The next stat numbers are wonky in 1.13. I put up my stats for deep affliction, and noticed that it was listing 50:50 for +shadow vs +fire. That can't be right, as ~10% of my dps is immolate.
So, I turned off immo and it went to 51:49 s/f. The weighting formulas are wrong.
On the TNS-X pages the Life Tap and Dark Pact frequency fields have become unlinked to the fields where you enter them on the DPS sheet. For those of you that do not want to wait for the next release but don't already know how to fix this problem do the following:
Open file
Select Format menu
Select Sheet then goto Unhide
Select a sheet to unhide and click ok
Repeat until all sheets with 'TNS' in the name are no longer hidden
Goto each TNS page and enter the following values into the following cells:
Cell D25: =DPS!D25
Cell D26: =DPS!D26
After you have updated both cells on all 4 pages 'the next stat' values on the DPS page should be correct.
Fixed the "The Next Stat" section. You can either download the 1.13.1 or if you don't want to re-configure your sheet again, you can do exactly what tetracycloide said.
I also fixed the comment on the "total crit %" cell according to Bakka's correct remark.
Thanks tetracycloide for you intersting comment. Information about sustainaible dps is indeed very interesting. As I said a few pages before, my primary goal was to make a good dps sheet, then add the pet dps, then start working on the "mana management sheet".
The relevant informations a WL would like to know are, I beleave :
- Unlimited mana dps and how long can you keep it (ie full->0 mana)
- Self-sutained dps using DP (if available) and LT only what you can heal yourself (Siphon Life, Soul Leach, maybe 3/3 FSW)
- Sustained dps with unlimited heal
- What is your max dps in a X min fight
About self-sutained dps, I didn't know Excel could do such nice things
But I beleave for a affliction lock, the logic is :
DP as much as your pet can provide you the mana, else LT.
I have a very very strong feeling that using my equations will result in a cycle that is only infinitly sustainable on the sheet because it assumes the pet always has mana. The version of the sheet I have locally has both of the equations as options and also builds them into the 'TNS' sheets so that LT/DP cycle adjustments get factored into TNS calculations. I will say that the limits are quite useful in theory but in practice, espeically when you start carrying them over to the TNS pages, they become difficult to troubleshoot. There's a reason the default setting is no cyclical equations and it's mostly because limits quite often go to infinity or to numbers that don't make any sense for the calculations you are performing.
I also added in a TNS-haste sheet on my version, it's kicking back some interesting results on the value of haste. So far it appears to be roughly .5 dmg or so depending on spec but unlike crit, which varies wildly from spec to spec, haste appears to be much more resiliant to different specing options.
I'll try and upload the sheet I have when I have access to a webserve and link it for you in a PM so you can review my work and decide if you want to incorperate any of it into your own.
So...despite being very heavy with shadow. I should still use immo? And Is the order it has the spells below the main body the order they should be cast in? Sorry for the questions. I would love to refine my cast order to where it would be most optimum.
Yes, for pretty much every realistic gear set you should cast immolate no matter what spec you are or even what talents you choose to take. The number you should look at on immolate is damage per cast time, as long as that is higher than shadowbolt's damage per cast time it should be cast.
The sheet does not deal directly with DoT rotations per see. It does allow for user error by factoring in DoT gaps, a variable you can change if you want. Basically the only really good rules of thumb in my experience are to cast longer duration spells first on your first cycle so there's more space for shadowbolts in between rotations. After you toss the initial set of DoTs on and they begin to expire, however, strict cycles tend to break down.
The relevant informations a WL would like to know are, I beleave :
- Unlimited mana dps and how long can you keep it (ie full->0 mana)
- Self-sutained dps using DP (if available) and LT only what you can heal yourself (Siphon Life, Soul Leach, maybe 3/3 FSW)
- Sustained dps with unlimited heal
- What is your max dps in a X min fight
When I was making my XLS ( Warlock DPS XLS ) , I wanted to be able to get as result most of those you listed. Since I based my calculations on simulating real period of time (X sec), it wasnt hard to add some of those.
I was auto-calculating number of needed LT to provide exactly needed mana, ie never to let warlock go oom. Until recently, only two spells whose number of uses I wasnt calculating were DP and Drain Life.
For DP problem was that I didnt have accurate imp info, but since I recently added calculations for Pet dps also, i used that info to calculate number of DP possible to use in given period (X) , based on imp initial mana pool and on imp regen rate - so I do not presume that pet always have mana. Auto-calc for LT then adds needed number of LTs to provide all mana.
For Drain Life, I set to auto-calculate number of DL just enough to keep warlock self-sustained for heals (taking into consideration warlock initial healt pool). Since all those 'auto-calc' spells I can chose to include or exclude from calculations ("0"/"1" list box), I normally have DL deselected, and select it only when I want to check self-sustained scenario.
Originally Posted by Bolche
But I beleave for a affliction lock, the logic is :
DP as much as your pet can provide you the mana, else LT.
this depends on what you want to maximize, although we usually always want to maximize DPS. As I said, I recently added pet DPS calculations (taking into account mana pools of pets, with other stats), and I added option to include pet DPS in your total DPS.
Now, if for afflic lock with certain stats and for fight of 300sec I chose to use 'DP', i get ~984 DPS (that is self-sustained scenario, no need for heals). If lock can get heals and decide to use imp in fight instead of using DP, he would get ~997 DPS. So, in non-self-sustained scenarios, in theory it would be better to use imp to fight. Of course, for self-sustained scenario, it is better to DP ,because using DL instead drops dps to ~979.
That is all theorycrafting about use of imp, since difference in dps if used in combat is not big, and it is probably not always easy keeping imp alive during boss fights.
But while we are on theorycrafting area, I tried to see how would non-self-sustained afflic stand on DPS if he use succy instead of imp. If my calculations are correct, in identical situation from above, dps would jump to ~1089. Interesting part is that even for self-sustained scenario (using DrainLifes since no DP) , with succy dps would be ~1069, so better than original self-sustained
So, conclusion would be that maybe afflic locks could do better DPS with succy than with imp (DP or not), even for self-sustained scenario.
BTW, since I'm not afflic, I could not check those numbers, especially pet DPS numbers that I calculate for afflic (succy is at about 105 DPS for that particular lock stats, and imp is at mere 14 dps). I was able to check numbers for demo pets, and they were correct for imp,succy and FG on lvl ~70 mobs.
1. one versus two roll system for shadowbolt. I thought it was one roll, this thread indicated it might be two roll, and then I read another that said that a 10% crit level 25 warlock was getting 10% crit, ~15% hit (rest miss) against high level mobs which would indicate a 1 roll system.
We have to settle this once and for all..... Or perhaps just have a toggle in the spreadsheet to try each one and see if it matters anyway.
2. ISB uptime calculations. There are two build in models. What is the difference? I know what the first model seems to be (infinite SB spam individual uptime including overlapping procs). But the reality is, especially for affliciton, that the ISB uptime differs because only half your time is spent casting SB.
For example, I will get a SB crit, and have some dot refreshing to do, and so I rarely consume all the charges and shorten the duration. On the other hand, I don't get to consume all the charges with shadowbolt. So I'm losing some dps because a smaller number of shadowbolts get a 20% boost but gaining dps because more of my DoT's would gain, and the raid gains too.
A model that includes simulations with dot refreshes, life taps, etc. may be required. This probably can't be captured in a simple formula, especially if we try to model the mixing of multiple warlocks with different shadowbolt duty cycle times and crit rates.
Nenad it's hard enough to keep an 8300hp 50% AOE avoidance Felguard up on most boss fights. Calculating the DPS of an imp or succubus is rather pointless.
2. ISB uptime calculations. There are two build in models. What is the difference? I know what the first model seems to be (infinite SB spam individual uptime including overlapping procs). But the reality is, especially for affliciton, that the ISB uptime differs because only half your time is spent casting SB.
For example, I will get a SB crit, and have some dot refreshing to do, and so I rarely consume all the charges and shorten the duration. On the other hand, I don't get to consume all the charges with shadowbolt. So I'm losing some dps because a smaller number of shadowbolts get a 20% boost but gaining dps because more of my DoT's would gain, and the raid gains too.
Shadowbolts adn ISB uptime are very complicated for one reason. Crit both increases the number of times you proc ISB while at the same time reducing the average duration of ISB by critting again before 4 bolts or twelve seconds.
Currently on the sheet
Model 1: Assumes all ISB procs last 12 seconds reguardless of crit and the rate at which you are casting ISB. This is moderatly accurate for solo/10mans situations in an affliction build, less so in solo situations/10 mans in a destruction build, and completely wrong for raids.
Model 2: Assuems ISB procs end after the average time it takes you to cast 4 more shadowbolts. This is slightly more accurate for solo situations/10 mans in an affliction build, less so in solo situations/10 mans in a destruction build, and completely wrong for raids.
Changes in overall ISB uptime, reguardless of the model you select, are completely different from the actual changes you would see in ISB uptime in a raid situation.
I customized my own version of the sheet with a more accurate model of ISB that looks something like:
total uptime = total time * ISB/sec *AVG duration of ISB
% ISB uptime = ISB/sec * AVG duration of ISB
This is pretty much as accurate as you can get for single lock ISB as long as you factor in max of 12 secs durations, with raids it gets significantly more hairy because you have to factor in every locks SB/sec and ISB/sec i.e. crit rate at the same time. Those two numbers change significantly with gear and spec so without knowing your raid makeup ahead of time coming up with an exact figure is difficult.
Last edited by tetracycloide : 06/14/07 at 4:57 PM.
A model that includes simulations with dot refreshes, life taps, etc. may be required. This probably can't be captured in a simple formula, especially if we try to model the mixing of multiple warlocks with different shadowbolt duty cycle times and crit rates.
I made simulation program to calculate ISB uptime for complex situations with several different locks and shadow priests. It shows that our current ISB uptime formulas for single lock are fairly accurate, while it significantly differ with 2-3 locks
Nenad it's hard enough to keep an 8300hp 50% AOE avoidance Felguard up on most boss fights. Calculating the DPS of an imp or succubus is rather pointless.
I made simulation program to calculate ISB uptime for complex situations with several different locks and shadow priests. It shows that our current ISB uptime formulas for single lock are fairly accurate, while it significantly differ with 2-3 locks
That is one of the reasons I'm a firm believer in multi-player simulation..... It's the only way to accurately capture all the player synergy.
I'll be releasing new class/spec dps/dpm rankings, class dmg/hit/crit scaling, and trinket/proc rankings shortly.
The simulator is pretty much done at this point..... Just creating all the scripts to drive it for report generation now.
ISB/sec = 0.04
average ISB duration = 0.25 + 0.45 + 0.6075 + 0.729 + 6.561 = 8.5975 s
%ISB uptime = 0.04 * 8.5975 = 34.39%
My thoughts
For this specific scenario, ISB could only be up at the time of a cast finish if one of the four previous shadowbolt casts had crit. Prior to that the debuff would already have fallen off.
There is a 90% chance that each cast did not crit. Over 4 casts, the chance of not having a crit would be 0.9 ^ 4 = 65.61%. This implies that the probability of one of them having crit, and thus the chance of ISB being up, is 34.39%.
Leulier spreadsheet
Putting 10% crit into Leulier's spreadsheet gives back a 39.69% uptime.
Why is this higher? The fifth previous shadowbolt cast finished 12.5 s ago; if it had crit its ISB debuff would have faded and shouldn't be considered. However, Leulier still counts the 2.0 s that it was up into its calculation (to make the full 12 s debuff), and instead calculates the uptime as 1 - 0.9 ^ 4.8 = 39.69%.
For this specific scenario, ISB could only be up at the time of a cast finish if one of the four previous shadowbolt casts had crit. Prior to that the debuff would already have fallen off.
There is a 90% chance that each cast did not crit. Over 4 casts, the chance of not having a crit would be 0.9 ^ 4 = 65.61%. This implies that the probability of one of them having crit, and thus the chance of ISB being up, is 34.39%.
Leulier spreadsheet
Putting 10% crit into Leulier's spreadsheet gives back a 39.69% uptime.
Why is this higher? The fifth previous shadowbolt cast finished 12.5 s ago; if it had crit its ISB debuff would have faded and shouldn't be considered. However, Leulier still counts the 2.0 s that it was up into its calculation (to make the full 12 s debuff), and instead calculates the uptime as 1 - 0.9 ^ 4.8 = 39.69%.
This is pretty much as accurate as you can get for single lock ISB as long as you factor in max of 12 secs durations
I noticed that you posted this in other thread, where we discussed simulations, but i didnt have time to check numbers then. Now I tried to compare this , and here is what i observed:
1) above formule result in lot of cases with average ISB duration of over 12sec, mainly in cases when SB is casted under 60%-70% of time, which is to say, in majority of cases. I believe that when you said "factor in max of 12 secs ", you basically use AVG ISB duration of 12sec if above formula result in longer. But in those cases your formula transform to very simple one:
2) only for unrealisticaly high SB/sec values, above formula is used completely. For 10% crit it requires over 70% cast time spent on SB, and for 20% crit it requires more than 60% cast time spent on SB, in order to use complex part of formula. Considering that usual time spent on SB is 45-55%, it means that even very simple model from above would cover most realistic cases
3) I tried to compare numbers of ISB uptime for several different SB% and crit%, between my simulatro, my xls, tertacycloide's formula and euilers xls, and results are something like:
in above example, "simple" version of tetra's formula is used for 40% and 54% Sb time, and "complex" for 81%. Aside from small variations, it seems that it is indeed coming closest to simulated values, if we consider simulation as most reliable of above ;p
Another thing that I don't believe anyone is trying to simulate yet revolves around the mechanics of mob debuffs and spells with travel time. The damage is always calculated at the time of the cast, rather than the time of contact. You could thus, in theory, have 1 charge left on ISB, and 5 warlocks all release Shadow Bolt at the same instant. Regardless of which shadow bolt hits first, even if none of them crit they will all receive the full benefit of the debuff. While this effect won't be a huge benefit to ISB, it will increase it's value slightly.
1) above formule result in lot of cases with average ISB duration of over 12sec, mainly in cases when SB is casted under 60%-70% of time, which is to say, in majority of cases. I believe that when you said "factor in max of 12 secs ", you basically use AVG ISB duration of 12sec if above formula result in longer. But in those cases your formula transform to very simple one:
Taking the average ISB duration as 12 secs when the formula comes up with longer is probably the easiest way to do it but there is another slightly more accurate way to do it.
Another thing that I don't believe anyone is trying to simulate yet revolves around the mechanics of mob debuffs and spells with travel time. The damage is always calculated at the time of the cast, rather than the time of contact. You could thus, in theory, have 1 charge left on ISB, and 5 warlocks all release Shadow Bolt at the same instant. Regardless of which shadow bolt hits first, even if none of them crit they will all receive the full benefit of the debuff. While this effect won't be a huge benefit to ISB, it will increase it's value slightly.
Are you CERTAIN about the "damage is always calculated at the time of the cast"?
Are you CERTAIN about the "damage is always calculated at the time of the cast"?
Yes. Unless ISB is different than every other debuff in the game. Throw a SB and spam CoS as it leaves your hands, then throw another SB. The results speak for themselves. Shatter, Misery, and shadow weaving all work the exact same way, in that their effects are not calculated till next completion of cast time for direct damage spells. DoTs calc on a per tick basis as far as debuffs are concerned).
Yes. Unless ISB is different than every other debuff in the game. Throw a SB and spam CoS as it leaves your hands, then throw another SB. The results speak for themselves. Shatter, Misery, and shadow weaving all work the exact same way, in that their effects are not calculated till next completion of cast time for direct damage spells. DoTs calc on a per tick basis as far as debuffs are concerned).
Have any parses to back this up? General rule is debuffs are checked when damage is delt and SB doesn't deal damage till it reaches the target.
Another thing that I don't believe anyone is trying to simulate yet revolves around the mechanics of mob debuffs and spells with travel time. The damage is always calculated at the time of the cast, rather than the time of contact. You could thus, in theory, have 1 charge left on ISB, and 5 warlocks all release Shadow Bolt at the same instant. Regardless of which shadow bolt hits first, even if none of them crit they will all receive the full benefit of the debuff. While this effect won't be a huge benefit to ISB, it will increase it's value slightly.
I believe it does not change ISB effect , from at least 2 reasons:
1) if we have small increase to effect of ISB if 5 locks cast nuke at last 'tick' of ISB, as you said in example, we have identical decrease to effect of ISB if those 5 locks cast nuke at same time without ISB, and nuke from 1st lock crit. In this case, none of others will receive ISB buff because their damage was defined before crit.
Therefore, we can ignore that 'travel time' issue (although it is interesting one) due to fact that it has opposite effect on start and on end of buff period.
2) also, it is possible that, if damage is calculated at time of cast, removal of one charge from ISB is also done at same time.
Have any parses to back this up? General rule is debuffs are checked when damage is delt and SB doesn't deal damage till it reaches the target.
Well, the -resist is a debuff but the +dmg% is a buff. Maybe it's calculated at different times. Percentage bonus when you start the cast and -resist in the roll system. I don't have any data for this, I'm just giving ideas on how to answer the doubts.
Another thing that I don't believe anyone is trying to simulate yet revolves around the mechanics of mob debuffs and spells with travel time. The damage is always calculated at the time of the cast, rather than the time of contact. You could thus, in theory, have 1 charge left on ISB, and 5 warlocks all release Shadow Bolt at the same instant. Regardless of which shadow bolt hits first, even if none of them crit they will all receive the full benefit of the debuff. While this effect won't be a huge benefit to ISB, it will increase it's value slightly.
So what is an average "travel time" for Shadow Bolt in your boss fights?
It is certainly easy enough to add an option to some spells for delayed damage.
I just have to throw another event on the queue instead of applying damage at that time.
But I -do- need a delta-time to use.
What other DD spells among the caster classes have travel times?
06/07/07, 12:09 PM
Similar Threads
