This is only easier if you know the exact duration of a fight ahead of time. Using my formula above you only need to know how much time you will spend in a fight casting SB as a percentage of total time which is readily avalible on the published spreadsheet and since you're dealing with percentages of total time the total time cancels out leaving you with meaningful information no matter the duration of a fight.

It's very hard to see how multiplying the time between each 2 spells in the rotation for every single SB across the entire fight is in any way 'easier' to calculate then just letting time cancle out of both sides and use percentages on both. As soon as your fight starts to last 2 mins or so you're upwards of what, 24 SBs to average together with individual calculations for the time between each one just for one affliction lock. Pulling stats directly from the spread sheet and plugging them in is advantageous, requiring a time limit before you can produce useful information is not.

The loss of uptime for the OTHER shadowpriests and warlocks makes the above statement not so clear cut.

20% boost to one mind blast is roughly equal to 2/3 of one tick of mind flay. Soooo.... 3 mind flay ticks in the raid total, or 1 corruption, one flay, and one SW:P or VT, then the gain to mind blast is smaller than the loss to the raid.

Simply put: If you are in a 5-man and you and a 'lock are sharing the charges, sure, MB is ok (if you aren't threat limited). If you are in a raid, with 2 or more shadow priests or a total of 4 shadow casting classes you are wasting raid dps and mana return.

A mind blast is roughly 2x the dps of mind flay. For 1.5 seconds. If that ISB proc can last an extra couple seconds by not using it you will gain more from leaving it up there longer. The VAST majority of the mana return of a shadow priest is from 2 dots and mind flay. The two nukes are a bonus, but 20% extra to the majority is > 20% to the minority.

Don't decrease ISB uptime and lower mana return with Mind Blast. If you are a 'lock and there is one charge left, leave it up and tap/pact so its up there longer. If you are in a 5-man, and both cooperate you can avoid using that last charge until near the end of the 12 seconds and both gain....

This is unfortunately flawed. It has to be a lot more complicated than that one equation.


Firstly, this is only true if both of the following are true:

1. That there are always at least 4 shadowbolts in a 12 second period. This is somewhat realistic in a larger raid. In Kara, with affliction locks, or in a 5-man the rule is broken all the time.
2. That all of the shadowbolts are coming in at equal intervals. Or at least in truly random intervals. Not possible.
3. That the average of all shadow DD casts ISB proccing frequency is also valid in the equation above.

#2 is important. If all you cared about was the % of shadowbolts getting the bonus, you could neglect it. (1-(1-crit)^4) is the chance that any individual SB will get a 20% boost.
But we want the UpTIME. Too many DoTs and channeled spells depend on that. The unevenness of the distribution has a strong effect since the equation is essentially the chance in terms of SB casts that are not evenly spaced in time. Over long periods, with casters who don't try to extend the uptime by changing cast priority, its ok however.

#3 breaks it. Two warlocks, both with 40% crit give a very different uptime number form two warlocks, one with 80% crit and one with 0%. Push the extreme, 2x 50% crit versus 1x100% and 1x0%. In the latter, we have 100% ISB uptime. In the former, 94%. If you mix the crit chance and casting frequency of multiple casters the formula can't be used anymore.
If you can assume that the number of casts from each warlock is the same, and the shadow priest (in your example) is as well, then you can. But this won't be true. Different spec warlocks will have different SB per minute rates, and the number of MindBlast/SW: D casts is absolutely less.

How could you figure in the following:

1 lock, affliction average 1 shadow bolt per 5 seconds 18% crit.
1 lock, destro, average 1 shadow bolt per 3.25 seconds, 24% crit.
1 SPriest, one shadow nuke per 10 seconds on average.

Well in a 12 second window, there will on average be (12/10 + 12/5 + 12/3.25) = 7.02 casts so assumption #1 is probably correct and we can neglect the "buff falling off without being used" case.

Now what? Well the average number of nukes out of the last 4 that each class contributed is:

aff lock: (12/5 / 7.02) *4= 1.3675
destro lock: (12/3.5 / 7.02)*4 = 1.954
SPriest: (12/10)/ 7.02 * 4 = 0.684

the odds that a given nuke is from each are just 1/4 of the above.
aff: 34% of nukes
dest: 49% of nukes
SP: 17% of nukes

Ok, now what? We know the average number of casts out of the last 4 from each caster, and their ISB proc rate. What is the chance that the last 4 didn't proc ISB?

First lets just calculate the odds of any one nuke proccing ISB:
The average nuke is 17% shadowpriest, 49% destro lock, and 34% affliction lock. The average ISB proc chance is
.17 * 0 + .49 * .24 + .34 * .18 = .178 = 17.8%

This is where the tricky part comes in. I will neglect the findings of criticism #3 above. IF the last 4 shadow damage spells represent these 3 _truly randomly_ then the 1-(1-chance)^4 rule applies. And the answer is about 45%. But things aren't truly random. The example used earlier illustrates this, and it is the flaw with this equation.
Two warlocks, one with 100% crit, one with 0% crit. If you assume total random events, then 94% is the "answer" which is the same as two 50% crit cases. However, the 6.25% time without it up represents times when warlock A with 0% crit gets 4 bolts off in a row without warlock B with 100% getting one. It is a flaw in the assumption of randomly distributed casts between all casters.

In the shadow priest case above, if the average time between casts is 10 seconds, and we use the 1-(1-chance)^4 rule and the average proc chance, we are assuming truly random distribution of casts. This includes impossibilities such as 4 shadow priest casts in a row without others casting (chance, 0.17^4) and other 'streaks' that just won't happen. The reality is, casting from the affliction lock will be in bursts of 2 to 5 in a row, 2.5 seconds appart, with breaks inbetween. The Destro lock will have long burst sessions with short breaks for dots / tap. And the shadow priest will use their nukes on specific cooldown intervals with small variance. All of those things break the 1-(1-chance)^4 model.

Again the issue is not with how to calculate it, it's with how to model the casting sequence. Which I agree is complicated to the extreme, but GIVEN a casting sequence, the calculation isn't hard. Make an estimated casting sequence, and get an estimated uptime. The calculation with chance of SB to not be up is as accurate as you can get with a given accuracy of the cast sequence.
I didn't claim to model anything accurately. But with a given cast sequence, calculating imp SB uptime is easy.

No warlock build in any boss fight currently in TBC will ever cast nothing but shadowbolts. Therefor, while it is correct to say that ISBuptime is the same as the chance that any specific shadowbolt will get the ISB buff when you spend 100% of your time casting shadowbolt, this formula is completely useless even as an ISB model for just one lock because 100% time on SB is never the case.

In a raid setting with several locks casting shadowbolts with different crit rates and different amounts of time spent casting shadowbolt the formula is even less useful in that it offers no way to calculate the raid ISB uptime without being given a complete casting sequence for the entire fight ahead of time. The other raid ISB model takes the information already avalible on the spreadsheet, like % casting time on SB and % time on NF, and produces meaningful ISB uptimes no matter how many locks you have.

The best way to illustrate that the raid model with %cast time taken into account is superior for theorycrafting was your post earlier in the '[Warlock] how do you do high DPS?' thread that came to the conclusion that adding locks at the same crit rate will not increase ISB uptime. Once you take %cast time into account you can come to the conclusion that adding a second lock with increase ISB uptime even with a lower crit rate. There are also so major implications for valuing +crit on gear that will vary wildly with models that overvalue +crit, like one where you cast nothing but shadowbolts, and ones that value +crit based on their actual contributions, like one where your mainting ISB as well as several other debuffs and are playing with other locks around.

I beg to differ on this slightly. We've already seen parses from Warlocks that do significant damage on fights casting nothing but shadowbolts from a 0/21/40 build. I believe there was at least one posted in the 'How do you do high DPS' thread, with the WWS here:

Wow Web Stats


I also use a 0/21/40 build, and very frequently get my highest DPS when I am doing nothing but casting shadowbolts, even on 'Movement' fights. My guild is not incredibly progressed yet, as we are still trying to down Gruul. I have managed to do pretty good DPS, with an acceptable DPS time.

Of course, I'm supplying my own CoS right now. Optimally, one of our affliction locks will respec for Malediction, and then I will be casting Doom as well, but I don't think a 1.5 second GCD once a minute changes the math on shadowbolt spam very much. In a latency free, movement free world, that's 23.2 possible shadowbolts instead of 24 every minute, or an ~3% change in the frequency of shadowbolts landing on a sustained basis.

Edit: I should have tested that link, the report seems gone now, or I've failed at linking.

The existence of a high dps WWS doesn't mean that's the right way to do it. With Dommius's gear, crit rate, etc., adding Immo into the rotation improves his DPS by about 5%. Damage per Casting Second is a difficult and stern mistress.

I don't play a warlock, I play a mage, but I think something to keep in mind is that heavy spec changes require (usually) heavy gear changes. What's good for one spec may be worthless for another, etc. I can't claim to know how warlock specs work, but keep in mind the other posts in the thread when you regear.

Come on, we are talking about spells you use when you are 0/21/40, aren't we ? if you include lifetap in, do we also need to include bandage time in ? :S

For 0/21/40 fire destro build, people know that they just need to immolate -> incinerate -> conflagrate. But some people don't know or are confused about what they should cast if they are shadow destro build. Spamming sb or Immolate -> shadowbolt ? OFC you will cast CoD or CoA when you are not CoX duty but that is not what we are talking about right now.

Before I can't figure out why people should spam shadowbolts(except CoA or CoD or CoS/E/R), then I think now maybe it is becos sb scales very good probably the best ? And with 5/5 S&F and DS maybe you just should cast it as much as possible in order to maximize the outcome of this build.

If you're spending time bandaging you're doing something wrong. Get on your healers to throw some renews around and start using more HSs and pots. In most fights there should be no need to bandage. Also bandages heal for a piddly amount considering the time they take.

For fire destro (or any build ever) you should NOT be casting Conflag. It's been proven many times that the part of an incinerate/SB + 1 immolate tick > conflag damage. Conflag = bad.

For shadow destro SB spam + curse & LT seems to be the best bet. The inclusion of corruption and immolate seem to add or subtract minimal amounts of DPS and leaving them out seems to be the general consensus. But as a matter of personal preference you can leave them in your spell lineup.

You spam SB because it has more dmg/CT and it will also proc ISB which will boost the entire raid's shadow damage.

This isn't necessarily true, as not all raids are built equal. If you are full shadow destro (No improved immo, no Emberstorm) and have shadow priests, but no fire mages adding fire Vulnerability, shadowbolt is significantly higher DPC.

According to the leulier spreadsheet, at 1140 Shadow damage, 23% crit and 14% hit and 990 fire damage with shadow debuffs but without fire:

The DPC on shadowbolt is 1259.79, using a 46% ISB uptime.

The DPC for immolate with such a raid, and build, is 1141.19

I suppose it could be said that spec'ing for emberstorm and improved immo, and losing cataclysm is better. According to the spreadsheet, you get about a 15 DPS gain from 1200 to 1215 under those conditions by casting immolate. That's what, a 1.25% DPS increase?

I bet you lose more than 1.25% just in managing your casts to make sure to get immo up every time it falls off. Yes, this is a stacked example, but the point remains that with full shadow buffs and a lack of others, you're better off just casting shadowbolt.

What I was saying is the spell you were talking about means sb incinerate immolate CoA etc. which I think lifetap and such do not count.

About conflagrate, I tried to but failed to find those threads out which prove that adding conflagrate into the rotation is just not worth it. And I do remember a fire lock from DnT saying that if your gear is pretty good ( which means having 1200-1300 fire dmg with fel armor like I do now ) that it is totally worth it casing conflagrate before immolate runs out. Also, Eoy from Forte did use conflagrate too. For me conflagrate is usually 1.9-2k non crit and around 4k or even a little bit more if it crits so I am kinda confused in here about that :/

Why wouldn't lifetap count? You spend global cool downs casting it so it counts, this seems like a really basic concept to me.

Conflagrate is a DPS loss because after you subtract the damage lost by consuming the last tick of conflagrate the damage per cast time on colfagrate is significantly lower than either shadowbolt or incinerate. It doesn't mater what level of +fire you have it is always a DPS loss to cast because the coefficients for conflagrate is lower than either spam skill.

Again you still don't get my point, do you ? Lifetap does have a GCD and it does count, everybody knows it ofc. But not in this case.

As I stated before, what that guy was saying was what _spells_ he should use if he is 0/21/40 shadow destro build ? Spamming sb only or immolate and sb. Then you replied his thread saying there are no builds that you can get maximum dps just by spamming sb. So that is why I said if you count lifetap in, why not count bandage and such in too and obviously I was not being serious... Because counting those things in has nothing to do with what that guy was talking about, it was totally ot tbh.

I don't really get your reason of conflagrate being a dps loss. Do you mean that spending those GCDs after casting conflagrate on casting incinerate will output more dmg than conflagrate gives? If that is what you mean then it kinda makes some sense to me.

I don't think you understand what the posts above were talking about then. The issue under discussion was not 'what spells should be used as 0/21/40 to maximize DPS' but 'what is the expected ISB uptime and how best can we model it.'

ISB uptime is extreamly dependent on the rate of SB/sec and therefor taking lifetaps into account is vital because it's a 13% reducetion in the rate of SB/sec at least. I didn't include bandaging becuase bandaging should not be necessary and even if it was would not make up nearly as large a percentage of time. Counting those things has everything to do with what we were talking about.

I think you may be confused because you came into the thread and commented on a response to a response. Perhaps if you tracked all of the responses back to the original source it would be clearer how the conversation is deveoloping.

As for conflagrate, yes, I mean that using the GCDs on casting incinerate/shadowbolt and simply letting your immolate tick to full will output more damage than conflagrate gives.

Except for Mother Sharaz I can't think of a fight where I would rather have Imp Immo and Emberstorm over Cataclysm, Nether prot, and 2/3 leech.

Do your affliction spells not have a chance to miss somehow ...?

Suppression should cover all your +hit worries for affliction spells. As your total +hit from gear goes up, the number of points you put in suppression can go down.

There was also discussion in one of the other threads here (I forget which one) about the effect of resists on DoT rotations. The conclusion was that a resist for an Affliction Warlock has less of a negative impact on DPS than it does for a Destruction Warlock.

On top there is no point talking about DoT resists for affliction as they should inevitably have maxed out 99% chance to hit. Hit for affliction therefore only affects non-affliction spells, which are a smaller % of your damage than they are for desto or demo.

I still don't see for a single lock how casting more shadowbolts incrases shadowbolt uptime, and none of the posts on any of the threads really explain WHY. Considering all that is being casted is DoTs and shadowbolts, so the only thing that applies and consumes ISB is the shadowbolts themselves, casting 2 million shadowbolts/sec, the chance for ISB to not be up is still the chance for the last 4 shadowbolts to have been non-crits, which is the exact number if I cast 4 shadowbolts in anything faster than 12 seconds. If there really is a difference for a single lock spamming 2 million shadowbolts/sec and firing 4 shadowbolts in 10 seconds, please show me what's missing. Note that this chance is for *any given time* not just when a shadowbolt hits.

More than just DOTs and Shadow Bolts are being cast. ISB uptime calculation is for a raid environment (as it is fairly trivial and, well, pointless, to calculate for solo/5-man play). So you have to worry about Shadow Priests using Shadow Word: Death and Mindblast. ISB charges may also be eaten Shadow Bolts incapable of proccing ISB (or cast by a warlock with a lower Crit % who is, in effect, leeching your ISB charges), Shadowburn, shadow wand damage, and, if Horde, Shadow Guard. Perhaps you even need to consider Lash of Pain, although I doubt there are many 0/40/21 Succubus specs raiding.

Given the large number of ISB charges consumed by something other than your own non-crit Shadow Bolts, a single Warlock casting more Shadow Bolts will increase ISB uptime relative to that same Warlock casting spells other than Shadow Bolt.

Edit: Missed a typo when I proof-read.

While the amount you can put in supression goes down, there really isn't anything else to get unless you are doing the CoA/Shadow Embrace route instead of Malediction.