent so it increased the damage done by wrath or the crit chance of wrath by 20-25% the tallent dps increase would be lower but in reality by increasing the haste cap by alot our total dps should increase. We would also be able to use a haste proc trinket any time we want rather then timing it with lunar cycle and heroism/bloodlust would give us a real increase for our entire rotation not just the lunar part.
This would affect Moonkin PvP. This is why they can't (won't) bump Wrath's cast time.

 12/02/09, 1:42 AM #2519 spiritryu Von Kaiser     Ryuujin Troll Druid   Ner'zhul I got onto the PTR for the last PTR build to find their change to compensate Moonkin DPS for the WE nerf. Turns out they changed Eclipse Damage portion for Wrath has been increased to 40% and the crit portion for Starfire has been increased to 40% 10% increase for both looks like it's just enough to make up for the loss of WE. Last edited by spiritryu : 12/02/09 at 1:47 AM.
12/02/09, 3:48 AM   #2520
Hamlet
<Druid Trainer>

Tauren Druid

Mal'Ganis
 Originally Posted by spiritryu I got onto the PTR for the last PTR build to find their change to compensate Moonkin DPS for the WE nerf. Turns out they changed Eclipse Damage portion for Wrath has been increased to 40% and the crit portion for Starfire has been increased to 40% 10% increase for both looks like it's just enough to make up for the loss of WE.
Nice buff. Looks like a typical person who switches from 2T8/2T9 to 4T9 on patch day can expect a small DPS increase, even with the loss of WiseEclipse. Here's a sheet with the new value.
Attached Files
 WrathCalcs 091202 3.3.xls (193.5 KB, 417 views)

Last edited by Hamlet : 12/02/09 at 3:55 AM.

 Elitist Jerks Moonkin Raiding