@Curan

The spell priority on the first page is questionable. Math has shown in another thread here that when you take into account the opportunity cost of casting MB on cooldown (you're not casting a MF), that DP comes out ahead in priority. I discounted that math when it was first shown, but have recently converted. My understanding is that the correct priority is:

VT
DP
MB
MF

I did the math, and while losing one dot is less than the extra overall VT damage you would gain on a big SP proc, the one or two ticks of mind flay you would lose while recasting it outweighs that gain.

Rawr is not a more realistic environment.

Get 2/2 Imp VE and then either do 2/3 Shadow Affinity, or 1/3 Shadow Affinity + 1/1 Psychic Horror. The extra threat reduction is really helpful for Mind Sear Whoring in AE situations and on fights like Hodir and Vezax where you can do absurdly high damage.

Imp PW:S is a terrible talent for a shadow priest. You should be doing damage, not wasting GCD's casting weak PW:S's on yourself. It's also a negligible amount of extra hit points gained on the shield versus what Imp VE allows you to heal yourself for.

You just have to accept our best spec is 14/0/55+2

Hi.

Been skimming over this thread. And one thing i have not notice is no one is talking bout partial resists.

I was wondering if there is a curse, stat, or class that brings something to a raid that can stop partial resists.

I raid with an ele shammy, boom chicken, 2-3 Mages (either FFB or TTW spec), and varying Locks. The locks i know are Destro but none seem to go ISB or put CoE up because it gets eaten by Ebon Plague and Earth and Moon.

Just wondering if there is any math or information that can help me.

EDIT: Im a Spriest now. So ignore the Lock tag to the side.

Level based partial resists cannot be overcome, except by levelling up. Bosses are always considered as +3 levels compared to you, so you suffer level based partial resists as if target is 3 levels higher. And there is no way to overcome it.

Shadowpriest.com is down, is there another website with BIS Info?

A few pages back there was a little bit of math that showed prioritizing Mind Blast over Plague would (slightly) increase your damage, while putting Plague over Blast will (slightly) increase your total available movement time. Casting Plague first doesn't cost that much damage, and it's not that much extra mana either. So for fights like Yogg Saron, Freya, or Hodir, it makes a lot of sense to push the priority of Plague higher.

For the record, I usually cast Blast over Plague though. I don't think I'd ever "not" cast a plague so I would have something to cast while moving, unless I knew I was going to move sometime in the next 3 or 4 seconds. Maybe my mind isn't trained to see situations like that, but I can't think of how often that happens.

I'm wondering about the rotation as well. I cast:

VT-MB-DP-SW-MF-SW:P(5 stacks)..then MB-SW when each come off cooldown, and replace VT-DP as necessary. MF fills in when I have nothing else to cast. I've been doing testing on a target dummy with self buffs and I haven't been able to get past 2667 peak and 2455 once the meter dies down and the dots have completely fallen off the dummy. I don't have latency issues, and I'm using the casting bar to make sure I'm not clipping my MF. I can't seem to figure out what I'm doing wrong...my dps seems low to me, but I don't really have any other shadow priest dummy testers to compare it too, lol. Anyone have any insight? It'd be much appreciated.

*EDIT* Just took SW back out of the equation. DPS went to over 2700. I think this is something frowned upon, but in case it isn't I'm going to ask anyways...what do you all typically get on a dummy with self buffs?

It happens quite often - for example, on Thorim, if you see the lighting strikes in your direction, you can usually fire off a mindblast quickly and then move during the plague gcd. You can do the same thing on mimiron hardmode as the fire is walking in your direction - you can tipically get off a last mindblast, then move away from fire during the plague/death gcds. If you plague first, you are stuck moving while mindblast is off cooldown not being casted.

On Hodir you can do the same with patches on the floor - if you are in a lightwell, you have plenty of time fireoff a mindblast and move during the gcd before you get knocked back.

My main point though is that VT/MB/DP are all close enough priority that you won't suffer a huge DPS loss for using one over the other, and mobility is a huge concern on a lot of fights, so mobility often trumps theoretical priority when thinking about what spells to cast.

Sections for fights that have special considerations. Vezax hard, for how you have focus on producing the most damage you can with your one mana bar. Looking at DPS means nothing. DPM is what matters.

Something else that is worth talking about is cast priorities on Vezax hard and Hodir hard. For Hodir, if you're forced to choose between Toasty Fire and Moonbeam, which do you use? How do your spell priorities change in either case?

I want to explicitly thank you for this post, even though my post will be mostly devoid of content.

I had been running sub-par on DPS for a few weeks on my alt shadowpriest and couldn't figure out what was wrong. This is my first DoT class to play and I think I was not completely aware of the issue with the timing of the ticks, especially the difference in SWP vs VT behavior. This point about not clipping the final VT tick is a really important one for beginning shadowpriests to be aware of. Once I started minding the VT expiry carefully and only refreshing after the last tick, basically having the VT cast in-flight as the last tick went off, I got significantly better DPS performance. The advantage is two-fold, first you do not lose the last VT tick nor the start-up time of the next VT, and second you end up being able to cast an additional tick of MF or possibly an extra MB or DP because you are delaying the VT refresh.

So basically, thanks for making this mechanic "click" for me.

We need to see actual numbers backing up this claim, but I've been thinking about it more. The "opportunity cost of not casting mind flay" doesn't seem to apply to the decision of Mind Blast versus Devouring Plague, because no matter which you cast, you're not casting Mind Flay either way. So you deduct the same opportunity cost from either one.

There was the suggestion that perhaps the best way to analyze a priority queue is to write down the optimal cycle for that queue and compute the average DPS, although this is a huge pain. I did say I would try to think of a better method, so here it is.

Priests have five damage spells they could use in a rotation:

• Vampiric Touch
• Devouring Plague
• Shadow Word: Death
• Mind Blast
• Mind Flay

(Shadow Word: Pain is omitted because Pain and Suffering refreshes it automatically.) There are 5! = 120 potential priority orderings of these five spells. Rather than consider all 120 options, we can derive the optimal priority by starting with a simple one spell cycle and adding one additional spell at a time. The simplest priority queue is:

• Mind Flay

Or just cast mind flay on cooldown. There are four possible extensions of this cycle:

• Vampiric Touch > Mind Flay
• Devouring Plague > Mind Flay
• Mind Blast > Mind Flay
• Shadow Word: Death > Mind Flay

(Technically there are 8 extensions, since (Mind Flay > Vampiric Touch) and (Mind Flay > Mind Blast) are mathematically expressible options. But these cycles can be ignored since Mind Flay has no cooldown, making them equivalent to the (Mind Flay) queue.)

For example, the (Vampiric Touch > Mind Flay) cycle means refresh Vampiric Touch when it expires and otherwise cast Mind Flay. Conceptually you can think of yourself as normally in the (Mind Flay) queue, but you switch to the (Vampiric Touch > Mind Flay) queue when that's an option, as that queue has higher damage.

The problem, of course, is what you do when you want to upgrade to both the (Devouring Plague > Mind Flay) and (Mind Blast > Mind Flay) queues at the same time. Do you merge into (Devouring Plague > Mind Blast > Mind Flay) or (Mind Blast > Devouring Plague > Mind Flay)? I posit that priority queue mergers are distributive. In other words, if (A > Mind Flay) is more damage than (B > Mind Flay), then (A > B > Mind Flay) is more damage than (B > A > Mind Flay). This is provable given a great amount of futsy math, but I'll spare everyone the details and skip to the upshot.

We can just analyze the relative damage values of each two spell queue, rank the queues, and merge them together for the official priority queue. Note that this implicitly takes into account the opportunity cost of not casting Mind Flay. Here are the DPS values for each of the two spell queues. In the math, MF is the total damage of Mind Flay, VT is the total damage of Vampiric Touch, and so on.

• Vampiric Touch: [ VT + (MF * 13.5/3) ] / 15
• Devouring Plague: [ DP + (MF * 22.5/3) ] / 24
• Mind Blast: [ MB + (MF * 5.5/3) ] / 7
• Shadow Word: Death: [ SWD + (MF * 10.5/3) ] / 12

Here are some actual values from a recent raid:

VT = 14423
DP = 19226
MB = 5959
SWD = 4497
MF = 8790

• Vampiric Touch: 3599 DPS
• Devouring Plague: 3548
• Mind Blast: 3153
• Shadow Word: Death: 2939
• Mind Flay Only: 2930

So that suggests the correct priority queue is in fact Vampiric Touch > Devouring Plague > Mind Blast > Mind Flay.

Note how this table shows that Shadow Word: Death is almost completely irrelevant, and that it's okay to priority Plague above Touch if you really need 1.5 seconds of movement.

Tedv, that is an excellent analysis. Thanks!

The opportunity cost I mention is actually the cost of having to skip another half-mindflay sooner for the rest of the fight due to casting the nuke sooner. If you do this enough times with a short cooldown ability, you will get to the point where you have lost an entire mind flay (or half) due to casting one ability before the other in collisions (as in you have cast one or two extra nukes). I go into much more depth in the other thread, but I believe the effect appears in your analysis:



shows clearly the contribution in DPS of the main nuke (in this case VT) and the filler spell (MF). As the nuke cooldown decreases, the total contribution of Mind Flay decreases.

Edit:

This is mathematically the same result as the effective damage per cooldown calculation. We can subtract the DPS of Mind Flay from each result to show only the increased DPS by using the Nuke.






Now, we can take this DPS lost and integrate it over one Global Cooldown (postponing it to cast another nuke), and get



which is identical to the result Baalor came up with in post #58 of the other thread. That post uses the method of Tifi (the OP), so I think that your separate method reaching the same conclusion confirms this as the current best method for determining cast order (assuming equivalent collisions).

That's an excellent analysis Tedv, and what is particularly striking is how remarkably close all those spells are in priority. It would be interesting to calculate how much damage do you lose for fucking up all your priorities except keeping mindflay last - or, cast VT/MB/DP randomly when they are up/not on the target, and mindflay when one of those is not up versus keeping an optimal cast sequence.

Kalman had written a very quick and dirty shadowcraft sim in matlab that is probably pretty easy to hack up, simcraft is excellent but a bit of a pain for this kind of analysis.

The "opportunity cost" does matters even for priority orders, because the priority order determines how often you'll have to pay that cost. I'll be strict on this point, you can read my posts in "Effective damage per Cooldown" thread.

However, your method is equivalent, and you compute the same results. Just remove the dps of Mind Flay, and you get Tifi's eDPC. In fact, you do take into account the opportunity cost (because you just fill your cycle with one nuke and only MF, for the nuke CD duration).

This being said, I'd like to remind everyone that all these methods assume that collisions of nukes are independant of your priority list, which is clearly not valid.
Given that dot's uptime (or the use of nukes) are a huge factor in SP dps, and that the eDPC is quite close for the main nukes (Touch / DP / Blast), I think that we ought not to ignore the collision factor. A good priority list would be the one that favor Touch > DP > Blast quite often, but also leads fewer collision if possible...

should shadow priest use sw : death in rotation?

This doesnt include SW: P?

I have just speced shadow and trying to find a general rotation for instances/raids.

I get the impression from reading that there is no ONE rotation that works in every situation and that it is based alot on the fight itself.

Is there a basic one i should start on and experiment myself, or would the one quoted above be worth sticking to with SW:P added in?

Of course you should use sw: p. But only after you have 5 stacks of shadow waeving. The rotation you basically start every fight with is: vt-> mb -> dp -> mf -> SW:P (add ve if you like it and need it from the very beginning of the fight)

From that point on it comes simplified said down to keeping your dots up as best as possible + keeping mb on cooldown. The reason why sw: p wasn't listed in your quote was that once you've applied sw: p it should be up the whole fight anyway.

It should be pointed out that on any fight where you have a +dmg or +crit modifier that might change throughout the fight, you should reapply SWPain once you have the modifier. Some of these include:

Thaddius charge stacking buff
Sparks on Malygos
Hodir crit buff
Etc

Since Shadow Word Pain retains all +dmg/+crit modifiers when it is refreshed (as far as my understanding goes) you can apply SWP one time with the buffs above and keep the modifier on your spell when you refresh it. Basically, on Hodir, if you get the crit buff once early on and apply SWP while you have it, your SWP will have the increased crit goodness as long as you never recast SWP.

This is purely based on how I understand the modifiers to work on SWP. Spell power IS changed dynamically (updates with refresh) while +dmg/+crit stays with whatever you had affecting you when you cast SWP.

Other than your opener there is no rotation only priority.
VT > DP > MB > MF > SW: P is posted around here somewhere to be the highest possible DMG opener based on perfect CD management and Perfect Debuffing using the simcraft model.
However, there is a slight Delay with regards to the Application of the Misery Debuff to the target (ingame) which can sometimes cause the 'On Application' Dmg of DP to miss even though the spell lands.
For this to not happen VT > MB > DP > MF > SW: P works better and is only marginally worse.
This allows Misery to pop up while MB is being cast.
My Priority is based on my opener - VT, MB, DP, MF. Its pretty solid. I get VE up on the move or i tag it to a cast of VT or MB after my opener has completed.

Last I checked, the improved DP misses were their own bug, unrelated to any delay associated with misery. I haven't checked recently, but I know there's been at least one fight where imp. DP missed 3 times, when I knew misery was up 100% of the time.

Recently I haven't been seeing imp. DP misses, but this could be any combination of me not paying attention or not looking for it, or a hotfix on Blizzard's side, or me getting lucky.

Imp. DP still misses.