I recently started playing around with shadow for the first time since leveling my priest. I'm pretty experienced with the two healing specs of the priest, but I'm very new to shadow (and have quickly learned I'm not very good at it).

I just have a simple question; When, if ever, should I interrupt MF channeling? On MB cooldown? When VT or DP needs refreshing? Should I interrupt right away, after next tick, or not at all?

In the beginning you probably shouldn't. Eventually you may get an idea of your "rotation" in advance and see that if you clip MF2 because VT is running out and MB or DP will be up for refresh after VT is applied. Keep in mind this is just minor cosmetics/optimization and not a direct cause to poor damage. If you're running with other spriests try to compare damage percent by spells. If there is a similarity you should look at amount of dot ticks and MB/MF hits. Don't focus too much on how hard the dots tick.

From the original context, Ghostcrawler was simply stating from a high level design perspective, spirit would not be as good as other DPS stats because it also provided regen, but that doesn't mean it cannot provide damage. In fact every cloth/leather DPS caster currently gains a DPS benefit from spirit, but right now spriests gain the least value from spirit by a pretty wide margin.

Hi, I did the research but can't find anywhere what I'm looking for... I will explain:

on Naxx25 gear I was over 330-350haste rating that makes me clip mf2 on every mb.

Now, on mid ulduar/trial gear I am over 430-440 haste rating plus 240 from t8 4pieces bonus.

The problem is... I'm on the same DPS aprox... wich makes me think I'm doing it wrong with the Haste thing...

I try to keep up dots and renew it on trinket procs but something it's not working... 2.9 DPS on the training dummie...

How many haste do we need to clip mf3 on every MB? From what i've written, is anything wrong?

I feel disappointed because i fell like i can't do it better...

I mathed this out awhile back. I believe you need something like 26% haste, before haste buffs, to achieve mfx3 in between MB cd's. It's impossible, or at least not worth doing.

However, two little points you should consider. Haste still *slightly* increases DPS even after you can fit in 2 full mind flays just because you can cast mind blast a tad quicker. I'm in a minority position on this, but haste (once you have enough to comfortably cast two mind flays between mind blast), drops significantly in value compared to Crit.

The other thing you should consider is going from 80% uptime on Mindblast to 90% uptime was theory crafted as a 5% DPS boost, if I recall correctly. The thread seems to have been removed. The numbers may be slightly off.

Long story short, stack hit rating as close to 289 as you can, then work on haste to ~300ish (cast 2 mindflays between MB) then stack crit>haste rating, while maintaining the highest possible spell power you can.

Unfortunately we don't have Glyph of Shadow procing off DoTs, ergo, we're stuck with an average 35-42% raid buffed chance to receive our 10 second SP buff... assuming MB crits are at a 38% chance to crit (pretty average on my collection of logs of spriests) and 7.5 seconds per MB (ideally)...

I'm not 100% sure WWS parses [Glyph of Shadow] procs correctly (80-90% uptime seems ridiculously high based on the constraints unless it isn't working as intended), but, if the number, on my napkin math is more around (7.5/.38) ~ 20 seconds between MB crits, with 10 second buffs... is about a 50% uptime, which feels much more in line with my experience (and buff counting) than going over WWS (used real sources from WWS/WOL as well...)

edit: no. it doesn't proc of DoTs... per below. per my patch note investigation, yes it was that way on the PTR...

That's not just bad, that's an awful conversion of spirit. our 50 spellpower boost we receive is cut down by half. Yet we're still moderately competitive at current content (not as GC put it where: all classes should be able to put the around the same dps if the skill is there... because we all know that isn't the case)...

The only universally good stat for us is: spellpower and will continue to be so. both haste and crit will undergo diminishing returns (until they give us haste scaling)

If blizzard is so concerned about spirit being useful dps, why isn't gear more tailored to dps to not include spirit? Because that'd exclude the benefits all other cloth classes (sans disc priests to a lesser extent) receive from spirit.

i noticed on my shadow priest a significant improvement on my dps in a quick and easy naxx 25 run last night. I dont recall what my dps was but i was second on the meter. i was kind of baffeled because usually i was happy to be in the top 10. i'm definetly gonna attribute my success to the improvement on the glyph of shadow.

Hey, I tried this on the dummie and get 500 dps more than with too many haste. I did 3150 dps on heroic dummie (yesterday was on 2,7) and the only thing i've done is lower the haste to 350...

I'll try to change a little my gear to fit 300 haste to recover all my dps.

Many thanks man.

Edit: Sorry if this felt like a thanks post. I just was trying to state that +170 haste (from 350 to 420) it's like a 500 dps penalty.

As of 3.2, Glyph of Shadow does NOT proc off dot criticals. This change was nixed before the PTR went live, and the tooltip of the glyph changed to exclude Damage over Time Critical Strikes.

Main, I believe you misunderstood me. Haste isn't bad, not at all. It's still a net benefit. Your DPS wont go up with less haste. It will increase with more Crit versus Haste when you can do two mind flays between Mind blast cool downs.

I suspect the reason your DPS increased was you were trying to clip MF, which is pretty hard to do and even with the best latency it's still hit or miss. It takes lots of practice. I don't do it, myself. It's a lot of micromanagement for a small DPS gain.

The difference is explained by Mind Flay. Crits will proc Glyph of Shadow.

That doesn't mean that a closed form solution won't be possible.
In fact, I see no reason not to have a closed form solution.
It will most likely be quite long and ugly to write and read, but it will exist for any given fight duration : given some spell priority, we can compute the miss/hit sequence, and the spell sequence, as well as the probability of such a sequence, and we have there everything, including the probability to spend your whole fight trying to apply (and failing to do it) SW:P and VT. That would be awfully painful to compute by hand, I know. But 1/ it does exist, as a closed form solution ; 2/ It can be quite easily done automatically with a program.

If (as it is in most cases) we are interested only about the mean dps, it's enough then to sum all damage done by every sequence, weighted by the sequence probability. Once again, that's perfectly (and easily) done by computer programming : programs can deal with formal probabilities, not only simulate a "likely" sequence.

That was for the "long and dummy" way. Most likely, clever shifting of events and expetation taking would be enough to get a really quicker result.

I was thinking about it again.
In fact, and unless I'm mistaken, things would be (at least for a first approach) quite better than one could expect.
If you don't allow MF clipping, then all the spells "cast times" are multiple of the GCD. It means that we can model it with discrete time, "counting" the GCD, and use discrete Markov chain in discrete time.
It means also that the "state" of the Markov chains is quite limitated : it's basically the product of all "cd's", ie a (10(VT) X 12 (SW:P) X 6 (MB) X 16 (DP) ) space. That's quite easy to deal with ;-)
That's obviously without taking haste into account : haste would be a lot more tricky, because it would kill this fine correspondance between GCD (and cast time) and cooldowns. But anyhow, the interest of hit shall not depend on haste ;-).

The importance of hit ABSOLUTELY depends on haste. What might not be important is the relative importance of hit with respect to crit, haste, and spell damage at different values of haste, but I am not even convinced the latter is obviously true. Haste obviously 'scales negatively' with more haste, in a % sense, but even excluding haste as a stat, I am not convinced that hit scales the same exact way relative to haste as damage.

You cannot discount latency (and variable latency) and things like button press speed etc. from your calculations. It does have an effect and it does mean that spells do not take an integer number of GCDs to cast.

Even with fixed haste one run might very well have a different spell sequence than a second run due to variable latency.

Without proper modeling of this, both closed solutions and a numerical solution are not going to be accurate.

Althor, I have been meaning to ask this, but since you handle the simcraft shadowpriest module, do you have a good place where one can find all the coefficients and the way they add/multiply for shadowpriest DPS? There are a few really wierd quirks with how some coefficients add/some multiply, and some coefficients are a pain to dig up - I wanted to write a really simple closed formula to have an easy way to derive parameters and see how much they differ from numerical calculations.

I could update that in the OP too.

I probably should do up a post at some point detailing the exact mechanics for each spell. Currently it's all in the sc_priest.cpp (with the mechanics of Replenishment in another C++ file) file as code.

With every major (and sometimes minor PTR) patch I go through each shadow spell and test on the dummies or other mobs to see if the numbers match the previous model exactly and if not, can I determine a model for it.

So I've got the coefficients for the various spells, and the quirks like how Improved Devouring Plague doesn't benefit from the 2piece T8 bonus nor from +spellpower from spirit via Twisted Faith or Glyph of Shadow.

In the Shadow Priest module at least mana costs (from what I recall) correctly floor() themselves appropriately with talents.

One thing the Shadow Priest module (and SimCraft as a whole) doesn't model 100% correctly however is latency and player error.
The current model has gcd_lag, queue_lag and channel_lag (along with std. dev variables for each) and these variables are used (in combination with rng) to determine how much latency to add to the end of the previous spell before the new one is cast. We also have some approximiate form of player skill which allows for the last tick of a DoT or channeled spell to be clipped or the best spell to not be chosen. And we also have a variable for reaction time.

I'm currently looking into improving our latency and player skill modeling to maybe more accurately represent how things work in game.

However in terms of Priest specific coefficients and quirks...as I've said, at the moment there's really only the code itself "documenting" my discoveries.

The 2 piece T8 bonus is added to the DoT part of Devouring Plague:

From line 859 of sc_priest.cpp:

However it's not there for devouring_plague_burst_t.

From line 785 in sc_priest.cpp:

And yeah, I've been /bugging and forum posting the bug on every PTR build since 3.1.

I've read alot of good material in this thread, but to me it seems slightly outdated. I have yet to read anything that relates to t9+ and our new set bonuses. I've been personally working on new rotations, trying to find the best one to fit the (2) set bonus, but also trying to work it out to minimalize the loss of dps from GCD's. Has anyone put any theorycrafting together in regards to this?

ATM the rotation I'm experimenting with is this:

MB-DP-VT-SW:P-VE-MB-MF-MF-MB-MF-MF-MB-DP-VT-MF-MB (rinse and repeat)

With the 6 sec additional time added to VT if cast in this sequence, VT and DP will always be on top of each other for CD's, and so far I've never had to clip any spells or cut my DoT's short to accommodate the casting of another spell. The one downside is that SW:P will be missing 2% dmg, but tbh, it's seems quite negligable considering it's our lowest dps spell. I choose to cast ve in between SW:P and MB rather than at the beginning because it fits quite nicely in the gcd for this rotation, where the only other options are a clipped flay and SW:D.

Any advice would be appreciated, as would any theorycrafting on the new tier set bonuses.

The bonus due to set bonuses was calculated by simcraft and is on shadowpriest.com.

The rotation does not change that much at all - VT > all, MB and DP are roughly equal, MF is filler, use SW: D when you have one GCD and nothing else to cast.

Our tier 9 is itemized horrendously, and has a ton of horrible spirit and haste, but item level trumps all.

I am quite convinced that you could write a simple script to calculate shadowpriest BiS by:

- Maximise itemlevel
- Do not overcap hit
- Minimise spirit
- Maintain set bonus

Spirit on our gear is such a waste of itemisation that a piece with crit/haste/spell damage is almost always better than an item one tier above it with crit or haste/spell damage/spirit.

Yeah I was thinking the same thing, Generally when itemizing my gear, I look for items with the most crit/spellpower available after hit and haste caps of course, but I also run into the trouble of what and where. I do want to get my 4 set t9 bonus to have a look at what that 5% crit to MF brings to the table, but as I've already picked up the two most transferable items, gloves and shoulders, I'm trying to figure out where to go from here. I was almost dead set on the pants, but last night I picked off the hit/crit pants from Anub, which is far better itemization than the hit/haste tier. However, my only other two choices then are the helm and chest, which are for lack of a better phrase, absolutely terrible.

I'm already rearranging gear to keep my hit cap as close to min as possible everytime I switch in a new piece, so to know that the two items I'm looking at have a crapload of hit and nothing else to bring to the table, I'm seriously considering whether or not it's worth it to even bother shooting for the 4 set bonus. As it stands, the chest will fall comparable to the chest already equipped, but the only other place that I can pull hit to make room is my neck, everything else is pretty integrated.

If the 4 set bonus is worth it's salt, should I just take points out of my talent tree and reallocate. Was never a big fan of this because there really isn't a useful place to put them other than 2 pts in imp ve, and maybe 1 pt in IF again....maybe as a push, getting silence or phorror for situational fights like the factions champs encounter.

In the end I don't know, I can figure most of what I need to know through testing and trial and error, but when it comes to the shear numbers, I'll never really be able to figure those out, so I come to you for help.

It's true that ilvl trumps all for holy, but not for shadow. Some of our T8.25 pieces (ilvl 226) are better than the T9.25 pieces (ilvl 245). For example, compare [Conqueror's Circlet of Sanctification] with [Zabra's Circlet of Triumph], assuming Runed Cardinal Ruby in one and Purified Dreadstone in the other.

The T8 piece gains 70 crit and 68 haste.
The T9 piece gains 19 int, 19 stam, 84 spirit, 82 hit, 16 spell power

The T8 piece's 70 crit and 68 haste are worth roughly 90 spell power (.65 value).
The T9 piece's 19 int and 84 spirit are worth 21 spell power (.2 value).

So the final choices are equivalent to:

T8: +90 spell power
T9: +37 spell power, +82 hit

Would you rather have 53 spell power or 82 hit? That depends an awful lot on whether you are hit capped. Also remember that the T8 set bonuses are much better than the T9. When you can pick the 254 ilvl helm instead of the 245 though, it's a clear upgrade over the 226 t8 helm.

That said, remember that hit rating is not amazing, but it's far better than spirit. If your choices are between a piece with hit but no spirit and a piece with spirit but no hit, you should take the hit piece, then find a way to drop hit from other gear slots.

Someone in Dalaran once advised me that for AoE I should stack crit; but wasn't certain and I couldn't find anything here - so I tried to figure it out myself and post it here for the benefit of others.

The nice part of mind-sear is that it is possible to figure out the maximum damage, the dps should on average be proportional to the raid-buffed stats (assuming 3 points in misery):

(220+0.286*1.15*spellpower)*(1+0.5*critChance%/100)*(1+haste%/100)

Or slightly better for crit due to increases in crit-damage, but not much since mind-Sear is not influenced by crit-bonus increases of Shadow Power or Shadowform.

So the relative benefit of each term is (=derivative of the logarithm):

Spellpower: 1.15*0.286/(220+0.286*1.15*spellpower)
Crit-rating: 0.5/(1+0.5*critChance%/100)/4591
Haste-rating: 1/(1+haste%/100)/3279

Basically each stat has separate diminishing returns, and relative values for 2500 spellpower and 20% crit/haste: spellpower 1, haste 0.8, crit 0.3 - or in other words don't trust random people in Dalaran. (And since spell-power is still the best stat it will mostly be the same as the normal gear.)