Jump to content


Photo

Moonkin PvE Discussion, 3.3.3


  • This topic is locked This topic is locked
328 replies to this topic

#41 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11,540 posts

Posted 11 March 2010 - 11:12 PM

Right, the spellpower was buffed. But since the ICD is actually higher than we'd thought it was, the two effects mostly cancel out.

#42 Guest_aceofsween_*

Guest_aceofsween_*
  • Guests

Posted 11 March 2010 - 11:34 PM

That seems completely counterproductive. Wasn't the idea of buffing Phylactery because it was under budget (as in it's average Spellpower came out worse than similar trinkets... as in DFO)?

Also conghaile, I still believe that there is a point toward the end of a Solar Eclipse where refreshing Moonfire might be the better option, but the difference is pretty minor. Plus, you also run the chance of having to reapply it before you even proc Lunar eclipses due to the RNG factor. It's more of a trade off than anything else. It'll still result in roughly similar uptimes of Moonfire, so the overall effect is more or less the same.

Finally, Arawethion, are you including Languish in those calculations? I'm fairly certain I remember reading reports about these ticks proccing Phylactery as well. With Languish, you have roughly 1.3 ticks per second, although because Languish can be refreshed before it ticks, I'd say it's closer to 1.25 ticks per second. I'll be honest, I'm not sure how to interpret your "5/6 DoT ticks per second" number.

Edit:
With the Simulator update, I wanted to also give a quick update on my rotation numbers:

Rotation 1: 11634.0 DPS +/- 31.0

Glyphs: Starfall, Starfire, Insect Swarm
Moonfire: Target does not have Moonfire, you do not have Solar Eclipse
Insect Swarm: Target does not have Insect Swarm, you do not have Lunar Eclipse


Rotation 2: 11637.6 DPS +/- 31.4

Glyphs: Starfall, Starfire, Moonfire
Moonfire: Target does not have Moonfire, you do not have Solar Eclipse
Insect Swarm: Target does not have Insect Swarm, you do not have Lunar Eclipse, and Solar Eclipse is ready


As you can see, the simulator is pretty up in the air for both rotations. Both are easily within the margin of error and so close that it is almost impossible to determine which one is truly better. The difference between the two essentially comes down to how often you apply Insect Swarm. The first rotation has about a 75% uptime of Insect Swarm while the 2nd rotation has about a 40% uptime.

#43 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11,540 posts

Posted 11 March 2010 - 11:46 PM

My guess is that the ICD was always 100s and nobody ever tested it properly, since we'd all assumed it was 90 beforehand, and the difference isn't noticeable at a glance.

When I did trinkets last patch, I assumed constant DoT refreshing. So each second, you have half an IS tick and 1/3 of a MF tick, so 5/6 DoT ticks total. 1/(5/6 * 30%) = 4, whch is where I got the 4s mean proc time from. The 6s above I totally pulled out of my hat since we don't know get what actual DoT uptimes will look like yet in 3.3.3.

#44 Guest_aceofsween_*

Guest_aceofsween_*
  • Guests

Posted 12 March 2010 - 12:13 AM

Ah, okay. I follow what you're saying now.

I believe I may have answered the question about DoT uptimes already at least. I can't find a rotation that generates higher DPS from the Simulator, although we'll have to wait to see how things work in game I suppose. Either way, it's safe to assume. Uptimes for the 1st rotation are roughly 76% for both Insect Swarm and Moonfire. For the 2nd rotation, the uptime of Insect Swarm drops to 40%, but Moonfire stays up around 76% because there's no real change there.

Additionally, while Languish has a 92% uptime, it only ticks 62 times over the course of the 300 second simulation, giving you roughly 1 tick every 4.8 seconds, much lower than I thought it'd be. My haste is just barely at the point where it will refresh it constantly during a Lunar Eclipse cycle which likely contributes to that.

It's probably best to break up the proc rate between Solar and Lunar phases though. During the Pre-Lunar and Lunar phase, you'll only have Moonfire and Languish up. During Pre-Solar, you'll have Moonfire, Languish, and Insect Swarm up. And then in the Solar phase, you'll have Insect Swarm for the most part and Languish with Moonfire about to expire in about 6 seconds or so.

#45 Yellowsix

Yellowsix

    Piston Honda

  • Members
  • 166 posts

Posted 12 March 2010 - 02:14 AM

For Languish, I don't have a moonkin with 4T10 that I can test it with myself... so I'm not 100% certain that I did it correctly in the simulator. Here were my assumptions:

- One application of Languish ticks twice at a 2 second interval.
- When Languish is applied, any damage from an existing Languish that has not ticked yet, gets averaged over the 2 ticks of the newly applied Languish.
- When Languish is applied, it starts the 4-second timer again, e.g. if Languish is applied at time 2.00, it will then tick at time 4.00 and 6.00.

Note that because of the 3rd assumption, if you were to get a crit from either starfire or wrath more often than once every 2 seconds, Languish would actually never do damage, because the timer would keep getting reset.

Is that 3rd assumption correct? That assumption is why the number of ticks seems low. If you look at the combat log and filter to "tier ten four piece", you'll see a whole bunch of "dot applied", "dot applied"... then a massive tick. On a hot crit streak, I've seen the simulator put out 5.5k or larger Languish ticks.

#46 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11,540 posts

Posted 12 March 2010 - 03:05 AM

That's all correct.

#47 Valardruid

Valardruid

    Glass Joe

  • Members
  • 13 posts

Posted 12 March 2010 - 07:10 AM

After a little over half an hour with the training dummy in Ironforge, I am certain the ICD is in fact 100 seconds. I used procodile, and the fastest ICD recorded was a little over 100 seconds. I do believe the proc rate might be slightly higher than assumed, with both dots up, and languish ticking at every proc, I only once saw a width of over 104 seconds between procs. (15 trials) Will test more tomorrow.

#48 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11,540 posts

Posted 12 March 2010 - 08:07 AM

The Robot results are interesting, and consistent with the last WC update. It seems the situation looks like this:
IS is best used either once per rotation (end of Lunar). Full refreshing loses a measurable amount of DPS. A good compromise might be to refresh after each Eclipse--that recovers a good bit of debuff uptime at a very small DPS loss. This is such a minor DPS variation that I don't want to overcomplicate the guide with three different options, but I'll try to summarize somehow.

MF: best usage so far "anytime except Solar," it seems. Probably actually "anytime except the beginning of Solar," but not only is that difference negligible, there's something convenient and practical about refresh at the end of Eclipses. Just to think about how this will play out:
You refresh as Solar ends. X seconds later (mean is around X=4), you proc Lunar and extend the MF. So it expires 24 seconds after you cast it, which is 9-X seconds after Lunar ends. But 4-5 seconds after Lunar ends is ether during Solar or during pre-Solar. If it comes up during Solar, you don't refresh until Solar ends, and this has simply reduced to the "refresh once after Solar" scheme that I predicted with WC. If it comes up during pre-Solar, I suspect it's still unideal to refresh, but nobody tested a condition that was so finely-grained. If I had to conjecture the optimal refresh rule, it would actually be, "refresh during pre-Lunar, during all but the last few seconds of Lunar, never during pre-Solar, and only during the last few seconds of Solar." Someone can feel free to tinker with that if you want. At the very least, perhaps take the MF rotation you were using above and see what happens if you stop refreshing during pre-Solar.

#49 Guest_aceofsween_*

Guest_aceofsween_*
  • Guests

Posted 12 March 2010 - 08:27 AM

I agree with you on pretty much all points.

If you have to reapply MF during the Pre-Solar phase, you will end up having to reapply it again after Solar Eclipse, which means it's probably not a good use of the GCD. It also seems like in simulation, Moonfire almost always lasts into the Solar Phase anyway. The only time it doesn't is when you have a poor RNG with the proc on Lunar Eclipse. I did test out an "Only when Lunar Eclipse is Ready" condition, but the results came up more or less the same (or at least within the margin of error).

#50 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11,540 posts

Posted 12 March 2010 - 09:07 AM

I'll probably update the guide accordingly then. One more question for the people testing rotations in Robot--did you construct your rotations to take advantage of the reaction time gained by casting instants during pre-Eclipse phases? i.e. say Solar is ending, so you're about to cast IS. What you really want to do is cast IS immediately after the final Solar Wrath, then check if it procced NG, then cast SF for your next spell if it did, otherwise Wrath, etc. Similarly, if we are in fact going to refresh both DoT's after Solar ends, is best to cast one, then a nuke, then the other (instead of two in a row)? That way you have to two chance for Eclipse to proc during a DoT CGD, preventing you from wasting a Wrath at the beginning of Lunar.

#51 Guest_aceofsween_*

Guest_aceofsween_*
  • Guests

Posted 12 March 2010 - 09:09 AM

I wanted to add that I think we're better off refreshing Moonfire outside of Solar Eclipse though. If you apply it during Solar, you are essentially gambling that you can get off a Starfire before it falls off, which isn't exactly a guarantee. The whole idea behind refreshing it during Solar is to maintain a higher uptime than otherwise possible, but I'm not convinced you can. The extension-gamble aside, if you follow this through, here's how it'll look:

Apply MF with Solar at 6 seconds --> Lunar Begins --> Starfire extends MF x3 --> Reapply MF during Lunar --> Starfire extends MF x3 --> Solar Begins --> MF falls off with Solar at 1 to 3 seconds

Basically, you won't be able to repeat the process because the timing is far too delicate. If you reapply MF before the 6 second mark on Eclipse, you won't be able to extend it at all. If you reapply MF after then 6 second mark (say at 4 or 3 seconds), then your MF will either last into Solar, or fall off when you'll only be casting 1 or 2 more Starfires (meaning you won't be able to extend it). In either case, you are better off going back to reapplying during the Pre-Lunar phase, which is a cycle that does repeat itself.

And to answer your question, I haven't looked into that. The chance of a single Wrath proccing Lunar Eclipse is something like 35% which to me is a little on the small side. If, however, I get two back to back crits Wraths, I know that in game I do start to cast Starfire because the likelihood of one of those getting a proc is pretty high.

Also, I'm not sure I'd advocate reapply IS after Solar Eclipse ends, especially if you're not glyphed for it, but that goes back to one of my earlier questions: what's the goal? It is a minor (but noticeable) DPS loss particularly if you aren't glyphed, but at the same time if you aren't glyphed and want to keep the hit debuff up, then I can understand why you might be curious about these things.

Also, doesn't this depend on the travel time of Wrath? If it's less that 1 second, only using one DoT will probably suffice to see if you pull off an Eclipse proc. If it is greater than 1 second, you would need to apply both to see if it procs.

#52 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11,540 posts

Posted 12 March 2010 - 09:13 AM

Right. Any rotation that depends on Eclipse winds up collapsing into some regular scheme that results in refreshes only as Eclipse ends. This phenomenon has going on ever since the Lunar-only rotations of 3.1, actually. It makes modeling in the spreadsheet quite a bit easier.

#53 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11,540 posts

Posted 12 March 2010 - 09:24 AM

And to answer your question, I haven't looked into that. The chance of a single Wrath proccing Lunar Eclipse is something like 35% which to me is a little on the small side. If, however, I get two back to back crits Wraths, I know that in game I do start to cast Starfire because the likelihood of one of those getting a proc is pretty high.

Also, I'm not sure I'd advocate reapply IS after Solar Eclipse ends, especially if you're not glyphed for it, but that goes back to one of my earlier questions: what's the goal? It is a minor (but noticeable) DPS loss particularly if you aren't glyphed, but at the same time if you aren't glyphed and want to keep the hit debuff up, then I can understand why you might be curious about these things.

Also, doesn't this depend on the travel time of Wrath? If it's less that 1 second, only using one DoT will probably suffice to see if you pull off an Eclipse proc. If it is greater than 1 second, you would need to apply both to see if it procs.


Once you see a single Nature's Grace, the chance of an Eclipse happening is 60%. So take the first case, when there's no DoT refreshing involved--you're casting Wrath in pre-Lunar, you see NG. You've already started one more Wrath after the one that crit, you switch to Starfire after that. That second Wrath will proc Eclipse ~30% of the time. So your decision to swap to Starfire is "right" around 70% of the time.

It's actually messy with DoT's. Now that I think about it, you can't do this with Moonfire, since if you Wrath-Moonfire and see an NG, you don't know whether it came from the Wrath. So only if you're using IS in pre-Lunar (for the debuff) does this come up. On the pre-Solar side, it doesn't really work out well, since Starfire has no travel time. You refresh IS with ~1s left on Lunar, so a crit Starfire there doesn't matter. So while the whole NG-watching idea works generally, it doesn't interact with DoT's as much as I'd hoped.

#54 Guest_aceofsween_*

Guest_aceofsween_*
  • Guests

Posted 12 March 2010 - 09:30 AM

Right, that's essentially the conclusion I came to when I started fooling around with that idea, although I can tell you that if you are using a Mod like MSBT, you can easily pick out when a NG procs from a Moonfire crit and when it procs from Wrath crits.

If only Eclipse were easier to predict it would solve a lot of these uncertainties but such is the life of a Moonkin, I suppose.

#55 Videl

Videl

    Piston Honda

  • Members
  • 111 posts

Posted 12 March 2010 - 09:34 AM

It doesn't really stack with NG clairvoyance, but you can still make sure you make some small use of the time to watch for eclipses.

For proccing solar, you can make sure you don't cast your IS until after your first solar-eligible starfire, rather than the last starfire of the lunar eclipse, that way you atleast get to use it to check to see if that starfire crit since spell queueing means you have to choose your next spell before you see if that one is going to crit.
For proccing lunar, when you see that first NG you'll already be casting another wrath, then you could follow that up with your moonfire rather than casting moonfire as soon as your solar eclipse is over. Using your number, 70% of the time it'd be the same as casting that moonfire at the beginning of your solar, a small dps loss. The other 30% of the time though it'd save you an entire starfire cast that wasn't under eclipse or able to proc eclipse. The tradeoff would be a 70% chance of losing the difference in 1 GCD of eclipsed starfire vs your overall average nuke dps or a 30% chance of gaining the difference in 2 GCD of uneclipsed starfire vs your overall average nuke dps. I'm using "overall average nuke dps" to mean the average dps of all your time nuking both to proc eclipse and during eclipse(it seems like there should be a good standard term for that since it comes up so much when theorycrafting actions during pre-eclipse phases). Using no real math whatsoever, the 70% vs 30% pretty much cancels out with the 1 GCD vs 2 GCD (using a little fudge factor for the low NG uptime for uneclipsed starfires), but overall average nuke dps is 2-3 times closer to eclipse dps than noneclipse dps, so on average you should lose less dps from casting moonfire at the start of your eclipse 70% of the time than you gain from not casting that starfire the other 30% of the time. Additionally, casting moonfire a little later gives some slight benefit from IIS and glyph of starfire should you happen to be unlucky with your eclipses.

I'm assuming that wrath is taking more than 1 second to travel and proc eclipse so you still don't know for sure if eclipse has procced by the time you finish the wrath after the wrath that procced NG. Once you factor in reaction times and lag it is my experience that you don't have to be very far out of melee range for this to be the case.

#56 Guest_aceofsween_*

Guest_aceofsween_*
  • Guests

Posted 12 March 2010 - 09:40 AM

Well, you can test this out pretty easily in Robot by assigning the Refresh values for the DoTs with the following conditions:

NOT <Moonfire/Insect Swarm> on target
AND Previous Action was Starfire

OR Previous Action was Wrath

AND Target has Nature's Grace
AND Solar is ready

OR Lunar is ready


That should do it. I'm not confident it will be a gain though.

#57 Najtrok

Najtrok

    Von Kaiser

  • Members
  • 44 posts

Posted 12 March 2010 - 01:48 PM

Trinket Time again (with updated values):

Setup:
1 Trinket Slot open
Spellpower: 3325
Crit: 981
Haste: 621
Spirit: 280
Int: 1387
4T10 / Lunar Eclipse
Glyphs: S / SF / MF
Rotations: dropped IS
Spec like in my Armory right now (2 iIS / 2 OF / 1 Typhoon)

I checked for DFO, Phyl and Reign in both Hero and non Hero. The Trinket in the other slot always was Reign (H), which I then simulated as wearing it double to get realistic DPS values.

Assumptions:
DFO: 5.5 stacks * 105 (or 121 for Hero) *20/55 (values from Hamlets Post)
Phyl: 1073 (or 1206 for Hero) * 20 / 103 (was a bit more optimistic)

The values for Spellpower I got:
DFO: 210
DFO (H): 242
Phyl: 208,4
Phyl (H): 234,2

If I didnt make any mistake the values should be as following:
Reign (NH): 11521 DPS
Reign (H): 11596,59 DPS
DFO (NH): 11521,91 DPS
DFO (H): 11605,41 DPS
Phyl (NH): 11504,99
Phyl (H): 11582 DPS
Muradins (H): 11489,22 DPS

Which would leave me with: DFO (H) and Reign (H).

Now I am not sure if it is okay to just "double" Reign (H) or if it destroyed the value for it, but I am pretty sure that the value would not have diminished and since I was pretty sure that it would come out best, I took it as 2nd Trinket.

Would like to see some comments if those results are realistic (which I think). Since I provided all values it should be fairly easy to test them if you want.

All calculations were obviously made in WrathCalcs (100310 3.3.3).

And I would like to add a guess (which seems not only to me I think): Phyl will be changed in any way. Its still too weak (commenter above already discussed that)

Edit: Added Muradins (H), thought it really is under all else (pay attention: I am over crit cap without muradins)

#58 Zifrelm

Zifrelm

    Von Kaiser

  • Members
  • 72 posts

Posted 12 March 2010 - 05:01 PM

That's more or less what I got last time I checked trinkets in WC, at least the order from best -> worst. I just left my second trinket slot empty, though I suppose putting either Reign in there wouldn't screw up scaling too much.

I think it's always worth including heroic Spyglass though, at least for comparing to non-heroic 25-man trinkets.

#59 conghaile

conghaile

    Glass Joe

  • Members
  • 8 posts

Posted 12 March 2010 - 06:11 PM

After getting help in the Team Robot thread, I played around with Moonfire refresh timing in the middle of eclipses. I tested until I got a timing that pulled even to the 10019.7 +-25.7dps I got testing aceofsween's ideal No-IS-glyph rotation:

ONLY EXECUTE MOONFIRE IF:

NOT (TARGET HAS [MOONFIRE DOT])
AND 
PLAYER MISSING [ECLIPSE SOLAR] OR LESS THAN 5 SEC REMAINS)
AND
(
  NOT (PLAYER MISSING [ECLIPSE LUNAR] OR LESS THAN 4 SEC REMAINS)
  OR
  NOT (HAVE >= 1 STACKS OF [ECLIPSE LUNAR])
)

It spits out 10012 +- 25.5dps, which isn't any better than any of the other rotations I tested. I had been playing around with the times on both eclipses at the same time using educated guesses, so when I hit this point I decided to focus on the solar eclipse timing only. I figured if there was an ideal cutoff point for casting moonfire during solar, than it must exist between 0 sec (applying when solar eclipse ends) and 5 sec as shown above, because as that time shrunk the probablility of moonfire being fully extended would increase from that point on. As I changed the 5 sec to 4, 3, and 2, the dps only decreased to points outside the confidence interval of the 5 second test. There is no peak, only a valley. When I tested PLAYER MISSING [ECLIPSE SOLAR] OR LESS THAN 1 SEC REMAINS, it jumped back to 10023 +- 25.7dps. This makes sense because that last wrath cast has a cast time range of 0.8-1.2 seconds thus probably won't be under the effects of Eclipse, thus the difference between casting moonfire at less than 1 second and at 0 seconds left on Eclipse is non-existent.

Seeing those results I decided to test times further away from the end of solar eclipse. At 6 seconds, I got 10033.1 +-25.6dps-- an increase, but within the confidence interval. At 7 seconds, 10026.4 +-26.4; At 8 seconds 10011.8 +-25.5dps. All in all, it points to a general range where it's ok to refresh moonfire, but hinging on two rules instead of one. In my opinion, it's just easier to wait until the end, that way you don't have to store the lunar eclipse timing rules in the back of your head as well for what amounts to 0 benefit.

#60 Najtrok

Najtrok

    Von Kaiser

  • Members
  • 44 posts

Posted 12 March 2010 - 06:52 PM

As it was suggested I put in Muradins (H). But as we expected, its slighty under everything else if you are over crit cap.
Hope I didnt make any mistake, while I reentered the values




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users