Jump to content


Photo

WrathCalcs - Moonkin DPS Spreadsheet


  • Please log in to reply
93 replies to this topic

#21 Erdluf

Erdluf

    Great Tiger

  • Members
  • 960 posts

Posted 06 December 2008 - 03:59 AM

Yay!

You undervalue Int a bit. Scaling from Int should be something more than (12% of scaling from Spellpower + 25% of scaling from Crit).

At least part of the problem is in the TotalSp formula:

TotalSp = ... + 0.04*Lun_Gui*(MkFInt-Int) + ...

When Int increases by 1, MkFInt shoul increase by 1.2 or so, but (MkFInt-Int) increases by only about 0.2.

As a workaround that gives what I think are reasonable results, I did:

In "Character Sheet":
E52: "ExtraInt"
E53: 0 <--- Named E53 "ExtraInt"
B53: Changed to "=ExtraInt+1"
Table at B52:C53, now has a column-input of ExtraInt

In "Basic Calcs", the "Total Int" formula now begins: "=(Int+ExtraInt+..."

With this change, Int DPS scaling goes from about 0.28 to about 0.41 (no other changes to your downloaded spreadsheet, unless I messed up).

#22 Adoriele

Adoriele

    Happy October 19th!

  • Members
  • 10,089 posts

Posted 06 December 2008 - 06:56 AM

Good call on that work around. A lot better than forcing people to input their pre-LG spellpower. I've got it changed locally, I'll upload it either tomorrow or Monday.

[edit] Doesn't fix the scaling numbers for LG itself, though. Right now it only shows the benefit from Furor.

#23 Erdluf

Erdluf

    Great Tiger

  • Members
  • 960 posts

Posted 07 December 2008 - 08:03 PM

[edit] Doesn't fix the scaling numbers for LG itself, though. Right now it only shows the benefit from Furor.


Instead of the data table stuff, you can just use

Lunar Guidance Benefit = Spellpower Scaling * 4% * Lunar Guidance Points * Raid Buffed Int

#24 Adoriele

Adoriele

    Happy October 19th!

  • Members
  • 10,089 posts

Posted 07 December 2008 - 10:37 PM

Instead of the data table stuff, you can just use

Lunar Guidance Benefit = Spellpower Scaling * 4% * Lunar Guidance Points * Raid Buffed Int


That might just work. I was gonna say that Spellpower's value diminishes as you add, but it doesn't. It's just that everything else's value rises.

#25 Namsar

Namsar

    Glass Joe

  • Members
  • 13 posts

Posted 07 December 2008 - 11:50 PM

Edited: out all my junk, need to drink more coffee before posting.

#26 crmccar

crmccar

    Glass Joe

  • Members
  • 18 posts

Posted 12 December 2008 - 09:56 PM

Hey Adoriele,

I'm a little confused about the different rotation types provided. All the rotations labeled with "filler" I assume define what you do while Eclipse is on cooldown. However, it doesn't seem to say anything about what's happening when eclipse is off cooldown or procced. Could you clarify what's going on here? I tried but failed to figure it out from the calc pages. Thanks!

#27 Adoriele

Adoriele

    Happy October 19th!

  • Members
  • 10,089 posts

Posted 12 December 2008 - 10:42 PM

Hey Adoriele,

I'm a little confused about the different rotation types provided. All the rotations labeled with "filler" I assume define what you do while Eclipse is on cooldown. However, it doesn't seem to say anything about what's happening when eclipse is off cooldown or procced. Could you clarify what's going on here? I tried but failed to figure it out from the calc pages. Thanks!


This is what the Eclipse Rotation Type selector is for, though I know it's a bit hard to understand what I'm getting at with that, so I've updated how it displays the rotation: When choosing an Eclipse Rotation Type, the form is pre-Eclipse, Eclipse, post-Eclipse. So choosing "W-SF-W" means that you'll cast Wrath to proc Eclipse, Starfire during Eclipse, and Wrath again during the cooldown. Note that this field DOES NOT UPDATE when you switch your filler type, and the post-Eclipse selection should always match your filler, so if you switch from, say Wrath Spam to Starfire Spam, you will need to re-select your Eclipse Rotation Type, or it will assume you're casting SF-W-SF (W-SF-W if you're using Wrath filler).

v1.1.2 also include a couple scaling fixes, Int and Lunar Guidance being the big ones. It'll be up on the OP in a couple minutes.

#28 crmccar

crmccar

    Glass Joe

  • Members
  • 18 posts

Posted 12 December 2008 - 11:00 PM

This is what the Eclipse Rotation Type selector is for, though I know it's a bit hard to understand what I'm getting at with that, so I've updated how it displays the rotation: When choosing an Eclipse Rotation Type, the form is pre-Eclipse, Eclipse, post-Eclipse. So choosing "W-SF-W" means that you'll cast Wrath to proc Eclipse, Starfire during Eclipse, and Wrath again during the cooldown. Note that this field DOES NOT UPDATE when you switch your filler type, and the post-Eclipse selection should always match your filler, so if you switch from, say Wrath Spam to Starfire Spam, you will need to re-select your Eclipse Rotation Type, or it will assume you're casting SF-W-SF (W-SF-W if you're using Wrath filler).

v1.1.2 also include a couple scaling fixes, Int and Lunar Guidance being the big ones. It'll be up on the OP in a couple minutes.


Ok, thanks for answering that. What still confuses me though is that none of the numbers change for the "filler" rotations when i change my eclipse rotation type - only the "spam" rotations do. If I'm understanding your explanation correctly, the "filler" numbers should change too.

#29 Adoriele

Adoriele

    Happy October 19th!

  • Members
  • 10,089 posts

Posted 12 December 2008 - 11:03 PM

Ok, thanks for answering that. What still confuses me though is that none of the numbers change for the "filler" rotations when i change my eclipse rotation type - only the "spam" rotations do. If I'm understanding your explanation correctly, the "filler" numbers should change too.


Eclipse is not fully-implemented for those rotations yet. It's been mentioned a few times, including the OP.

#30 thedopefishlives

thedopefishlives

    Don Flamenco

  • Members
  • 482 posts

Posted 13 December 2008 - 05:24 PM

Eclipse is not fully-implemented for those rotations yet. It's been mentioned a few times, including the OP.


If you want some numbers for the "filler" rotations and Eclipse, you can try Rawr. I cannot guarantee that the math is 100% accurate, but it should at least give you a ballpark figure for how much benefit Eclipse will give you.

#31 Adoriele

Adoriele

    Happy October 19th!

  • Members
  • 10,089 posts

Posted 22 December 2008 - 08:56 PM

If you want some numbers for the "filler" rotations and Eclipse, you can try Rawr. I cannot guarantee that the math is 100% accurate, but it should at least give you a ballpark figure for how much benefit Eclipse will give you.


The numbers are generated on the fly, so I'd have to model the rotations completely before the numbers meant anything. That said:

V1.1.3 is up! V1.1.3 now includes (hopefully) strong support for Eclipse when choosing the IS, MF, Filler rotations. Note that this includes the previous caveats for choosing which Eclipse Rotation Type, in that it assumes you're using the "dumb" rotation unless you specifically choose the "smart" rotation each time you change your filler type. That said, some interesting observations:

The third point in Eclipse is showing strong value even with a high native crit rate when using a smart Starfire rotation (which is the highest DPS rotation available) We're talking on the order of 100 DPS, like the other two points.

The smart Wrath rotation has lower DPS than the dumb Wrath rotation, likely because of Starfire's higher DPET and larger benefit from Eclipse. This means that you really do want to make the switch to a SF filler rotation as quickly as possible.

The Starfire Idol seems to have higher value than the Wrath idol on all but the smart Wrath rotation, likely because of the former observation.

The model makes some assumptions, namely that you will refresh your DoTs the moment they drop. This is NOT OPTIMAL. However, trying to get it coded optimally is going to take a huge amount of effort, which will take time. I wanted to get at least a modicum of support for Eclipse in ASAP. To work out casting DoTs, the model takes the total time spent casting filler, divides it by the DoT durations, then allocates the appropriate partial GCDs to each part of the rotation based on time spent. The duration for Moonfire is assumed to always be full-length on the Starfire filler rotation, but only one Moonfire per rotation is extended in the Wrath rotation (I assume you will get in 3 Starfires during Eclipse in the smart rotation, extension is based on number of pre-Eclipse SFs cast in the dumb rotation).

Now that this set of Eclipse rotations is in, the other DoT-based rotations should come pretty quickly; I just have to copy the pages and remove stuff, rather than make new models. Hopefully I'll be able to get them in during my days off for Christmas, but no promises.

#32 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11,540 posts

Posted 23 December 2008 - 09:00 AM

SpamEclipseRotCalcs.E6 reads:
(1-Eclipse/3*IF(SmartEclipse,0.6,1))*IF(SmartEclipse,WCritChance,SFCritChance)*TotalHit

I believe it should read:
(1-Eclipse/3*IF(SmartEclipse,0.6,1))*IF(SmartEclipse,WCritChance,SFCritChance)*TotalHit/(1-D6)

Explanation:
Say you have chance H on a cast to hit, chance C on a hit to crit, and chance E on a crit to eclipse.

Chance to eclipse is HCE. (what you have in D6)
Chance to crit but not eclipse is HC(1-E). (what you have in E6)

But the probability of a crit, provided a spell has not procced eclipse (what you're doing in E6-E7) is not HC(1-E), but HC(1-E)/(1-HCE).

-------------

WNGCast seems to be equal to:
=MAX((WrathMidCast-0.5)/(1+TotalHaste),1)+Penalty*Latency

What's "Penalty"? It seems to be set to 2. This makes my Wrath NG cast time longer than my normal Wrath cast time, which is a bit silly.

-------------

E21 references E6, but I think it should reference E20.

#33 thedopefishlives

thedopefishlives

    Don Flamenco

  • Members
  • 482 posts

Posted 23 December 2008 - 12:57 PM

WNGCast seems to be equal to:
=MAX((WrathMidCast-0.5)/(1+TotalHaste),1)+Penalty*Latency

What's "Penalty"? It seems to be set to 2. This makes my Wrath NG cast time longer than my normal Wrath cast time, which is a bit silly.


Penalty is a multiplier because the spell queuing system doesn't work for spells that clip the GCD. That's my understanding of it, anyway.

#34 Adoriele

Adoriele

    Happy October 19th!

  • Members
  • 10,089 posts

Posted 23 December 2008 - 04:01 PM

Penalty is a multiplier because the spell queuing system doesn't work for spells that clip the GCD. That's my understanding of it, anyway.


This. A penalty of 1 is (unless I've made an error somewhere), actually not a penalty, but normal, as all spells are subject to latency (though you can make it very small with AHK if you're not worried about Blizzard finding out). A Penalty of two suggests that, since you can't queue a spell while under GCD, you actually have to suffer through the full roundtrip latency effect, which is double your ping. It's a quick-fix, as I'm not sure that 2 is the correct answer, but I'm very sure that 1 is the wrong answer. I made it into a variable, though, so it would be easier to change. You're right that, at high latency, it causes Nature's Grace to actually be a DPS loss.

Actually, now that I think about it, Wrath's cast time should always be the GCD or less, which means it should never be queueable. I'm going to look into this, because that doesn't sound right, as I'm sure I've been queueing it.

As for the calcs in E6, I'm on the fence about them. I wanna run some sanity checks before I upload it, but your theory's definitely sound.

E21, you're probably right. My SOP for generating the Wrath numbers is to copy the whole Starfire section and swap stuff around. Looks like I missed a spot.

#35 hquest

hquest

    Glass Joe

  • Members
  • 19 posts

Posted 14 January 2009 - 03:18 PM

Any chance of getting Owlkin Frenzy added to calculations? Or the spreadsheet works most for Patchwerk -like encounters?

#36 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11,540 posts

Posted 14 January 2009 - 03:28 PM

Any chance of getting Owlkin Frenzy added to calculations? Or the spreadsheet works most for Patchwerk -like encounters?


How would you suggest this be modeled?

#37 Ashaera

Ashaera

    Piston Honda

  • Members
  • 141 posts

Posted 14 January 2009 - 03:39 PM

Modeling Owlkin Frenzy in pretty much hopeless (Or will be very situational) - For common moonkin use I dare claim that theres no good reason to pick Owlkin Frenzy over Brambles (Malygos 5min being the main exception).

Perhaps if they start adding more raid damage later on, but for now its not a very useful talent. Both are fairly low dps boosts, but Brambles is if nothing else more reliable atleast.

---
If you want to model it then figure out what averaged % you wanna examine (Check your wws's for number of hits in the fight you wanna look at, which of the hits can proc it & match with proc chance, uptime etc.)

Look at Moonfury to get an idea of what 10% gives & devide that number by whatever % you calculated to be a realistic Owlkin Frenzy averaged gain.

---
http://elitistjerks....623-post54.html - Is my quick approximation over its value in Naxx fights (Was done after our first clear so fight lenghts are way off).

#38 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11,540 posts

Posted 14 January 2009 - 04:00 PM

Modeling Owlkin Frenzy in pretty much hopeless (Or will be very situational) - For common moonkin use I dare claim that theres no good reason to pick Owlkin Frenzy over Brambles (Malygos 5min being the main exception).


I'd say Sarth3, more so. This does mean it can be reasonable to spec OF, as for many people (myself included) this is the only challenging thing we're working on each week.

Also, let's generously say Brambles adds 1% average DPS for 3 points. For OF to give that much, it would need to have 10% uptime, which means it only has to proc every 100 seconds. Fights where you take damage every 15 seconds on average don't seem too outlandish. There aren't too many in Naxx because it's so easy, but it's entirely possible that hard fights in later zones will have that much raid damage.

For contrast, imagine how nice a talent OF would be if the current content included Kil'Jaeden, M'uru, Twins, Kalecgos, and Illidan.

#39 Ashaera

Ashaera

    Piston Honda

  • Members
  • 141 posts

Posted 14 January 2009 - 04:52 PM

I went over my wws from the last 4½-5 naxx clears to check Frenzy value again.

Ill stick to the fights where Im pretty sure what procs it (Leaving sapphiron out fx as I got no idea if the aura triggers or not).

Anub'rekhan
10 hits // 532 sec
1.5proc = 15 sec
Uptime = 2.8%
Averaged %-gain = 0.28%

Faerlina
18 hits // 435 sec
2.7procs = 27sec
Uptime = 6.2%
Averaged %-gain = 0.62%

Maexnna
13 hits // 453 sec
Averaged %-gain = 0.43%

Patchwerk
0 hits

Grobbulus
48 hits // 1353 sec
Averaged %-gain = 0.53%

Gluth
18 hits // 826 sec
Averaged %-gain = 0.33%

Thaddius
48 hits // 1241 sec
Averaged %-gain = 0.58%

Razuvious
59 hits // 959 sec
Averaged %-gain = 0.92%

4h
61 hits // 1335 sec
Averaged %-gain = 0.69%

Noth
25 hits // 1025 sec
Averaged %-gain = 0.37%

Heigan
10 hits // 945 sec
Averaged %-gain = 0.16%

Kel'thuzad
53 hits // 1537 sec
Averaged %-gain = 0.52%

Malygos
A heavy night of 5min attempt wiping with Owlkin Frenzy spec shows me about 0.8procs pr attempt - I guess we called it a wipe around 3mins averaged - So roughly a 0.5% gain there.

Sartharion+3 (Low sample size)
8 hits in 441seconds.
1.2procs = 12sec.
Uptime = 2.7%
% gain = 0.27% -> Sample size might be unfair dunno.

On the same Sartharion my trees list in at 4.33%. If I would have had Brambles (old parse so I dont) that would have been 4.95% / or a total gain of 0.65% damage done over the full fight.


Edit : Sorry for hijacking the thread!

#40 Erdluf

Erdluf

    Great Tiger

  • Members
  • 960 posts

Posted 09 February 2009 - 04:23 AM

I ran some target dummy tests, hoping to determine if Wrath Eclipse is additive with Moonfury or is a global multiplier. I found a few other things in the process.


Tooltips for SF and Wrath seem to show that Moonfury provides a bit more than 10% benefit to base damage. This was true in BC also.

Untalented SF (according to MMO) is 1028 to 1212. With Moonfury, my tooltip says 1141 to 1345, about an 11% increase.
Untalented Wr (according to MMO) is 553 to 623. With Moonfury and LK Idol (+70) my tooltip says 689 to 767, about a 10.7% increase.



Bug in Wrathcalcs 1.1.3:

WrathCalcs Moonfury is 3,6,10% for SF. 2,4,6% for Moonfire. 2,4,7% for Wrath.



WrathCalcs is using 76%/6 = 12.667% coefficient for each IS dot. I'm finding that 12.70% gives better answers:

IS formula (215 + spellpower*coef) * 1.3(glyph) * 1.13 (e&m) * 1.03 (e&m) *1.04 (mss)

1746 SP, no MSS: Ticks are 660 or 661. 12.7% coef says 660.8 vs. 659.9

1841 SP, with MSS: Ticks are 706 or 707. 12.7% coef says 706.2 vs. 705.3

2431 SP, with MSS: Ticks are 824 (only value observed): 12.7% coef says 824.1 vs. 822.9.


Test values for Wrath Eclipse. None of these had IIS. All had LK Wrath idol, Moonfury, personal E&M and target E&M. No Owlkin Frenzy. There were only a few casts at each spell level.

SP 1746, no MSS, damage range 2769-2802.
SP 2336, no MSS, damage range 3359-3418
SP 1841, MSS, damage range 2953-3040.

By my calculations, the bottom end of those damage ranges are too low for Moonfury and Eclipse to be multiplicative. They are in-range for Moonfury and Eclipse being additive. I used Moonfury=10% for these calculations, but worked off of the tooltip base-damage numbers.




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users