galzohar

Isn't it that the damage caused by YOUR stats (buffs, spell damage from gear, talents) is decided when you cast, and stuff that depend on the mob (debuffs, resistances) are determined when you hit the mob?

 

TheOnly

No, but as a warlock who has leveled his own character to 70 I've noticed this a billion (ok, 10,000) times. Bolt or soulfire or whatever.... then curse in mid flight. Damage is calculated based on mob debuffs at the time of cast. More precicely, the time of cast on the server.

Basically, "travel time" doesn't exist in determining what the damage is. "Damage is delt" when you cast, but is only subtracted from the target after flight time.

Its not just your buffs versus target debuffs in this timing.
You can see this with a LOT of things. A mob puts up a magic reflect shield. If your spell was already cast and in mid flight it DOES NOT reflect to you.

PvP is a little different, but for PvE that is how it works.


For ISB however, the server doesn't recieve 5 warlocks shadowbolt casts at the same time-- and the first one could remove the debuff before the next 4 casts register. Its the time of cast -- according to the server, not the client.

 

Bolche

Ok, 1.14 is out :

- Added haste rating in the next stat section (thanks to tetracycloide)
- added oom-time
- Added options for LT/DP frequencies (thanks to tetracycloide)

I had to rewrite au lot of things, so there may be some bugs.
Thanks to tetracycloide for providing me a nice way to balance your DP/LT frequency. I just added options :
DP :
0: Disabled
max: the max amount your pet can regen
custom : use the value set in the Custom LT/DP freq table
LT :
0: Disabled
max_regen: the max amount your self regen alows
needed: Frequency req to sustain your mana use
custom : use the value set in the Custom LT/DP freq table

http://www.leulier.com/warlock_dps_v1.14.xls

 

Demi9OD

Nextstat haste rating just doesn't seem right. Highly overvalued from what I can tell. In a 0/21/40 Shadow Destro build (Best case scenario for haste rating), if I take http://www.wowhead.com/?item=32338 and http://www.wowhead.com/?item=32587, comparing the stats on Item Compare show the Blood Cursed coming out way ahead. If I look at Nextstat and multiply by +dmg coefficients the tables are turned and Nimble win out.

Edit: Ok found it, TNS-Haste was adding 1cr as well as 1hr. Sadly haste is still over budget after the nerf and pretty much not worth wearing unless you are hit capped and cant find a comparable crit alternative for the slot.

 

Bolche

Fixed thanks.

 

Bolche

I made some quick tests on imp mana regen, and the following values are what I have found :
with 263 spirit (base) : ~220 mp5
with 313 : ~245 mp5
with 327 : ~257 mp5

What is strange is that the mana tics are not equals. for the 263 spirit, I had the following tics :
219, 220, 209, 222, 220, 220, 220

while 219/220 are common (the way blizzard handles a number between 219 and 220), I don't see where the 222 and the 209 came from.

Anyway, I changed the default values for the pet ooc mp5

 

Kruthal

What you found matches what I found a while back when I tested it. I wrote down some number in a spreadsheet if you're really curious, but it has almost no text, so I doubt you'll get much out of it:
http://folk.uio.no/oysts/Wow/imp-spirit-scaling.xls

Seems just shy of 400 mp5 is the most you can expect out of your imp, fully buffed. He's a pain to buff though, and the gains are fairly minimal (i.e. even unbuffed, SL will net you more hp than you can spend with the extra Life taps you need to do, all buffing your pet does is turn more of SL into overhealing, or give you more self-healing in case of AoE)

 

tetracycloide

I've made some additions to the sheet I am hoping to get feedback on, most noteably including a model for raid ISB uptime.

Usage is fairly straightforward, but the number of warlocks in each spec you expect in your raid and average the crit rates for each spec together and place that number beside it. This should generate a resonable aproximation of raid ISB uptime.

The model assumes all three locks are casting their full rotations with affliciton speced locks using 45% of their time on SB and Demo/Destruction speced locks using 60% of their time on SB. It also assumes the the locks being counted have 5/5 ISB and are using SB exclusivly as filler. If you have a destruciton lock that likes to spam incenerate *bleh* then don't count them. The model automatically accounts for changes to player haste skill based on the stats added to the sheet but assumes the other locks in the raid being counted have no +haste. At some point I had to really draw a line on how many factors for the other locks were going to be included because it's really not useful at all to have to armory your entire raid before you can start using the sheet for yourself.

The Raid ISB uptime is included on all of the TNS numbers so you can use the 'Raid' ISB setting and get valid results for the next crit in the context of what the additional point of crit will do to Raid ISB uptime in addition to improving damage from individual skills. The 'Raid' figure is a dynamic equation which replaces the old slot for 'custom' which was static.

My hope is that this will provide reasonably appropriate numbers for the effect that +crit has on damage with ISB factored in a raid setting. Previously the choice was between single player models that skewed dramatically in favor of crit, over valuing the stat, and a custom setting that made ISB fixed and skewed dramatically against crit, under valuing the stat. The new raid model will yield a value of crit somewhere inbetween relative to the other stats.

I've also added a box next to overall DPS that shows the players effect on raid shadow damage as a result of their additions or detractions from overall ISB uptime. This places low crit builds with a large number of outside locks maintaining ISB uptime for you in the proper raid context.

Sheet with Raid ISB

Just to be clear this sheet is not exclusivly my own work and the vast majority of the work was done by leulier, not myself. I simply added a more complex and dynamic raid ISB model to what was already an excellent spread sheet. Thanks again for all your excellent work leulier.

My vanity is justified.

 

Demi9OD

Interesting work Tetra. I would just like to note that all of our guilds destruction locks are not specced for emberstorm and imp immolate, and do not cast immolate in their cycle, only SB and Curse. This makes the 60% cast time you state in your Demo/Destro catagory (which shouldn't be a catagory, since Demo and Destro casting sequences are often different) quite a bit lower than in practice.

 

tetracycloide

Really? Is there something about demonology I'm missing then? I was under the impression that a high demo build basically used corruption, curse, shadowbolt, and occasionally immolate just like a destruction build. What about the cast sequence is that different? Every time I've used the sheet to model demo DPS corruption and a curse, at least, are always worth casting. Keep in mind the only thing that has an effect on the model is the ratio of SB cast time to total cast time since the differences in crit can just be averaged out. Do demo build really spend significantly more or less than 60% of their time on shadowbolt? significant being 10% or greater difference.

My vanity is justified.

 

Demi9OD

The 1.14 spreadsheet had corruption being a DPS loss, probably from the inflated ISB uptime from casting more shadow bolts. That, coupled with the 30 yard range, prevented me from putting it in my cast sequence. Your new sheet models even a 0/5 corruption as an increase in DPS though for a 0/21/40 build, which changes things. I do believe my reluctance to spec imp immo/emberstorm and leave Immolate out of the rotation is somewhat unique though. I probably still won't start casting corruption with the new ISB calcs because the 8dps increase isn't worth the 6 yard hassle. But yes, to answer your question I spend 88-95% of my time shadow bolting depending on whether or not I have a shadowpriest.

 

◊ Kyth

I use corruption only when it's convenient. I'd never close range just to cast corruption.

I don't see any better place for emberstorm's points, really, so yes I personally do use immolate.


I'm very often running into threat issues, so skipping corruption has that additional benefit as well. Possibly slightly lower dps from not casting it, but 10%-lower-threat damage can be cast in its place. (e.g., we get all channelers down before Mag spawns, so the elemental shaman always ends up dropping tranq air for myself and him partway through the fight even when we're both on cubes. Obviously I don't have threat issues on fights where there's some casting interruption.)

 

• frmorrison

For Locks that want to quickly calculate dps, I found this http://ferox-horde.de/downloads/4/warlockdps.php

Not as good as the one here, but it is web-based.

 

Kargoyle

This spreadsheet is really a cool toy, especially when coupled with the "Advanced Item Search" on allakhazam.

I am just a bit curious about the way +hit and dots are handled :
The spreadsheet considers that only the application of a dot can be resisted.
My understanding was that, actually, each tick of a dot can also be resisted. This would have a noticeable impact on the dps of the dots.

A formula is worth a thousand words, so it's probably best just to indicate how I believe the formulas could be (homeopathically) altered :
A reasonable way to model this would be simply to multiply the current formulas by the chance of hitting the mob for the affliction dots: (1-MAX(1;base_miss-hit-suppr*2)/100)
For Immolate, only the dot part of the spell would have to be multiply by : (1-MAX(1;base_miss-hit)/100)

Am I just being overwhelmingly pessimistic (I knew watching Titanic again wasn't such a good idea) ?

 

tetracycloide

This is true although your hit rating won't have anything to do with these partial resists. As far as I know everything on the sheet can be partially resisted except CoD (which I am only assuming cannot be because I've never seen it) so while the total DPS number is wrong for a boss level mob the relative values of all the skills are still correct. The error in the total DPS would be a fixed percentage so if you were comparing two different DPS figures from changes in gear or spec then the bigger DPS is still the bigger DPS.

My vanity is justified.

 

Ele'

CoD can indeed be resisted, I've seen a lot of 25-50% resists from Quagmirran while soloing it. (When you suck at jumping the little cliff and take 3 hours to kill the beast, you use a *lot* of CoD, let me tell you :p)

 

Silmeria

Partially resisted? I've never seen a partially resisted CoD in my entire life. You're thinking of missed CoDs probably.

 

Bolche

CoD can both miss (= be fully resisted when applied) and be partialy resisted when it ticks. I have seen quite often numbers like "7500 {2500 resisted}" on my SCT.
No partial resist is modelled in the spreadsheet, as they affect all spell by the same value (with an overall 8% reduction on a mob with 0 shadow/fire resist, I believe).
What is missing in the spreadsheet is the damage you loose (or win ) when a CoD misses. Currently its assumes that you recast it 1.5s later, wich is not possible due to 1 min CD. I should probably model a CoD miss as a cast of 3 CoA

 

Jinda

Silmeria, I will also support Bolche in saying that CoD can be partially resisted, as I saw this happen using SCT in my last Magtheridon fight. Around 2500 damage was partially resisted when one of my Curses of Doom ticked.

 

Deelit70

I have made a spreadsheet as well which I am modelling using a time index (You make your choices of what spells you prefer, and then it models out in half second intervals over a t=0 -> t=600 line. At each half second it decides if you're casting, recovering, or free to cast again.

I've been modelling my dots under the same theory as DDs - you cast and it does an average amount of damage per cast. I do this because it is impracticable to have less than half-second fractions for my sheet (which is what you would get if you modeled a 1.5s cast with a 83% hit rate). I am wondering if you have any way to solve this problem without refining t.

I also thought I would post because you seem to just pick a number for SB uptime, and my sheet models it (granted, there are so many variables and so little control in a raid setting it is not much better than actually guessing, but it's a start) which you might want to look at. Also your Lifetap is confusing me. Neither of the included models are optimal. It should be modeled to only kick in once your mana drops below some number, say 1000. Otherwise you lose a good 10-20% of your dps during the first minute or two (In my simulations I have found that after minute 3 warlocks settle in at ~75% of non-tap dps). Putting your stats into my sheet I consistently get slightly above your numbers, but similar.

Warlock_Spreadsheet_V4.2a.xls - FileFront.com

 

Anastotte

I just downloaded this today and i have a circular reference error. Have others encountered this? Is there a way to resovle this?

 

Trickykid

I've seen one with the LT/DP cycles. If that's where you saw the circular ref, choose "Custom" and manually enter numbers (I find a LT cycle that puts my OOM at 10 minutes).

 

tetracycloide

Alternativly you could use the intended circular reference to your advantage by adjusting your excell settings to iterate formulas for you. If you have taken any calculus this is basically just letting excell calculate the limit of seconds per dark pact as mana consumption approches 0.

You can change the setting in excel in windows by going to the tools menu (alt + t shortcut) and selecting options from the drop down display. In the window that opens select the 'Calculation' tab and put a check in the box next to 'Iteration.' The default values for 'Maximum iterations' and 'Maximum change' are 100 and .001 on my machine and they both work fine. Also make sure the radio button next to 'Automatic' is selcted instead of 'Manual.'

The reason excel has iterative calculation dissabled by default is becuase they are literally never ending equations from an algebraic stand point and the sheets are incapable of properly evaluating limits through anything other than brute force. Limits can also be a little wonky sometimes and produce the wrong answer if evaluated with brute force algebra. The equations for infinite mana cycles in the sheet have been carefully constructed to avoiding any issues.

If you start getting errors in the DPS calcs or in the 'TNS' section of the sheet you can 'reset' the calculations by setting all of your casts to '0' and then adding them back in.

My vanity is justified.

 

Glaurung

Any possibility of expanding the demo, and more specifically felguard, support of this spreadsheet? As near as I can tell, here are few improvements that could be made:

1) Lack of Demonic Knowledge support. - I know you can just manually punch in your total at the end, but it's nice for gear testing to be able quantify how much this talent is worth.

2) Demonic Knowledge Syngergies with gear and raid buffs. - Sort of derivative of the last problem, but raid buffed vs. unbuffed is a significant difference for demo locks. I've calculated that BoK, fort, MotW. and AI on yourself and your pet gives you roughly +65 dmg, and also +65 pet AP (65*.57+BoK's 1.1 Str to felguard).

3) No consideration of melee buffs for pets. - This is a huge consideration, as once again, even without massive raid stacking for your benefit (ie enhancement shamans, dps warriors, marksmanship hunters all in your group), you still can gain from the typical raid buffs like BoM. BoM alone gives you a 50 dps edge (40 with boss level glances and miss/dodge rate) that no other warlock spec can effectively tap into. Possibilities for battleshout, SoE, unleashed fury, and trueshot aura allow for a huge gap in calculations and performance.

4) No considerations for melee debuffs on mobs - Bloodfrenzy, improved hunters mark, sunder armor (or expose armor), faerie fire, curse of recklessness, and hunter's expose weakness. The expose weakness potion would probably a major pain, but the others are reasonable possibilities.

I have some calculation work done on these. If you're interested, let me know and I'll post them.

 

Silverstorm

Deelit, I've looked at your spreadsheet (though back at 4.0 or 4.1), and it's quite nice. I'm working (slowly) on some personal improvements to it, such as variable time increments based on spell duration and latency options. Another thing I've done is add the ability to break out DoT damage into individual ticks to have a smoother representation of the damage done. I stuck with the .5 second ticks for determining when those happened for now, though I hope to be able to have resolution down to the tenth or hundredth of a second. Basically I want to look at when the next event (health tick, mana tick, damage dealt, DoT tick, etc.) happens and add that much time before calculating the next row.

I'd also like to get a gear selector sheet written that includes a good list of gear. I had one at level 60, but I don't know the source.

I have not looked at Leulier's sheet yet, but I will do so, and that might help with my sheet as well. Once it's decently usable, I'll post a link.