Jump to content


Photo

Shadow Priest -- Cataclysm


  • This topic is locked This topic is locked
375 replies to this topic

#21 Inu

Inu

    Glass Joe

  • Members
  • 23 posts

Posted 04 January 2011 - 11:00 PM

Do you mind sharing why you think the casting rules skew stats weights?, and why you are willing to bet that crit is as good (possibly even better) than haste?

I'm not saying your not right, but I miss some reasoning to support your statements.


Haste is slightly less valuable in fights where you can't refresh dots. If a dot falls off the only extra ticks your going to gain out of that cast is if you break the few "soft-cap" points on Vampiric Touch and Devouring Plague. Assuming you stack haste there is a good chance you only have enough for 1 extra tick.

Also in order to supposedly maintain that haste simcraft value you'll need to refresh at 0.5 seconds remaining. Can you do that everytime? There are very few "straight-burn" fights. Straying from this lowers your mindflay time etc.

#22 alinna

alinna

    Piston Honda

  • Members
  • 193 posts

Posted 04 January 2011 - 11:29 PM

It may be worth noting that SimulationCraft's default action list for a shadow priest doesn't include casting Shadow Word: Death when low on mana. This can result in odd stat weights, especially for lesser geared characters. I typically use this command near the top of the priority list to include this behavior, change the percentage as you see fit: actions+=/shadow_word_death,if=mana_pct<15

#23 Cadfael

Cadfael

    Witch doctors park in gear

  •  Patrons
  • 3,819 posts

Posted 05 January 2011 - 09:21 AM

I am a bit puzzeld as to why SW:D should only be cast when moving (while > 25% HP). Assuming the self-inflicted damage is not relevant for healing (ie. is self-healed by VE relatively fast) and I don't need to move at all for a longer periods (say Maloriak encounter), then SW:D has the highest DPET and DPE values of the possible set of SW:D, MB, MF (and MS).

Of course it's not optimal to cast SW:D such, that you are on cooldown when 25% is reached. But otherwise even when mana is not a problem, it's the highest damage spell per time possible to cast when you don't need to refresh dots.
"Take the universe and grind it down to the finest powder and sieve it through with the finest sieve and then show me one atom of justice, one molecule of mercy. And yet you act as if there were some sort of rightness in the universe by which it may be judged."

- Discworld: Hogfather


#24 tedv

tedv

    Observation: I am awesome

  • Allied Members
  • 1,796 posts

Posted 05 January 2011 - 04:16 PM

I agree on the Mental Agility, but I did take 2 points in Sanctum merely for the 4% spell damage reduction for now. You certainly don't need all three points.

Masochism is very useful when you need to multi-DOT at a minimum...and keep in mind the mana return is only half of the effect. Reducing the damage taken from SW:D by 50% is substantial during execute phases. Get double 18kish crit (for 72k damage total) and you take 72k damage instead of 36k. Even taking 18k instead of 36k (assuming neither crits) can be important on fights with heavy raid damage.


Masochism isn't tied to reduced Shadow Word: Death damage anymore. Here's the talent: Masochism - Spell - World of Warcraft

This effect got moved to Pain and Suffering, which is of course mandatory: Pain and Suffering - Spell - World of Warcraft

#25 Snowy

Snowy

    Mitt Romney?

  • Moderators
  • 10,236 posts

Posted 05 January 2011 - 04:46 PM

This is where it gets tricky, since there's still a lot of things higher on the list than SWD: Refreshing dots, casting Mind Blast if you have an orb when Empowered Shadows is about to expire, and then the big question is, if you DON'T have any orbs and Empowered Shadow is about to drop off, is Mind Flay still a better choice since you need to get an Orb?

#26 tedv

tedv

    Observation: I am awesome

  • Allied Members
  • 1,796 posts

Posted 05 January 2011 - 06:03 PM

This is where it gets tricky, since there's still a lot of things higher on the list than SWD: Refreshing dots, casting Mind Blast if you have an orb when Empowered Shadows is about to expire, and then the big question is, if you DON'T have any orbs and Empowered Shadow is about to drop off, is Mind Flay still a better choice since you need to get an Orb?


Plus Mind Flay also reduces your shadowfiend cooldown. When you have no orbs and dots are refreshed, what else can you do but spam mind flay? You could mind blast to keep up replenishment, but I doubt that's worth it. Spamming Mind Flay will give you an orb fast enough.

I do think it's a legitimate question of whether you want to let dots expire and just spam mind flay until you get an orb, or refresh your low quality dots without empowered shadow. The answer certainly depends on how much mastery you have. The more you have, the larger the DoT damage differential from having the buff, and the sooner you want the buff up.

You'd think that extra haste would also encourage orb fishing because you have more orb proc chances per second from mind flay (since it casts faster). But you're also losing proportionate DoT damage from the fact that your hasted DoTs aren't on the target.

Anecdotally, when orb fishing I've seen that you normally get the orb pretty quickly but there are some long tail cases where you're fishing for a very long time without DoTs up (except obviously shadow word pain). This suggests that orb fishing has a high chance of causing a minor DPS increase but a low chance of causing a massive DPS loss. The optimal cast pattern could depend on your tolerance for damage variance in a particular section of the fight.

For example, suppose you can choose one of these two damage profiles:
Orb fishing: 90% chance of 16k DPS, 10% chance of 10k DPS (15.4k avg)
Refresh DoTs: 100% chance of 15k DPS

Then suppose you have a burn phase where you need to break a damage shield to interrupt a cast, and you need to sustain 14k DPS to do so. Choosing the Orb fishing option is ludicrous because you have a 10% chance of wiping due to low damage. But in a fight without a short term DPS threshold, the outliers should be mitigated by a sufficiently large sample size to get an overall DPS increase.

Again, that's assuming Orb fishing is even a DPS gain, which we'll have to simulate to determine.

On another note, I have a sneaking suspicion that because there are so many internal tensions in the rotation, the question of how you prioritize the lower priority options doesn't matter much. The high priority things are obviously "keep up your dots", "keep up empowered shadow", and "shadow word: death in execute range". It might be that "Mind Blast when you have 3 orbs" isn't a huge deal, given the opportunity cost of a Mind Flay and all its associated benefits. It's not like blast hits THAT hard.

#27 Cadfael

Cadfael

    Witch doctors park in gear

  •  Patrons
  • 3,819 posts

Posted 05 January 2011 - 06:28 PM

Yeah it's a bit tricky, I fully agree. I just am not sure if a blanket "keep SW:D ready for movement and execute range, never cast it otherwise" is good advice, as SW:D keeps being largest DPET outside execution range and thus worthy of being casted. Obviously not at the expense of DoT refreshing or ES refreshing.

The reason I asked is that I found myself unable to properly monitor the critical variables with EventHorizon, which I used for the longest time for all my needs. So I wrote a little addon which looks like the old Daemon debuff monitor of TBC times which displays in a small frame those three things: HP of target, color coded to signal SW:D on cooldown, execute range and when outside execute range it is save to cast too (ie. cooldown will not reach into execute range), an ES indicator, showing when it's on, when it's on and I could refresh it with 1 or more orbs and MB off cooldown and when it's off and/or I can't cast MB due to cooldown or no orbs and finally a simple orb counter.

With these three indicators on a very small spot I feel I can better react to the new conditions and I do cast SW:D quite a lot, wheter I need the mana or not since those 10K+ non crit hits for 1 GCD aren't wasted. You also get to notice those bad luck strings that tedv mentioned very well when you start to stare on the ES indicator to finally light up and cursing when you haven't gotten an orb after 10 seconds into the fight.
"Take the universe and grind it down to the finest powder and sieve it through with the finest sieve and then show me one atom of justice, one molecule of mercy. And yet you act as if there were some sort of rightness in the universe by which it may be judged."

- Discworld: Hogfather


#28 Nimiks

Nimiks

    Von Kaiser

  • Members
  • 54 posts

Posted 05 January 2011 - 06:28 PM

About hit capping and aside all theory, in practice I found it very annoying not to be hitcapped on Cho'gall fpr exemple, where all the adds are tagged as bosses. If you need to assist on the oozes (blood of the old god or whatever), mind spike is really efficient but you need the hit cap (or near, 15%~should be ok). You also loose a lot of mana when multidotting with low hit and bad luck (and a lot of fights allow multidoting atm) and can be forced to SW:D or keep your SF for mana instead of pure dps.
So all in all, I would say it's not really a matter of opinion, I tried to raid with between 5 and 10% hit on the first raids, because i couldn't reach the hitcap without sacrifying too much crit/haste (or trinkets, like the excellent jewelcrafting one), and if it showed good DPS results, it was really annoying because of the reasons I stated above. For now I run something like 15% and lost quite a lot of crit, but i've much less to care if my fucking VT missed for the third time on theralion or not, and can use mind spike properly.

#29 Balkoth

Balkoth

    Piston Honda

  • Members
  • 244 posts

Posted 05 January 2011 - 10:32 PM

This effect got moved to Pain and Suffering, which is of course mandatory: Pain and Suffering - Spell - World of Warcraft


Oops.

Yeah it's a bit tricky, I fully agree. I just am not sure if a blanket "keep SW:D ready for movement and execute range, never cast it otherwise" is good advice, as SW:D keeps being largest DPET outside execution range and thus worthy of being casted. Obviously not at the expense of DoT refreshing or ES refreshing.


I just went and spammed SW:P, SW:D, MF, and MB only to keep Empowered Shadow (12% for me) up on a dummy for a bit. As far as I'm aware, raid buffs would only skew the results further in MF's favor and these numbers don't need to be precise anyway.

Average MF tick: 4042
Average SW:D hit: 3603

You get 1.5 MF ticks per SW:D, meaning MF would have to tick for about 2400 for SW:D and MF to offer equal DPS contributions. Obviously MF is doing nearly twice that, so SW:D is a DPS loss when cast (as even a single tick of MF would seem to be better) unless you need it for mana. If you need it for mana, you would ideally want to only use it while moving to boot and not use it at all if mana isn't an issue on a fight. So you only use SW:D as much as you need to for mana above 25% and no more.
The brightest light casts the darkest shadow.

#30 Whatev

Whatev

    Von Kaiser

  • Members
  • 77 posts

Posted 06 January 2011 - 03:05 AM

Has anyone had experience dropping Masochism? I find that with proper cooldown management, I don't run out of mana too often. In theory this could let us take Phantasm, or maybe 1 point in Improved Mind Blast (which has the best returns, since it reduces the cooldown by 1 second). Or you could take a full PvP spec complete with Paralysis without any real loss in raid utility.


This is incorrect, Imp MB is still 0.5/1/1.5; you can confirm this on the talent tree in game. Wowhead has some kind of weird error at the moment where they display it as 1/1/2, perhaps due to rounding. You can also see this on Divine Fury where Wowhead states that the cast time reduction is 0.15/0.35/1 where the actual third point still brings it to 0.5.

#31 jjloraine

jjloraine

    Glass Joe

  • Members
  • 5 posts

Posted 06 January 2011 - 04:32 AM

I've been sitting and playing around with a target dummy for a bit now, and I respecced a few times, and I'm not really seeing any significant disparity in DPS between trying to max out the damage of my DoTs and just spamming mind spike x3/mind blast. The only variations I'm seeing are when there are crits and good/bad orb procs, but that'll make things change up either way and I really can't control that. Am I missing something big here about how mind spike plays into good shadow DPS? With the mind melt talent turning mind blast into an instant with a 90% crit chance, the spamming is actually delivering rather high damage.

Edit: Disregard. I did more research and got some add-ons to help me minimize DoT downtime. Surprisingly more difference than I expected in the DPS output. However, I'm guessing mind spike is our weapon of choice for small trash and adds that need nuking? The RNG involved in damage output is also driving me quite batty. Orb procs seem to follow the whims of the anti-priest gods. :(

#32 Elimbras

Elimbras

    Don Flamenco

  • Members
  • 477 posts

Posted 06 January 2011 - 01:38 PM

Anecdotally, when orb fishing I've seen that you normally get the orb pretty quickly but there are some long tail cases where you're fishing for a very long time without DoTs up (except obviously shadow word pain). This suggests that orb fishing has a high chance of causing a minor DPS increase but a low chance of causing a massive DPS loss. The optimal cast pattern could depend on your tolerance for damage variance in a particular section of the fight.


When spamming Mind Flay, the time to the next orb is an exponential random variable. For each three MF ticks, you have 1 SW:P tick. With a proc chance per tick of p, the probability that the orb procs at the Nth MF tick is : (1-p)^(N*4/3 - 1) * p (the 4/3 represents the additional ticks from SW:P. Depending on the "phasing, it could be the floor or the ceil of that value).

The main point is that this distribution is not heavy tailed. The probability to wait a long time decreases quickly.

#33 revulva

revulva

    Piston Honda

  • Members
  • 105 posts

Posted 06 January 2011 - 02:53 PM

About hit capping and aside all theory, in practice I found it very annoying not to be hitcapped on Cho'gall fpr exemple, where all the adds are tagged as bosses. If you need to assist on the oozes (blood of the old god or whatever), mind spike is really efficient but you need the hit cap (or near, 15%~should be ok). You also loose a lot of mana when multidotting with low hit and bad luck (and a lot of fights allow multidoting atm) and can be forced to SW:D or keep your SF for mana instead of pure dps.
So all in all, I would say it's not really a matter of opinion, I tried to raid with between 5 and 10% hit on the first raids, because i couldn't reach the hitcap without sacrifying too much crit/haste (or trinkets, like the excellent jewelcrafting one), and if it showed good DPS results, it was really annoying because of the reasons I stated above. For now I run something like 15% and lost quite a lot of crit, but i've much less to care if my fucking VT missed for the third time on theralion or not, and can use mind spike properly.


I agree.

I ran a whole bunch of simulations on SimC with my own character (ilevel 333 with a couple 346 items) and also with the best in slot 372 gear. I found that for the low-level geared character, hit was actually coming out as weighted higher than haste/crit.

When I loaded up the best in slot 372 gear, the stat weights that resulted were obviously those in the OP. Based on those stat weights, I changed the best in slot gear to what it "should" be, since the gear they have in there by default favors hit to the 17% cap. After equipping what ought to be the best in slot gear, I had only about 9.5% chance to hit, but it did simulate out to 200 more DPS.

So, in summary, you lose 0.8% DPS on an ideal simulation with the best gear available in the game for getting hit capped.

I think it would be sound to say that getting hit capped or very near it (15%+) should still be considered a best practice and priority for gearing shadow priests.

I have a longer thread here:
Shadow Priest Stat Weights

I do note in my thread that I was seeing some anomalous results from SimC. I know it is still in beta, so I'm trying to figure out what could be causing it so I can hopefully help out the developers.

#34 tedv

tedv

    Observation: I am awesome

  • Allied Members
  • 1,796 posts

Posted 06 January 2011 - 02:54 PM

When spamming Mind Flay, the time to the next orb is an exponential random variable. For each three MF ticks, you have 1 SW:P tick. With a proc chance per tick of p, the probability that the orb procs at the Nth MF tick is : (1-p)^(N*4/3 - 1) * p (the 4/3 represents the additional ticks from SW:P. Depending on the "phasing, it could be the floor or the ceil of that value).

The main point is that this distribution is not heavy tailed. The probability to wait a long time decreases quickly.


Sorry I was using long tail in the exact opposite sense of what I meant. I meant that there are a few very rare, very long delay occurrences.

#35 Elimbras

Elimbras

    Don Flamenco

  • Members
  • 477 posts

Posted 06 January 2011 - 03:28 PM

With regard to hit value, it is useful to realize that the more movement / adds / silence a fight includes, the more the value of hit. One of the main reason for the low value for hit is that the penalty for missing dots is "low", because one can recast them (in fact, this is true for any sequence with different priority spells, when high priority spells can be recast). But if the proportion of time that one spends casting high priority (read dot) spells increases, then the value of hit has a lower decrease, up to a point where one cast only high priority spells and hit has a "normal" value.

#36 revulva

revulva

    Piston Honda

  • Members
  • 105 posts

Posted 06 January 2011 - 03:37 PM

I understand the theoretical argument for the low value of hit in a simulated tank and spank environment. That makes sense to me, in theory. I was simply pointing out that, for all practical purposes (0.8% if far less than the RNG in real raid environments), you don't lose any DPS, even in a perfect world, by getting hit capped with the gear currently available.

That is why I (and others) are proposing that it is safe and correct to say:
Shadow Priests: hit is better than haste until you are near the cap (say 15%). And, really, you are not going to see a DPS loss even if you get to the cap.
Team Robot developer.
Follow Mr. Robot on Twitter or Facebook for updates, feature releases, bug fixes.
Mr. Robot is now available on your Android and iPhone

#37 Elerion

Elerion

    Great Tiger

  • Members
  • 861 posts

Posted 06 January 2011 - 03:49 PM

Consider also the human error that occurs when you miss a dot and don't notice it immediately due to the fight mechanics requiring your full attention, which further increases the value of hit.

I think it's safe to say that hit is underrated somewhat by SimCraft relative to realistic scenarios.

My personal gearing philosophy is to prioritize hit marginally above all secondary stats until cap. If I value haste at 0.50PP, I value spirit/hit at ~0.51PP. This mainly affects reforging and some close item decisions (haste/spirit > haste/crit items).

For what it's worth: There's no reason to stop chasing hit as you approach the cap. The value of hit is constant all the way from 0% to 17%.

#38 Snowy

Snowy

    Mitt Romney?

  • Moderators
  • 10,236 posts

Posted 06 January 2011 - 05:14 PM

The way I view it is I prefer to be hit capped, but it's not absolutely essential like it was before Cataclysm. Before, if I was 1% shy of the hit cap, I'd probably wear clearly inferior pieces of gear that had hit on it. It was that important. As I mentioned in the guide though, the hidden variable is how quickly you react to missing a spell. If you're hit capped, it's one less thing you have to worry about watching, and we have enough things to watch already.

#39 Brekk

Brekk

    Piston Honda

  • Members
  • 149 posts

Posted 06 January 2011 - 07:19 PM

This is where it gets tricky, since there's still a lot of things higher on the list than SWD: Refreshing dots, casting Mind Blast if you have an orb when Empowered Shadows is about to expire, and then the big question is, if you DON'T have any orbs and Empowered Shadow is about to drop off, is Mind Flay still a better choice since you need to get an Orb?


Don't forget the remaining duration of your currently active VT and DP. With both of them locking in Empowered Shadows it dropping momentarily is not a large concern, yes you lose it from SW:P, and MF if you're casting it, but as long as it doesn't effect your VT and DP rotation the damage loss can be minimized.

#40 Stormraiser

Stormraiser

    Glass Joe

  • Members
  • 4 posts

Posted 06 January 2011 - 07:51 PM

The way I view it is I prefer to be hit capped, but it's not absolutely essential like it was before Cataclysm. Before, if I was 1% shy of the hit cap, I'd probably wear clearly inferior pieces of gear that had hit on it. It was that important. As I mentioned in the guide though, the hidden variable is how quickly you react to missing a spell. If you're hit capped, it's one less thing you have to worry about watching, and we have enough things to watch already.


I concur. I mentioned on another thread that my personal dummy tests resulted in a very small dps loss when I used my regular, 8.5% hit set vs an "inferior" hitcapped set, and I think that gap would only broaden in an actual raid environment, where movement and overtaxed reaction times come into play.

Not being capped drives me insane, anyway. Missing two flays in a row during a Rohash Storm Shield and watching the raid wipe because of it = murderous rage.

Edit: Does anyone have any hard numbers on the dps loss of glyphing Dispersion instead of Death? I've had the former since xpac release due to mana and survival issues and have only recently (having reached full-ish 359 gear) considered returning to GoSWD; I've always thought that an occasional pre-25% SWD and the glyphed Dispersion (available, though not used for its duration on CD, as a rule) were better dps gains overall than glyphing Death and having to use full Dispersions, save Fiend for mana, etc. Additionally, my limited experience with Cata's hardmodes has shown me that 100% of wipes are "Hey idiots, keep yourselves alive" wipes instead of "We all lived but hit the enrage timer due to a lack of 1% dps" wipes. Do those reasons seem sufficient to consider Dispersion as a sometimes glyph choice?




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users