Demonology in Cataclysm (4.3 Release) - Page 10

slay12288 · 04/26/11, 3:42 PM

Originally Posted by Shermyth

I notice you say mastery is greater than haste in the original post. Yet right after, in the reforging section, I see you recommending reforging mastery to haste. Am I missing something?

Before 4.1, Haste > Mastery until the heroic raid levels, where it became Mastery > Haste.

Either the OP had not updated the reforging section, or it is like it is in the reforging section because what matters more is your current haste level. If you're close enough to the next haste threshold to gain another dot tick, then I believe it would be more beneficial to reforge mastery to haste in order to gain that extra tick.

On the otherhand, if you're stuck in the middle two haste thresholds, reforging haste to mastery would be more beneficial for the demon buff, and not significantly lowering cast times.

But I'm going to go under the assumption that since this patch JUST came out, there may be a typo error or two, or a copy/paste messup (assuming the OP had been preparing a 4.1 update).

Originally Posted by suriah387

Will I do more damage using my Felhunter during a single-target fight, even if I have my Felguard glyphed at the time, instead of Glyph of Incinerate?
My question is for the occasions I am unable to replace glyphs for a fight, whether it is better to simply use my Felguard in spite of the buffs to the Felhunter.
Thanks, Suri

Is there ever a time that you don't have 10s to swap one glyph? If you mean heroic dungeons, then yea, people want to be in and out of those so they move fast. However, at this point, I don't feel that it matters that much (nor should you waste the gold) on swapping one glyph for a silly heroic dungeon. For raids though, there should be plenty of time to swap glyphs, like during trash.

(I did not answer the first question since I do not know, still learning these changes myself. Merely wanted to point out that when it matters, there will be time to swap glyphs)

revulva · 04/28/11, 1:46 PM

I was working on an update for the default demonology stat weights on the Mr. Robot site, and I am noticing some interesting results from SimC.

First: At the 359 gear level, Haste is coming out as clearly superior to Mastery. The scale factors come out to something around Haste - 1.5, Mastery - 1.25. I tried a few different sets and confirmed that the simulated DPS backs this up.

At the 372 gear level, mastery is being weighted higher than haste... but sets with haste perform substantially better. Here are the profiles I tested:
SimC Default BiS 372 PTR - 28180 DPS
Set 1 (using SimC Calculated stat weights) - 27659 DPS
Set 2 (favoring Haste) - 28522 DPS

This is a substantial DPS increase... and SimC is STILL indicating that adding mastery at this point would be more beneficial... I made a plot of haste vs. mastery and it showed mastery as providing more DPS over the whole range of the graph. I tried dropping a little haste for mastery in this set:
Set 3 (drop a little haste for mastery) - 28504 DPS

Still no increase in DPS. And, the lower you go in item level, the more and more haste takes over as superior. SimC is the tool of choice for analyzing warlock gear... and right now it is showing haste as the clearly superior stat to mastery for this spec.

VoidStar · 04/28/11, 8:39 PM

You can't use the stat weights in the manner you describe: the stat weights for Haste and Mastery Rating (and all other stats except Hit Rating) represent the value in DPS of adding a unit of that stat and (this is where you went wrong) says nothing about the impact of removing any amount.

Fundamentally it's likely that the reason you're getting strange results is that you're using the numbers in an unintended and inappropriate manner.

Zimeron · 04/29/11, 2:02 PM

Originally Posted by revulva

This is a substantial DPS increase... and SimC is STILL indicating that adding mastery at this point would be more beneficial... I made a plot of haste vs. mastery and it showed mastery as providing more DPS over the whole range of the graph. I tried dropping a little haste for mastery in this set:
Set 3 (drop a little haste for mastery) - 28504 DPS

What type of plot did you run? We added the option for reforge plotting for this very scenario. Like the previous posted said, if you're plotting stat weights, it's not counting the trade off, where as the reforge plot will.

Also, please post your results, either the HTML or SimC profiles (with options included), so we can reproduce your results. The graphs themselves for the plotting would also be useful to see.

revulva · 05/04/11, 1:59 PM

Here are the settings I use for testing:
Patchwerk Style Fight, 450 sec, 20% variation in length. I use the action priority that loads by default:

actions=flask,type=draconic_mind
actions+=/food,type=seafood_magnifique_feast
actions+=/fel_armor
actions+=/summon_succubus
actions+=/dark_intent
actions+=/snapshot_stats
actions+=/blood_fury
actions+=/volcanic_potion,if=buff.bloodlust.react|!in_combat|target.health_pct<=20
actions+=/metamorphosis
actions+=/immolation,if=buff.metamorphosis.remains>10
actions+=/bane_of_doom,if=!ticking&target.time_to_die>=15&miss_react
actions+=/immolate,if=!ticking&target.time_to_die>=4&miss_react
actions+=/corruption,if=(remains<tick_time|!ticking)&target.time_to_die>=6&miss_react
actions+=/fel_flame,if=buff.tier11_4pc_caster.react
actions+=/shadowflame
actions+=/hand_of_guldan
actions+=/incinerate,if=buff.molten_core.react
actions+=/soulburn
actions+=/soul_fire,if=buff.decimation.react|buff.soulburn.up
actions+=/summon_doomguard
actions+=/life_tap,if=mana_pct<=50&buff.bloodlust.down&buff.metamorphosis.down&buff.demon_soul_succubus.down
actions+=/demon_soul
actions+=/shadow_bolt
actions+=/life_tap,moving=1,if=mana_pct<80&mana_pct<target.health_pct
actions+=/fel_flame,moving=1
actions+=/life_tap

First I tested this set (I imported from chardev, since the set embedded in SimC has an error):
chardev 8 - WoW Cataclysm 28108 DPS, mastery 1.37, haste 1.03

This tells me that if I add up to 300 mastery, and leave everything else the same, I will get more DPS than if I add 300 haste, and leave everything else the same. I understand that.

The question I am trying to answer is: what set of stat weights will lead us to the BiS t11 heroic set of gear? There is a set of gear that will do more damage than any other set, and there is a combination of stat weights such that the highest total EP value is obtained with that specific set of gear. That set favors haste over mastery.

Here is a set that is close to what I think the BiS is:
Demonology Warlock - Mr. Robot - World of Warcraft (stat weights are at bottom of page.)
chardev 8 - WoW Cataclysm 28613 DPS, mastery 1.34, haste 1.28

There is something going on that is more complicated than what the stat weights tell us. I tried one of these reforge plots, since it could theoretically tell us something about the options available on the gear. It won't tell us the whole story, though, since gems and enchants need to be included in the analysis. I will be honest that I am unsure how to read the plot, and I don't have anywhere to quickly host an html page. I tried to plot haste reforge options and mastery reforge options... but it has somehow combined both of these on one axis. I ran the test against the default profile that I listed first, if anyone cares to try it.

My whole point here is this: Demonology Warlocks do more damage by favoring Haste, based on simulations in SimC. The stat weights being calculated indicate some bias towards Mastery... but that gearing strategy just does not work out to a DPS gain. I don't really care if Mastery or Haste is better, I just noticed a discrepancy between my observed results and the suggestions in this post. Usually these posts are spot-on, so I am just wondering where this disconnect is coming from. I searched through these posts, and haven't seen anyone put up sets of gear that favor Mastery which out-perform sets that favor Haste. I also haven't found any math that would explain it either. The information must have come from somewhere, I'm just wondering where it came from. The results are showing that, in this case, we cannot simply rely on the stat weights calculated by SimC to determine the correct stat priority.

CaseyTheRetard · 05/04/11, 5:24 PM

Originally Posted by revulva

The question I am trying to answer is: what set of stat weights will lead us to the BiS t11 heroic set of gear? There is a set of gear that will do more damage than any other set, and there is a combination of stat weights such that the highest total EP value is obtained with that specific set of gear.

I think we are all with you on "There is a set of gear that will do more damage than any other set", but it is not necessarily that case that "there is a combination of stat weights such that the highest total EP value is obtained with that specific set of gear." The mapping from stats to DPS is non-linear. A set of stat weights is a linear approximation and therefore insufficiently general to describe that mapping over the entire space of possible stats. As you said, "There is something going on that is more complicated than what the stat weights tell us."

Say that DPS was really simple, and you could calculate it as DPS = c * (1 + haste) * (1 + crit) for some constant c. What set of stat weights will lead to optimal gear choices? Say you are starting from 0 haste and crit and have enough stat points to get 100% of a single stat or 50% of each. 100% haste or crit will give you 2c DPS, while 50% of each is 2.25c DPS. In fact, the ideal gearing strategy in this scenario is to add more to the stat of which you currently have the least. Optimal gear will be as close as possible to equal haste and crit. Any set of stat weights you choose will either have (a) haste > crit, user tries to maximize haste, (b) haste < crit, user maximizes crit, or (c) haste = crit, user decides it simply doesn't matter what stat he chooses. No possible set of stat weights will calculate BiS gear or lead to optimal piece-by-piece gear choices unless by accident.

Given that WoW has more than 2 stats and significantly more complicating factors in calculating DPS, any set of stat weights you choose will be no more applicable than in this simplified scenario.

revulva · 05/05/11, 1:32 AM

Theoretically nothing you say is incorrect. In a practical sense, you can use stat weights to get within an extremely small margin of error. Take that set I just posted above. Can you find a set that is better than that, according to SimC? I used stat weights to find it, along with an automated optimization tool. You will be hard-pressed to find anything that does more than 0.5% more damage (150 DPS) than that set - and I would be willing to bet there isn't a set that even does that much more. On top of that, do we really know that SimC models the game so perfectly that 100 simulated DPS will translate to a noticeable gain in-game? It's a great tool, but it is still a model made by humans based largely on empirical evidence... there IS some margin of error. I'm not trying to argue that your mathematical theory is wrong... I just argue that the significance of the effect is much smaller than you make it out to be.

Still... I really would like to know where the conclusion was drawn that Mastery is better than Haste for Demonology Warlocks, and how that conclusion was reached. The DPS data still doesn't back it up - I've continued to run more tests just to make sure I'm not crazy. Am I the only one seeing this?

CaseyTheRetard · 05/06/11, 12:02 PM

Originally Posted by revulva

Theoretically nothing you say is incorrect. In a practical sense, you can use stat weights to get within an extremely small margin of error. Take that set I just posted above. Can you find a set that is better than that, according to SimC? I used stat weights to find it, along with an automated optimization tool. You will be hard-pressed to find anything that does more than 0.5% more damage (150 DPS) than that set - and I would be willing to bet there isn't a set that even does that much more. On top of that, do we really know that SimC models the game so perfectly that 100 simulated DPS will translate to a noticeable gain in-game? It's a great tool, but it is still a model made by humans based largely on empirical evidence... there IS some margin of error. I'm not trying to argue that your mathematical theory is wrong... I just argue that the significance of the effect is much smaller than you make it out to be.

Still... I really would like to know where the conclusion was drawn that Mastery is better than Haste for Demonology Warlocks, and how that conclusion was reached. The DPS data still doesn't back it up - I've continued to run more tests just to make sure I'm not crazy. Am I the only one seeing this?

It's ironic that in the same post you both call the accuracy of Simulationcraft into question and claim without support that "you can use stat weights to get within an extremely small margin of error." In any case, I think we've beaten the issue far enough into the ground. You are welcome to chop down trees with a screwdriver, but don't be surprised if (a) you get strange results sometimes, and (b) people tell you it's a bad idea to chop down trees with a screwdriver when you complain about those strange results.

It seems likely that the assertion in the OP that Mastery is better than Haste is the result of someone else misinterpreting stat weights. It is more important that EJ spec threads be timely than perfectly accurate. Presumably the OP was updated in response to someone running the BiS gearset through Simulationcraft with 4.10 mechanics and noticing that the Mastery stat weight was higher than the Haste stat weight.

revulva · 05/07/11, 1:12 PM

Originally Posted by CaseyTheRetard

It's ironic that in the same post you both call the accuracy of Simulationcraft into question and claim without support that "you can use stat weights to get within an extremely small margin of error." In any case, I think we've beaten the issue far enough into the ground. You are welcome to chop down trees with a screwdriver, but don't be surprised if (a) you get strange results sometimes, and (b) people tell you it's a bad idea to chop down trees with a screwdriver when you complain about those strange results.

It seems likely that the assertion in the OP that Mastery is better than Haste is the result of someone else misinterpreting stat weights. It is more important that EJ spec threads be timely than perfectly accurate. Presumably the OP was updated in response to someone running the BiS gearset through Simulationcraft with 4.10 mechanics and noticing that the Mastery stat weight was higher than the Haste stat weight.

Well, I think it is realistic to acknowledge that SimC has some margin of error. From all indications and comparison to in-game data, we believe that margin of error is very low.

Using stat weights to determine "BiS" gear sets, and using SimC as my measure for how good those sets are, I have shown that you can use stat weights to get within an extremely small margin of error of what SimC indicates as the "best" set of gear.

If mastery is weighted more because of stacking buffs, procs, and trinkets not taken into account when you sim... how could the sim be calculating that higher weight? And then following that weighting results in a DPS drop?

Something is not adding up right now: SimC keeps calculating stat weights that indicate Mastery is worth more, point for point, than haste. But, sets of gear that favor Haste consistently out-perform sets of gear with Mastery, according to the same simulator. Either the weights are right, or the DPS outputs are right - but they can't both be right. This is all I am pointing out, in a nutshell. I was hoping someone could point me to some sound reasoning for believing SimC's stat weight calculations over SimC's DPS output, because, it seems to me that the DPS outputs would be a more reliable source of information in a simulator.

• Meaning · 05/08/11, 4:18 PM

Note: I just shit-canned the whole movement speed on boots discussion. Two things:

1) It pretty much is personal preference these days, because there are so many movement options. And if you have good situational awareness, it really, really is a toss up.

2) I have no idea why somebody randomly replied to a 5 month-old post, and in this thread no less, instead of the simple questions/simple answers thread.

Now back to your regularly scheduled Simcraft discussion.

Vakki · 05/10/11, 12:03 AM

The biggest problem with Felhunter is your positioning. If you can stay in meleerange, you have the advantage to use shadowflame on cd. If not, the Felhunter loses a lot of his dps potential.

Succubus seems to be a little weird atm. In Nefarian Fight i sent her to Onyxia and after i switched to Nefarian, she attacked him from 50+ yards range and i had no chance to use whiplash ability.

I also wanted to say, that the BIS 372 demonology from simulationcraft has a lot of mistakes. Ring with Mastery and Haste is much better, 10 Haste Socketbonus is not worth to sacrifice 20 INT for it and one Trinket was reforged to Haste even without Hitcap. Another big hit in the eye is the priority list. Every demospecced lock should start with felguard and swap to another pet very soon in the fight. I posted that prio list and my chardev in the other thread and the result was much more dps than the official BIS sims.

What do you guys believe is best pet for fights with huge HASTE buffs like Cho'gall and Sinestra. I thought to use DS(Imp) in these fights, because you cast much more Destrospells when DS is up, but I'm not 100% sure. Theralion's Mirror seems to be also a great improvement in these two bossfights. You have Meta ready for every Trinketprocc. It really hurts, that the Masteryprocc is no on-use-effect.

Meegosh · 05/10/11, 1:56 PM

I agree with the Felhunter/Succubus dilemma.

My early 4.1 tests shew that, indeed, the felhunter DPS was over the succubus's and indeed, you gain one glyph. But that glyph, whichever you choose to use depending on the fight, is not that important (20-30k-ish maximum over the fight), and the DPS increase of the Felhunter is not that high.

I find that there are 2 situations where the succubus is superior :

1) Encounters where you can't go,for any reason, at melee range

2) Encounters where your target is flying

(There is also the problem of heavy switching encounters where the succubus has the advantage with its range, but the felhunter in most cases remains superior)

I have been convinced of this since the release of the 4.1 patch, but never took the time to test it in a raid situation, but I just stumbled on this, reviewing some WoLs :

World of Logs - Real Time Raid Analysis
World of Logs - Real Time Raid Analysis

Both of the fights have the exact same lenght, he is using Felhunter and I am using succubus and the succubus comes on top.

Edit : I just realized that one glyph I never really considered is actually better than the glyphs of Incinerate/Felguard : the Glyph of Corruption. The following calculations are based on a haste rating of 1993.

I)Glyph of Corruption vs Glyph of Incinerate

First off, immolate and corruption should tick roughly at the same rate so for a given fight, you should have roughly the same number of ticks. Let's take a 150 ticks fight.
Shadow Trance (Glyph of Corruption proc) should proc 150*4/100 = 6 times. On each proc, tou gain (casting time - GCD) of shadow bolt casting time hence you gain 2 - 1.2 = 0.8 seconds. Thus, you gain 0.8/2 = 40% of one shadow bolt damage, 7k. Damage gain = 7*6 = 42k (considering an average of 17.5k shadowbolts).
Molten Core should proc 150*6/100 = 9 times. On each proc (in the best case scenario) you gain 3 incinerates so you gain 9*3*(0.05)*19000 = 25650. Damage gain = 25k6 (considering an average of 19k incinerates).

II)Glyph of Corruption vs Glyph of Felguard

I don't really think that the glyph of corruption is superior to the glyph of felguard, but that's only in a really lenghy AoE fight, not a fight with some limited AoE phases. The only encounters I can think of that justify the use of the glyph of felguard are Magmaw (and only if you DPS the parasites), Maloriak (on heroic mode only) and Sinestra.

Also, there is one other benefit (which may not really be a DPS increase) : you gain an instant cast over a lenghy cast, allowing you to move without any loss for a bit (there are always these moments when you don't really HAVE to move but you would like to, and you're stuck casting fillers, with a little bit of luck, it can help - like the 4p proc -).

Draust · 05/19/11, 7:24 PM

Sorry to break up your conversation, but I have a question regarding Improved Immolate and Improved Corruption as viable talents.

On Spell power coefficient - WoWWiki - Your guide to the World of Warcraft it currently states that Corruption has a 120% benefit to SP, and Immolate has a 120% total SP benefit, but only a 100% benefit for the DoT portion of the spell.

Since HoG refreshes Immolate, meaning we usually only have to cast it once, wouldn't Imp Corruption be a better choice? Or is the base damage for Immolate that much greater than Corruption, that Imp Immolate is the way to go?

Thanks ahead of time for an answer.

Flamingcloud · 05/19/11, 8:58 PM

Originally Posted by Draust

Sorry to break up your conversation, but I have a question regarding Improved Immolate and Improved Corruption as viable talents.

On Spell power coefficient - WoWWiki - Your guide to the World of Warcraft it currently states that Corruption has a 120% benefit to SP, and Immolate has a 120% total SP benefit, but only a 100% benefit for the DoT portion of the spell.

Since HoG refreshes Immolate, meaning we usually only have to cast it once, wouldn't Imp Corruption be a better choice? Or is the base damage for Immolate that much greater than Corruption, that Imp Immolate is the way to go?

Thanks ahead of time for an answer.

Improved Immolate gives +10% per rank and Improved Corruption gives +4%, so corruption would have to be 150% better for the talent to be equivalent.

Morninglory · 05/20/11, 8:55 AM

Originally Posted by Meegosh

I)Glyph of Corruption vs Glyph of Incinerate
.....

I don't really think that the glyph of corruption is superior to the glyph of felguard, but that's only in a really lenghy AoE fight, not a fight with some limited AoE phases. The only encounters I can think of that justify the use of the glyph of felguard are Magmaw (and only if you DPS the parasites), Maloriak (on heroic mode only) and Sinestra.

I was recently thinking the same thing, so very glad to see a post on the subject. However since I'm bringing the SP buff for my guild I'm running demo on every fight. Which means I also see a use for glyph of felguard on Omno (use felstorm for tox adds, still debatable if your raid takes these out quickly), halfus, and cho'gal (you can send him in where you dare not tread -- bloods).

I have another more detailed question, but it's a separate subject and thus I'll make a separate post.