Neat, that puts the numbers roughly where they should be.

Althor, does simcraft assume perfectly clipped mindflays, and is there a way to make an option so that simcraft clips mindflays, there is a non-trivial chance that it fucks up the clip or clips it half a second later or so?

I am wondering why when I compare simcraft performances of various classes, shadowpriests seem to show a bigger gap between the theoretical numbers and the real numbers than other classes.

I've been wanting to implement "proper" latency modeling and "proper" spell-choosing with the possibility of screwing it up and properly supporting clipping etc. It's not an entirely trivial change however.

There is a "cut_for_mb" option I added a while back for Mind Flay but it's not overly tested.

And yes, I mostly wanted to implement those discussed changes because I believe also that Shadow Priests can more easily perform less than what would be suggested by theory if they play less than perfectly than some other classes.

And it's that belief that makes me think that Shadow Priests being where they are in SimCraft with 3.3.0 *might* not be too bad (from a single target PoV) because we stand to lose more DPS in actual play than other classes.

Is there an option in Simcraft to turn off mindflay clipping?

It doesn't clip by default. In effect, the default SimulationCraft model is to use [nochanneling] macros for all the spells following Mind Flay (including itself) with a default latency of 250ms +/- some delta (which follows a gaussian distribution curve).

Aha, interesting. Out of curiosity, are you placing a hardcap on the STD_DEV of the gaussian to avoid ending up with negative lag? That would be an interesting situation!

It's 2 std. dev's. And yes, afaik you can't get a negative lag.

Poisson is discrete distribution, which would be quite strange (I'm pretty sure there is a continuous distribution which follows the same "slope", but it has another name).
However, we don't really care about the precise distribution : I don't think it has a huge impact (the mean has an impact, and maybee the standart deviation, but the whole distribution should not have a huge impact, unless it is "wierd").
More than that, Gaussian is one of the most legitimate candidates, if you consider that lag is a sum of different independant effects.

In fact, the most legitimate candidate would be some hypo-exponential (aka generalized Erlang, or sum of independant exponential rv with different parameter) distribution. There is significant litterature saying that on Internet link, the delays are a minimal delay (caused by serialization time + transport time on the link), plus a waiting time composed of an atom a 0 and an exponentially distributed density. There are reason to think that it would be the same for software / hardware queueing on servers and client sides. But once again, I don't think it really matters. What matters is the shape : a strictly positive mean, and a exponentially decaying density at both sides of the mean.

From the 10/16 changes to the PTR patch notes:

Enchant Weapon - Black Magic: This enchantment now sometimes increases haste rating for the caster rather than inflicting the caster's target with a damage-over-time effect. It is also now triggered by landing any harmful spell rather than inflicting damage with a spell.

PTR analysis so far has shown that this is a haste increase of 250 for 10 seconds. If the internal cooldown is still 45 seconds and the PP of haste is 1.31, this enchant would be worth roughly 72.7 dps.

Would this now become a viable or even preferred enchant for shadow priests?

Over 63 spell damage? Not really, but:

- SW: P & haste interaction probably is the same and SW: P and crit interaction, inflating the value of temporary haste procs.

- Haste makes your rotation 'messier' assuming you optimized your gear to be at a good cutoff for haste. We can only fit in so many single GCD spells.

Again, as for all complex issues, we can estimate an answer with napkin math, but we will need proper modeling to get a good answer.

which rotation do you perfer? I played a lot of time in Naxx, Ulduar and PDC. But i dont get to the 5k dps, i can do a lot and had a good supportet group but my highest dps in a 25 man raid was 4,2k. and you`re talking about the 5k at the beginning of this post.

And about the mana problems, i had a lot if i´m not in a well supportet group and this by bosses like emalon.

It's worth about 71 haste (35s ICD, 10s duration), and with the current scale factors, that makes it, on average, better than 63 SP, and 81 SP to staff (it's worth roughly 93 SP, probably closer to the mid-80s in a haste-optimized set). Consider also that if you "lock it in" to your SW:P that's 179 more haste contributing to 11% of your dps (same post), you can add 19.69 haste or 26 SP bringing it up to ~119 equivalent SP. Tell me if my math is wrong, but that seems like a pretty significantly powerful enchant, especially for 1h weps and especially if you can time your rotation around refreshing your dots while it's up (10s is a small window of opportunity so your mileage may vary and it will likely need field testing).

When I get around to it, I'd like to do a comparison of the two major haste proc trinkets (Embrace of the Spider and Shard of the Crystal Heart; I don't believe I'm forgetting any EDIT: Elemental Focus Stone, too (though to be honest I don't see it or Shard being useful because of how much hit will likely be on our tier)) and see how they stack up to the current top trinkets depending on whether or not you can lock the haste into SW:P.

Also, I'm not spending an awfully large amount of time on the PTR so if the mechanics behind SW:P "remembering" haste as it does with crit on live change, let me know so I can alter my calculations accordingly.

It's based on what I read on wowhead, so it may change, but it seems more in line with Berserking's uptime.

Also, engineering seems like an exceedingly good choice for 3.3 assuming that the SWP refresh mechanics aren't fixed by the time it launches. Going off those same (slightly too good) scale factors:

Inscription/LW/Alch still 46 PP
BS = 40 Haste = 52.4 PP
JC = 42 Haste = 55.02 PP
Tailoring = 73.8 PP (lightweave best possible scenario) - 30.13 PP (greater speed) = 43.67 (sorry everyone who rerolled tailoring! It'll probably be on par with Inscription/LW/Alch when we get more haste... more numbers are needed)
Engineering is more complicated. Depending on the scale factors when we have more haste, it might be a wash between the two cloak enchants (27 PP and 30.13 PP). Hyperspeed accelerators is 340 haste *12 seconds /60 seconds = 68 Haste (-28 SP) = 61.08, making it BiS baseline. And again with locking it into your SW:P... SW:P would gain 80% more benefit at 100% uptime of the 340 haste, and at 11% of our dps that equals 30 haste or 39 PP. Engineering is worth 100 PP if you stack its haste onto SW:P, more than twice as good as inscription/LW/Alch and almost twice as good as BS/JC. I wouldn't reroll it until the patch goes live because they may change things, but it just looks CRAZY good right now. And let's not forget that nitro boots are easily the BIS foot enchant.

Keep in mind that the more stackable haste benefits you have (black magic, PI, Hyperspeed, lust, speed potion) the lower the likelihood that the stars will align and you'll get a 1 second PW:S. Even if you can convince a disc priest to hold a PI for you until lust, you should be using Accelerators every cooldown... and black magic is just random. You're almost guaranteed to have one or both up at least for a little while during Lust, but you're more likely to be sacrificing some DPS.

I tested Embrace of the Spider on PTR.

TEST 1:

Cast SW:P
Refresh via MF spam (occasional Spider procs)

Number of ticks: 29
Shortest Tick Interval: 2.562s
Longest Tick Interval: 2.750s
Average Tick Interval: 2.647s



TEST 2:

Spam MF until spider procs
Cast SW:P
Refresh via MF spam

Number of ticks: 66
Shortest Tick Interval: 2.250s
Longest Tick Interval: 2.422s
Average Tick Interval: 2.331s


It appears that SW:P "remembers" your haste rating at cast (just as crit works), but is not updated via MF refreshes.

This was already established in this post. It's still possible that this will be changed before the patch goes live, but there's no indication yet that it will.

For what it's worth: EotS = 98 SP + (assuming it procs roughly every 50 seconds with a 45s ICD and) 101 haste = 230.31 PP. At this point it's sitting above IotDS and behind Flare of the Heaven. Now, if you get it to lock in with SW:P, it's another 44.44 static haste or 58 PP, making it 288 PP and sitting in between the Heroic and non-Heroic version of the Unliving trinkets as 2nd BiS (though I'd still take Unliving over it because of the unpredictable nature of haste procs). As a plus, you will almost always have it up for at least part of Bloodlust.

(Ah, scale of fates. I knew I was forgetting one. I'll crunch those numbers later.)

The numbers are useful though, as they raise the question of whether it's properly scaling with haste. 505 haste = 15.4% haste. 2.7 (unhasted period) / 1.154 = 2.294, which is below the average hasted period (although it's within the interval range). What did you use to measure the intervals, out of curiosity? And could you do it again with 0 haste from gear?



Scale numbers; 125 SP + 72 Haste + 39.6 Haste (if locked into SWP) = 271 PP, 219.32 if you don't lock in SWP.
Abyssal rune is 218 (107SP from proc + 84*1.31 Haste)
Eye of the Broodmother is 198 (crit PP increased slightly) and a wash with IotDS

2x Unliving is probably still BiS because of the RNG nature of haste procs, but those of us still rocking IotDS or EotB will be replacing it for Abyssal Rune, Scale, or EotS.

Would like to confirm the 35s cooldown on Black Magic. I hit stopwatch right when I got it and it was up at 36 seconds again.

An additional consequence of the haste locking with casts is the completely screwy effect that this has on haste clickies and troll berserking:

Since haste affects dots now, you can massively increase shadow word pain damage by doing a few things:

On the pull, use troll berserking (20% haste), engineering gloves (?? haste), get ToT (15% damage) and a speed potion (+500 haste rating) and cast shadow word pain. As long as you keep it rolling with haste, your pain dot will get those bonuses massively locked in.

Pretty much, for min/maxing sake, every priest should reroll engineering, since the haste clickie is very very powerful, when you can lock it in, and you either need a disc priest to PI you or you need to a troll. I hope they fix the mechanic, since having trolls be the master PvE race is annoying.

Can someone bring this up on the US forums to see if they will adress this?

I'll make a post as a suggestion and bug report, and sadly I suspect that blizz will just flat-out remove haste scaling from SW:P instead of fixing it because it wasn't as big a deal with crit + %dmg as it will be with haste procs. For the record, we have:

Berserking (20%)
Bloodlust/Heroism (30%)
Speed Potion (15.25%)
Hyperspeed Accelerators (10.36%)
Embrace of the Spider (15.40%)
Scale of Fates (13.17%)
Black Magic (7.62%)
Power Infusion (20%)

131.8% haste, and I'm sure I'm forgetting something with as ubiquitous as haste procs are in WotLK. Factor in 8% from raid buffs and 25% from gear and you'll have 165% haste on your shadow word pain (if the stars align, that's a 1.13s period), and an additional 15% damage from TotT.

And if you'd refer to my earlier post, Engy is the BiS profession even without SWP locking in, but it's roughly 50% better when you can. I'd say as a rule of thumb, haste procs are worth about 1.5x their nominal value (which is already impressive for shadow priests in 3.3) when you can successfully lock them into SW:P.

Berserking/heroism/PI do not stack. The rest stack, but I am not very sure if embrace of spider/scale of fates can proc at the same time ( I suspect they cannot).

One thing that people however are not keeping in mind, is that with ~100% haste, keeping SW: P rolling on multiple targets will require you to weave mindflay ticks like a ninja, since it wears off incredibly quickly.

Another thing that you have not accounted for in your analysis is tricks of the trade from a rogue, AND the possibility of doing a weapon swap with a low spell damage high crit/haste weapon with lots of sockets- something like the staff from twins 10 man heroic.

On the topic of massive haste buffs locking into SW:P on cast, this could quickly lead to infinite mana situations when glyphed for SW:P. Not that shadow has mana issues now mind you, but it kind of goes against the whole idea of mana as a resource system.

Since priest base mana is 3863 at level 80 if I remember correctly, then that's 12.88 mana per second returned with 0 haste.

At 55% haste (30% of that coming from procs/speed potion at fight start and about 25% from gear since I'm assuming heroism won't always be blown at the encounter start), the period of SW:P would be 1.94 seconds if I calculated correctly. This value of haste brings the glyph from 64.4 mp5 to 99.6 mp5.

Considering Meditation for a typical raiding shadow priest is worth somewhere between 150-250 mp5 according to some random samples of PvE shadowpriests I've armoried, I was thinking it might be worth it to free up a point or two (or all) from Meditation and spend those in utility talents in the shadow tree like Improved Shadowform or Silence, as a single shadowfiend is usually enough to offset any dearth of mana over the course of the encounter.

Improved Shadowform is borderline required for raiding. Rather than cutting meditation, the best place to trim is Focused Mind. Check my spec (armory link to the left). It has Silence, Psychic Horror, Improved Shadowform, and 3 minute Shadowfiends. Only drops 1 point in Focused Mind to do it.

They don't stack? That's news to me, but makes sense I guess. I guess the thing about this is it would be mandatory to be troll because it's too difficult to wait until everything stacks and cross your fingers that there's a lust up. I sincerely doubt that EotS and SoF won't be able to stack because there are no other situations I can think of where an internal cooldown procs an external one--maybe the other way around, but if you wait for spider and then click scale you should be good. I'd test it but I own neither trinket.

Currently on the PTRs, Mind Flay resets SW:P to 18 seconds regardless of the haste it was cast with. I don't have time to test this ATM, but I suspect it will either tick until it drops off or tick 6 times and drop off regardless of how much time is left on it.

A weapon swap would be very interesting, though I suspect that if you swap out the weapon you got Black Magic with (ostensibly a 1her since it's so vastly better with the new choice in enchants) then the buff will cancel. If anyone has the time to test any of these 3 situations today on the PTRs (or on Live with the trinkets), it would be much appreciated. I'll get to it myself later this week if nobody else can.

Focused Mind is ~50 mp5 by the last approximation I read, and is frankly nearly worthless except in fights where you're going to be mind searing frequently, as the value is significantly better in that case. Since its reduction doesn't apply to DoTs, it's going to be even worse in 3.3. Imp Spirit Tap on the other hand is going to be much better in 3.3 with spirit being 55% spellpower's value, and honestly I couldn't imagine going without spirit tap on fights with adds, esp. H Anub.

The threads asking for a fix to haste scaling to SW:P are at:
World of Warcraft - English (NA) Forums -> Fix DoT refreshing!
World of Warcraft - English (NA) Forums -> Fix DoT refreshing!

If you have a US account, please post there to keep them visible and get this issue sorted out.

Indeed they do not stack. PI can be active at the same time as heroism, but you'll only receive benefit from the spell cost reduction portion of it.

@ Tedv - You'd be dropping Spirit Tap/IST as well, as far as talents that people often pick up in raiding specs. Not that it isn't a piss poor dps talent for 5 points, but it'd still be another change from what he's currently specced. Plus, it will be considerably less awful next patch!

Berserking stacks multiplicatively with both of the other two, but PI and Heroism do not stack together.