01/14/10, 9:45 AM #16 Warlocomotif Don Flamenco     Warlocomotif Human Warlock   Emerald Dream (EU) There's a big difference between playing perfectly and assuming uptimes/scenarios that even a computer player could never match. Frankly assuming such things makes the end result irrelevant in my eyes, as even perfect play would not match it. Anyway you should be able to calculate the overlap probability with the formula above. The mind is like a parachute. It works best when it's open.
01/14/10, 12:57 PM   #17
purefury
Glass Joe

Human Warlock

Azjol-Nerub (EU)
 Originally Posted by Warlocomotif There's a big difference between playing perfectly and assuming uptimes/scenarios that even a computer player could never match. Frankly assuming such things makes the end result irrelevant in my eyes, as even perfect play would not match it. Anyway you should be able to calculate the overlap probability with the formula above.
And i did ;p. thats tha 18% uptime.

What i had in mind by perfect play, is no human+internet lag. In practice, we must consider that immolate is more important than incinerate, but its alot of times when we use 2proccs, and got 1.2second left on immolate... either clip 1dot, cast incinerate without immolate, or cast shadowbolt before incinerate.
I'm sure simulation can answer the question easily, and give results, but its not a scenario you can model into sensible eqations.

To sum it up, either use simplified scenario equation, roughly 18% of time spent on incinerates, or use simulation. DPS isn't a single formula anyway. Potential DPS will be for a caster always higher than real one, can't expect perfect LT usage, perfect decimation awarness, perfect DoT uptime and then say that human can't possibly do perfect incinerates ;p

My suggestion would be to rip the numbers of casts from a huge simcraft sample, and consider them constants. Gear wont affect them, only the damage dealt by Incinerate/Shadow Bolt.

 01/20/10, 2:06 PM #19 Cerula Von Kaiser   Mahida Tauren Warrior   Mal'Ganis Can you merge this thread with the Warlock Spreadsheet 3.3.0 now that it's started?
 01/20/10, 2:18 PM #20 Aramezzet Von Kaiser   Täj Gnome Warlock   Andorhal Haste is entirely factored out for this calculation, and latency / remaining cast time aren't entirely factored in, which screws with equation 2. I think it's a good start, however. 1. Average number of MC procs per cast of Corruption: sum{from n=0, to n=6} of {0.88^6-n*n *0.12^n} e: I tried putting this term together with Latex but couldn't figure it out. I'll clean it up later and hope it makes sense. I think this figures to 0.085 procs per cast. 2. Average number of casts benefitting from each Core proc, mostly discounting latency and remaining cast time: 0.88^2*3+0.88*0.12*1+0.12*0=2.429 3. DPS benefit per use = DPS benefit * duration Numbers pulled from Simcraft SampleOutputT9T10 35%+: (incinerate_dps-shadowbolt_dps) * incinerate_cast_time = (7898-5796) * 1.923 =4038D 35%-: (MC_Soul_Fire_dps - Soul_Fire_dps) * soul_fire_cast_time = (5435) * 5435.04D (I don't think the Simcraft sample does casts Corruption when the target is below 35% HP, which would skew Soul Fire's damage upward a little bit. I'm not sure.) Damage benefit over a Corruption from Molten Core: procs_per_cast * benefits_per_proc * damage_per_benefit 35%+: 832.8D 35%-: 1120.85D Last edited by Aramezzet : 01/20/10 at 5:13 PM. Reason: Edited for a typo in the formula.
 01/20/10, 2:45 PM #21 Mupki Glass Joe   Fearsom Orc Warlock   Mal'Ganis There are still things I need to fix that are high on my priority list. Nibelung procs, Reign of the Dead, and some other trinket procs. I also need to include Black Magic for those that want to see the differences. TNS also doesn't scale pets yet. If you have any formulas that would help me with these things, it would be much appreciated.
 01/20/10, 3:36 PM #22 • Hamlet     Hamlet Tauren Druid   Mal'Ganis No need to split out the binomial in step 1. Mean procs per Corruption cast is 6*0.12 = 0.72. (Also, your sum is wrong--you need to include the binomial coefficent 6Cn. And multiply by the number of procs n in each case. The correct sum is: $\sum_{n=0}^6{n\binom{6}{n}0.12^n*0.88^{6-n}}$, which equals 0.72)
 01/20/10, 5:07 PM #23 • Hamlet     Hamlet Tauren Druid   Mal'Ganis Anyway, on to the problem from the OP. Best I can tell (I don't really know anything about Warlocks), Incinerate cast time before haste is 2.5/1.3 = 1.92s. Corruption tick time is 3s before haste. Both decrease equally with haste, so that ratio shouldn't change. Simplified outlook to start: all cast time is spent on either Shadow Bolt or Incinerate. At any given moment, Molten Core will be up if it has procced within the amount of time it takes to cast 3 Incinerates. Given a bit of reaction time, etc., let's say that there always 2 Corruption ticks in the time it takes to chaincast 3 Incinerates. In other words, if at time T, one of your last two Corruption ticks has procced MC, you'll still be casting Incinerate. Otherwise, you'll be casting SB (since two Corruption ticks is the time it takes to use up your 3 MC charges). One of your two Corruption ticks will have procced MC with probability (1-0.88^2)= 22.6%. So a good first-order estimate is that you'll be casting Incinerate 22.6% of the time, and Shadow Bolt the rest of the time. Curiously, this is constant with respect to haste, which may be counterintuitive at first but makes sense when you think about it. Note that 22.6 is very close to Mahida's above empirical estimate of 27%*82%. ------- The big wrench in the works is all the execution time spent on things other then Incinerate or SB (including Corruption itself). I think the best way to deal with this is to average in the cast time spent on anything other than nukes in the form of added Corruption ticks (since, from the perspective of the Immolate/SB question, the time spent casting other stuff only matters insofar as it provides extra time during which Corruption may have ticked). Essentially if you spend a fraction X of your time casting things other than SB/Incinerate, perform the same computation I did above, but imagine that Corruption ticks 1/(1-X) times more quickly. So, above, I assumed that the ratio of Incinerate cast time to Corruption tick time (call this R) is 2/3. Here let's start with the real value of R=C/3 (where C is the pre-haste cast time of a buffed Incinerate, 25/13). And, using the approximation from the last paragraph, we'll instead model the ratio as R=C/(3*(1-X)). Now basically repeat what I did in the top half of the post. Since Corruption will tick R times during one Incinerate cast, it will tick 3R times during 3 Incinerate casts. So, at any time, you'll find yourself casting Incinerate instead of SB if any of the last 3R Corruption ticks have procced MC. This leads to a time spent casting Incinerate (as a fraction of total time spent casting nukes) of: $1-0.88^{3R}$, or $1-0.88^{\frac{C}{(1-X)}}$, where C = 25/13 (pre-haste cast time of Incinerate). Last edited by Hamlet : 01/20/10 at 5:13 PM.
 01/20/10, 7:07 PM #24 Vidiz Glass Joe   Vidiz Human Warlock   Xavius (EU) Hey there - first of all may I say very well done and I comment you for your hard work and effort! However, when I try loading my character from the armory into the spreadsheet it says: " 48667 is not a supported Back please use Custom Gear for it " Why is this error message being displayed (maybe I just don't understand what it's asking me to do) and how do I fix it? Many thanks Vidiz
 01/20/10, 7:17 PM #25 Mupki Glass Joe   Fearsom Orc Warlock   Mal'Ganis It means that the item isn't on the spreadsheet, it loads items based on their ID. I'm still working on adding every item I can for the next update, but it takes awhile. There are a lot of alliance items that aren't in there from ToC and probably some others I have missed.
01/20/10, 7:30 PM   #26
Hamlet
<Druid Trainer>

Tauren Druid

Mal'Ganis
 Originally Posted by Mupki Nibelung procs
Base damage range for normal Nibelung is 1591-1785. It's affected by level-based partial resists, and by caster debuffs on the target (Earth and Moon, Imp. Scorch, Totem of Wrath). 50% crit bonus, 0% base crit rate, 0% miss. Casts on average 17 Smites per proc if it stays still. 1% proc rate, but a bunch of weird talents and things proc it; you'll have to experiment some yourself.

 01/21/10, 2:04 AM #27 Acabar Glass Joe   Gnomercee Gnome Warlock   Fizzcrank I get the same error for a couple of my armor pieces. My main concern is I'm getting a ton of #N/A errors in cells where there's supposed to be DPS/Damage numbers. Anyone else having this issue? Here's a picture of the spreadsheet after I import my armory to avoid confusion:
 01/21/10, 2:51 AM #28 Cerula Von Kaiser   Mahida Tauren Warrior   Mal'Ganis You're getting those errors because you have 2 items that aren't in the spreadsheet's database. Also ran into a bunch of bugs with loading a character...Ill try and fix all this by tomorrow.
 01/21/10, 3:15 AM #29 Vidiz Glass Joe   Vidiz Human Warlock   Xavius (EU) I understand how it could be hard to add every single item. For me, the item in question was the Shawl of the Devour Crusader - the 258 cloak from ToC 10 HC Insanity.
01/21/10, 11:55 AM   #30
Acabar
Glass Joe

Gnome Warlock

Fizzcrank
 Originally Posted by Cerula You're getting those errors because you have 2 items that aren't in the spreadsheet's database. Also ran into a bunch of bugs with loading a character...Ill try and fix all this by tomorrow.
That's what I figured - just wanted to report in case it was something else.

The two items for me were: Bindings of Dark Essence - wrists from ToC 25 and the Icehowl Cinch from ToGC 10.