Theck's Simcraft Thread - WoD/6.x

This thread contains a variety of calculations relevant to paladin tanking, primarily focused on threat and damage output. The first 10 posts or so are summaries of the results of these calculations, which you can access via the table of contents below.

This thread is for discussion of analyzing Warlords of Draenor using Simulationcraft. In previous expansions I maintained my own MATLAB code base, but I've retired that in favor of the Simulationcraft suite, which has a number of benefits, not least of which being increased accuracy and detail. For information on the older versions that modeled Mists of Pandaria, Cataclysm or Wrath of the Lich King mechanics, please see the Mists of Pandaria/5.x thread, Cataclysm/4.x thread or WotLK/3.x thread.

The code is hosted on Google Code, and is available for public review/consumption. For more details about the Simulationcraft project, please consult the home page or the documentation.


Table of Contents

• Basic Information
  
  
  
  1. Mechanical Changes
  2. Default Gear Set
• Calculations
  
  
  1. Ability Damage
     
  2. Single-Target Rotation
     
  3. AoE Rotation
     
  4. Talent Comparison
     
  5. Glyph Comparison
     
  6. Gear & Weapon Comparison
• Other Stuff
  
  
  1. Glossary
  2. Simulation Notes
  3. Outstanding Requests - Call to Arms thread

Default Gear Sets

My current plan for WoD is to stick to one gear set per tier. Simulationcraft will obviously have profiles for Normal, Heroic, and Mythic raiding, but the general trends we're looking for don't usually change when you swap from e.g. Normal to Heroic. As such, my plan is to focus on the Heroic gear sets, since those give the best approximation of what you're wearing during Mythic progression.

For our preliminary T17 beta work, I'm using the following gear set, which was compiled earlier in beta. It's not necessarily the same as the T17H profile in SimC, as we may have updated that slightly since this set was created. Once we're done tweaking profiles I may go back and update it and re-run the sims.

T17H_gear.simc:

Code: Select all

head=helmet_of_guiding_light,id=115568,bonus_id=566
neck=vertebrae_protector,id=113900,bonus_id=566/448,enchant=gift_of_haste
shoulders=unstable_slag_shoulderplates,id=113884,bonus_id=566/448
back=charbreath_firecloak,id=113916,bonus_id=566/448,enchant=gift_of_haste
chest=battleplate_of_guiding_light,id=115566,bonus_id=566
wrists=bracers_of_visceral_force,id=119331,bonus_id=566/448
hands=gauntlets_of_guiding_light,id=115567,bonus_id=566
waist=uktars_belt_of_chiming_rings,id=113976,bonus_id=566/448
legs=legplates_of_guiding_light,id=115569,bonus_id=566
feet=sabatons_of_fractal_earth,id=113936,bonus_id=566/448
finger1=seal_of_unquenchable_flame,id=113922,bonus_id=566/448,enchant=gift_of_haste
finger2=phosphorescent_seal,id=113866,bonus_id=566/448,enchant=gift_of_haste
trinket1=forgemasters_insignia,id=113983,bonus_id=566/448
trinket2=horn_of_screaming_spirits,id=119193,bonus_id=566/448
main_hand=kromogs_brutal_fist,id=113927,bonus_id=566/448,enchant=mark_of_warsong
off_hand=kromogs_protecting_palm,id=113926,bonus_id=566/448

Topic 11 Placeholder

Topic 12 Placeholder

Glossary

This section will define acronyms that appear over and over again.

Current Terminology (See the 4.0/3.0 threads if you want discontinued Cataclysm/Wrath acronyms)

• AS - Avenger's Shield. Can also be used to represent Glyph of the Alabaster Shield when referring to glyphs.
• AM - Active Mitigation
• CS - Crusader Strike.
• Cons - Consecration, or Glyph of Consecration when referring to glyphs.
• DivPurp - Divine Purpose.
• DP - Divine Purpose or Divine Protection. Generally clear from context.
• FoL - Flash of Light
• GC or GrCr - Grand Crusader.
• GoBH - Glyph of the Battle Healer.
• GbtL - Guarded by the Light.
• HA - Holy Avenger
• HaNova or HammerNova - The AoE portion of Hammer of the Righteous.
• HotR - Hammer of the Righteous or Glyph of Hammer of the Righteous depending on context.
• HoPo - Holy Power.
• HP - Holy Power, or on rare occasions Hit Points. This should be clear enough by the context.
• HW - Holy Wrath.
• HoW - Hammer of Wrath
• J or Jud- Judgement, or Glyph of Judgement when referring to glyphs.
• SH - Selfless Healer
• ShoR Shield of Righteousness the Righteous, deprecated in favor of SotR. You may still catch me using it by accident though.
• SoI - Seal of Insight.
• SoR - Seal of Righteousness
• SoT - Seal of Truth.
• SotR - Shield of the Righteous. In rotation simulations, SotR# refers to using SotR at #-holy-power. In other words, SotR5 means, "SotR if and only if we have 5 holy power."
• SW - Sanctified Wrath
• WoG - Word of Glory

Simulation Notes

Write something insightful about simulation vs. formulation here.

Outstanding Requests

See the Call to Arms thread.

As of today, I'm opening this thread for discussion. I've spent the better part of the summer implementing things in Simulationcraft and keeping the paladin module up-to-date. We've got a number of fancy new features we can use to analyze results compared to last expansion, most of which are documented on the Simulationcraft For Tanks wiki page.

This thread is for discussion of simulation-related topics, both about the accuracy and features of Simulationcraft itself and about the results that it produces. Most of this thread will probably end up being focused on using those results to optimize and guide our play, but discussions of bugs and features is also encouraged. However....

Questions such as "What rotation/talents/glyphs/etc should I be using at level 100" will be deleted.

This is not a beginner's thread. It is not a "help me learn to paladin" thread. It is a hardcore, number-crunching, analyze-the-shit-out-of-the-data thread. I will moderate useless posts as I see fit. You've been warned.

Contributing to the thread
If you're contributing data/reports to the thread, please try and be as complete as possible. For example, don't just take a screenshot of the one graph you want to draw attention to and call it a day. Also upload the html report somewhere that everyone can access it, so that we can look at other parts of the report that may be relevant to the graph you shared. There are plenty of free services you can use for this (google drive, dropbox, etc.)

Contributing to Simulationcraft
If you are interested in helping contribute, there are various ways you can do so. Developers are always handy (though, obviously, I have the paladin stuff pretty much covered). But it's also handy to have people participating in a non-coding role. For example:

• Comparing sim results to in-game tests for consistency
• Analyzing results and identifying bugs.
• Keeping the default profiles optimized, both in terms of gear and rotation
• Updating the documentation (dear lord do we need this)

The best place to start in this regard is to talk to me or one of the other developers. The Participate page of the documentation has details on who to talk to about getting repository access and has the information for the irc channel, which is often the fastest way to talk to us.

So, I promised a data dump, here it is.

First things first: before we can reasonably compare talents, glyhs, or anything else, we have to figure out our rotation. In particular, we need to develop the ideal action priority list, which we'll then use for a default in SimC. This is a more daunting task than in previous expansions, because we need to be a little more complete. If you look back at old threads, I primarily focused on getting a basic, default priority set up using a particular talent combination, and then used that to assess other things.

That worked well enough for rough approximations. It doesn't work as well if we want an APL that is optimized for any talent and glyph configuration though. For example, one would expect that the new version of Sanctified Wrath would significantly affect the priority of Holy Wrath (spoiler: it does). The same could be true of many of our talents, often in subtle and easily-overlooked ways. For example, Sacred Shield and Selfless Healer cost GCDs every so often, while EF doesn't - but it has a Holy Power cost. One might assume that a GCD is a trivial cost, until you consider that Empowered Seals eats up all of our free GCDs - at which point the loss of GCDs actually decreases the effectiveness of one or the other. At which point, the ideal priority of Empowered Seals and/or Sacred Shield might change!

A little quick combinatorics tells you that simulating every permutation of all of our possible skills under all talent configurations is not feasible. So we need a more directed method that gets us close enough to optimal that we can finish the job with smaller, more controlled tests. We used a very similar procedure in MoP and before with the MATLAB code, but we did it all in one simulation and made a lot more assumptions. This time, we're going to be a lot more thorough.

That necessitates splitting everything up into a number of different simulations though, which we will use to slowly patchwork together our ideal APL. Many posts incoming, handle them.

First things first, let's get a handle on the basic rotation stub that we're going to be using. In MoP, this was the "CS>J>AS>HW>Cons" that you all know and love. But even that isn't sacrosanct - HW's cooldown changed to 15s, and the damage values of all of these abilities have been shuffled. So let's see how they stack up now.

We're going to make heavy use of the Automation Tool to perform these tests. That's why I wrote it, of course - I wanted to replace my MATLAB code with something more easily accessible and maintained. Originally I thought I'd use python, until a collaborator pointed out that we could just write it all in C++ and build it into the GUI so anyone could use it, even if they didn't know python. A++ for that suggestion, by the way (Thanks Seth!).

So, we fire up the automation tool and choose the following settings:

Comparison Type: Rotation
Class: Paladin
Spec: Protection
Race: Blood Elf
Level: 100
Default Talents: 0032323
Default Glyphs: alabaster_shield/focused_shield
Default Gear: T17H_gear.simc (see first page)
Actions Header:

Code: Select all

actions.precombat=flask,type=greater_draenic_stamina_flask
actions.precombat+=/food,type=talador_surf_and_turf
actions.precombat+=/blessing_of_kings,if=(!aura.str_agi_int.up)&(aura.mastery.up)
actions.precombat+=/blessing_of_might,if=!aura.mastery.up
actions.precombat+=/seal_of_insight
actions.precombat+=/sacred_shield
actions.precombat+=/snapshot_stats

actions=/auto_attack
actions+=/holy_avenger,if=talent.holy_avenger.enabled
actions+=/shield_of_the_righteous,if=holy_power>=5|buff.divine_purpose.react|incoming_damage_1500ms>=health.max*0.3


Actions Footer:

Code: Select all

actions+=/sacred_shield

Rotation Configurations:

Code: Select all

CS>J>AS>HW>Cons name="CS>J>AS>HW>Cons(Basic)"
CS>J>AS>Cons>HW
CS>J>AS+GC>Cons>AS>HW
CS>J>Cons>AS>HW
CS>Cons>J>AS>HW
Cons>CS>J>AS>HW
CS>J>AS+GC>HW>AS>Cons
CS>J>HW>AS>Cons
CS>HW>J>AS>Cons
HW>CS>J>AS>Cons
J>CS>AS>HW>Cons
J>AS+GC>CS>AS>HW>Cons
J>AS>CS>HW>Cons
AS>J>CS>HW>Cons
AS>CS>J>HW>Cons

Footer: Empty

Final imported profile: http://pastebin.com/ZQj132hw
HTML Report: https://www.dropbox.com/s/atykkv5480i33 ... .html?dl=0

Relevant plots:


Commentary:
First of all, note that this sim was run in build 18865. Nothing significant has changed in b18934 to change these results (the buff to HS will change numbers, but not relative rankings). Who knows what will happen in today's build, but we can more or less expect these results to hold until one of these spells is changed, at which point we'll re-run.

Also note that many of these sims will have TMI stat weights. I stopped doing that after a while because they take much longer than a 25k sim w/o stat weights, and in practice we don't really need stat weights all along the way. Once I re-run those stat weights will probably go.

Default talents were chosen to make the situation as simple as possible. Sacred Shield is there because sticking it on the end of the rotation is a fairly simple thing to do. Unbreakable Spirit is pointless since we don't use anything, but it's a fairly obvious default. Divine Purpose acts independently of the rotation, L90 talent is irrelevant since we're not using it, and Holy Shield is similarly passive.

Default Glyphs were Alabaster Shield (again, basically passive) and Focused Shield. Note that this will make AS 30% stronger for single-target; we'll probably want to go back once we have a more complete APL and see exactly where AS fits in with and without the glyph.

I don't think there's anything too controversial in the Action Header/Footer. I'm going to be abusing the Automation Tool's flexibility and ability to parse shorthands throughout these posts, so get used to it now!

Let's talk about the results. First of all, DPS accuracy is about +/- 3 DPS, TMI is repeatable to ~28 (not 28k... 28).

CS>J>AS>Cons>HW seems to be the DPS winner. Even though HW hits like a truck, it's not that much more truck-like than Cons:

And Cons's shorter cooldown means that we lose more DPS by delaying a Cons than by delaying a HW. Not an unprecedented result, anyone who remembers Wrath-era ret will recall that same situation with CS.

That said, the difference between Cons>HW and HW>Cons isn't huge. And understandably, we wouldn't expect much difference in DTPS or TMI rankings either. Any differences here can pretty much be attributed to SS uptimes and refreshing. More efficient use of GCDs generally means higher DPS, but fewer GCDs for refreshing SS. That said, the rotation here is so sparse that we have little trouble keeping SS up 100% of the time. Also note that SS reporting on the "Healing and Absorb Stats" table is mildly bugged in the version of the sim I used to run this (errantly showing up as having direct results and tick results). We've fixed it since then.

So what we can conclude from this is,

• CS>J>AS is still our first priority
• Cons>HW is winning for a basic rotation, but the difference is tiny

That second point will be important because we expect the order of Cons & HW to depend on several other factors - SW obviously, but also how many free GCDs we have to work with. If we end up being GCD-constrained, such that we can only cast HW or Cons, and can't cast another one for quite a while, then HW is the obvious choice. As such, we may decide to sim both HW>Cons and Cons>HW versions of rotations once we add some spells.

Next up, where do we fit HoW? Easiest thing to do here is just sim the basic rotation with HoW stuffed into every possible slot.

All of the simulation settings are the same as the previous post, with two notable exceptions:

Rotation Configurations:

Code: Select all

#hw>cons
CS>J>AS>HW>Cons
CS>J>AS>HW>Cons>HoW
CS>J>AS>HW>HoW>Cons
CS>J>AS>HoW>HW>Cons
CS>J>AS+GC>HoW>AS>HW>Cons
CS>J>HoW>AS>HW>Cons
CS>HoW>J>AS>HW>Cons
HoW>CS>J>AS>HW>Cons
#cons>hw
CS>J>AS>Cons>HW
CS>J>AS>Cons>HW>HoW
CS>J>AS>Cons>HoW>HW
CS>J>AS>HoW>Cons>HW
CS>J>AS+GC>HoW>AS>Cons>HW
CS>J>HoW>AS>Cons>HW
CS>HoW>J>AS>Cons>HW
HoW>CS>J>AS>Cons>HW

Footer:

Code: Select all

enemy_initial_health_percentage=20


(this just keeps us in execute range the entire time, so we get a larger variation between results and thus more accuracy)

Final imported profile: http://pastebin.com/Rj6hZkV7

HTML Report: https://www.dropbox.com/s/1bj4l7d8zvx6y ... .html?dl=0

Relevant Plots:



Commentary:
Again, this is using build 18898. DPS error is ~4, TMI error ~42.

As far as DPS, it's clear that HoW is our highest priority. Both HoW>everything_else queues tie for top DPS spot. The tie is because the rotation devolves into HoW-CS-J-X, where X just alternates between AS, Cons, and HW, so there's no real clashing and the HW>Cons vs Cons>HW distinction fails to mean anything.

On the TMI side, using HoW at all is a survivability loss because we're eliminating empty GCDs that we could be using to refresh Sacred Shield, leading to lower SS uptime. We're not too concerned with that, since we'll deal with SS refreshes in a later sim. So as far as survivability, we're more concerned with the minor differences between the rotations involving HoW. In this case, it looks like slotting HoW behind Avenger's Shield is the best trade-off between survivability and DPS. CS>J>AS>HoW>(HW/Cons) seems to be the runner-up in the DPS department, and performs rather well in TMI since it doesn't impact holy power generation.

So, the conclusion here is that HoW should fit in right behind AS, giving CS>J>AS>HoW>(HW or Cons).

While we're still trying to figure out the Holy Wrath / Consecration stuff, it seemed like a good time to figure out how the Glyph of Final Wrath fits in.

All of the simulation settings are the same as the previous post, with two notable exceptions:

Default Glyphs: alabaster_shield/focused_shield/final_wrath

Rotation Configurations: (FW here is holy_wrath,if=glyph.final_wrath.enabled&target.health.pct<=20

Code: Select all

#hw>cons
CS>J>AS>HoW>HW>Cons
CS>J>AS>FW>HoW>HW>Cons
CS>J>AS+GC>FW>AS>HoW>HW>Cons name="CS>J>AS+>FW>AS>HoW>HW>Cons"
CS>J>FW>AS>HoW>HW>Cons
CS>FW>J>AS>HoW>HW>Cons
FW>CS>J>AS>HoW>HW>Cons
#cons>hw
CS>J>AS>HoW>Cons>HW
CS>J>AS>HoW>FW>Cons>HW
CS>J>AS>FW>HoW>Cons>HW
CS>J>AS+GC>FW>AS>HoW>Cons>HW name="CS>J>AS+>FW>AS>HoW>cons>HW"
CS>J>FW>AS>HoW>Cons>HW
CS>FW>J>AS>HoW>Cons>HW
FW>CS>J>AS>HoW>Cons>HW


Footer:

Code: Select all

enemy_initial_health_percentage=40


(this splits the sim into half execute range and half outside, to make sure the difference between FW and HW is clear)

Final imported profile: http://pastebin.com/AmNWfxeU

HTML Report: https://www.dropbox.com/s/n6budp6mipiq5 ... .html?dl=0

Relevant Plots:



Commentary:
DPS error ~5, TMI error ~46, build 18898 again

Glyphed Holy Wrath is obviously a heavy hitter during execute range. The TMI cost of prioritizing it ahead of everything doesn't appear to be that large, but to be fair neither are the DPS gains. The queue that gives the highest DPS while still being tied for minimum TMI is
CS>J>AS>HoW>FW>Cons>HW, though again, the difference between Cons>HW and HW>Cons (or really, any of these rotations) is very small. Still, that rotation makes some sense. It still prioritizes HP generation, but tries to fill as many GCDs as possible with heavy-hitting HoW & FW during execute.

So, conclusion here is that CS>J>AS>HoW>FW>Cons>HW is going to be our basic go-to rotation for general tanking. We can tweak that for higher DPS later by prioritizing things like HoW and FW, but let's focus on getting the basic survival rotation down first.

Believe it or not, the next thing to test is our L90 talents, mostly because they're easy. So we may as well get them out of the way.

We'll do one sim for each talent, checking where it fits in the rotation. Setup has changed only slightly. Default Talents are irrelevant because I hardcoded the talents into the rotation specification. Action Header/Footer haven't changed.

Default Glyphs: alabaster_shield/focused_shield/final_wrath

Rotation Configurations:
ES:

Code: Select all

CS>J>AS>HoW>FW>Cons>HW
CS>J>AS>HoW>FW>Cons>HW>ES talents=0032333
CS>J>AS>HoW>FW>Cons>ES>HW talents=0032333
CS>J>AS>HoW>FW>ES>Cons>HW talents=0032333
CS>J>AS>HoW>ES>FW>Cons>HW talents=0032333
CS>J>AS>ES>HoW>FW>Cons>HW talents=0032333
CS>J>AS+GC>ES>AS>HoW>FW>Cons>HW talents=0032333 name="CS>J>AS+>ES>AS>HoW>FW>Cons>HW"
CS>J>ES>AS>HoW>FW>Cons>HW talents=0032333
CS>ES>J>AS>HoW>FW>Cons>HW talents=0032333
ES>CS>J>AS>HoW>FW>Cons>HW talents=0032333


LH:

Code: Select all

CS>J>AS>HoW>FW>Cons>HW
CS>J>AS>HoW>FW>Cons>HW>LH talents=0032323
CS>J>AS>HoW>FW>Cons>LH>HW talents=0032323
CS>J>AS>HoW>FW>LH>Cons>HW talents=0032323
CS>J>AS>HoW>LH>FW>Cons>HW talents=0032323
CS>J>AS>LH>HoW>FW>Cons>HW talents=0032323
CS>J>AS+GC>LH>AS>HoW>FW>Cons>HW talents=0032323 name="CS>J>AS+>LH>AS>HoW>FW>Cons>HW"
CS>J>LH>AS>HoW>FW>Cons>HW talents=0032323
CS>LH>J>AS>HoW>FW>Cons>HW talents=0032323
LH>CS>J>AS>HoW>FW>Cons>HW talents=0032323

HPR:

Code: Select all

CS>J>AS>HoW>FW>Cons>HW
CS>J>AS>HoW>FW>Cons>HW>HPr talents=0032313
CS>J>AS>HoW>FW>Cons>HPr>HW talents=0032313
CS>J>AS>HoW>FW>HPr>Cons>HW talents=0032313
CS>J>AS>HoW>HPr>FW>Cons>HW talents=0032313
CS>J>AS>HPr>HoW>FW>Cons>HW talents=0032313
CS>J>AS+GC>HPr>AS>HoW>FW>Cons>HW talents=0032313 name="CS>J>AS+>HPr>AS>HoW>FW>Cons>HW"
CS>J>HPr>AS>HoW>FW>Cons>HW talents=0032313
CS>HPr>J>AS>HoW>FW>Cons>HW talents=0032313
HPr>CS>J>AS>HoW>FW>Cons>HW talents=0032313

Footer: None

Final imported profiles:
ES: http://pastebin.com/c5JGv0kD
LH: http://pastebin.com/t0mmQMbb
HPr: http://pastebin.com/JYAft3n4

HTML Reports:
ES: https://www.dropbox.com/s/rjbv1ls5p2x09 ... .html?dl=0
LH: https://www.dropbox.com/s/nryyfb8scjmd7 ... .html?dl=0
HPr: https://www.dropbox.com/s/u6lq52efu8o35 ... .html?dl=0

Relevant Plots:









Commentary
DPS error ~5, TMI error ~43, build 18898 again

Probably not all that surprising that ES>all is top DPS given its insane DPET. A very close second is HPG>ES, which also isn't that surprising. Since this is also tied for best TMI, I think it's the clear winner, with the caveat that it almost doesn't matter where you put ES for TMI because one cast every minute is a very minor impact. As long as it's behind HPG, it really can't hurt you at all.

Mental note: need to check SimC and make sure that ES is updating dynamically with buffs and such. I think it is, but tbh I don't think I got around to checking that, either in-game or in SimC. If someone has time and can test that and get back to me, that would be great.

LH is much less powerful than it used to be, with a DPET of less than half of ES. That's obviously on single-target though, for AoE that should go up fairly dramatically. Practically speaking, it hardly matters where you slot it in as long as it's in a filler spot. I chose to go with HoW>LH>FW because that was top DPS and min-TMI, but it can really go just about anywhere with minimal impact. Which is actually a bit of a plus for the talent in my mind. If you're taking LH you're probably using it as a raid cooldown somewhere (well... pre-Resolve anyhow, maybe not anymore...) so you'd be casting it on demand rather than just slipping it into the rotation.

Holy Prism has an absolutely abysmal DPET for a L90 talent. It's worse than Holy Wrath and Avenger's Shield, on par with HoW. I'm not sure what caused it to get such a large nerf in WoD, but whatever it was, it made this talent fairly weak. The new Resolve mechanics mean that it's raid healing side-effect is basically useless. It's still useful as a self-heal and AoE snap aggro, presumably, though we're not testing either of those here. We're just testing single-target damage and splash self-healing, and it's pretty pitiful. And it has very little effect on TMI. So, to the back of the priority list it goes.

Note that the LH results may be a little tainted by the fact that it's healing all of the other paladins in the sim, which means that its self-healing (which is all that matters for TMI) may be artificially low because of the 6-target cap. I don't think this will be significant enough to change our results though, so I'm not going to bother to re-test it. Feel free to do so on your own if you think it matters.

Conclusion: Default rotation updated to
CS>J>AS>ES>HoW>LH>FW>Cons>HW>HPr

Next test is level 75 talents. Divine Purpose (which has been our default all along so far) doesn't really need to be re-tested, as all of its effects are off-GCD and should have no real impact on the rotation. Likewise, Holy Avenger is pretty much all off-GCD as long as we're prioritizing holy power generators, which we are. So this category really comes down to testing the new Sanctified Wrath implementation.

So, we set our default talents appropriately and do just that. Actions Header/Footer unchanged, Footer empty.


Default Talents: 0032223
Default Glyphs: alabaster_shield/focused_shield/final_wrath

Rotation Configurations:

Code: Select all

CS>J>AS>HoW>LH>FW>Cons>HW
CS>J>AS>HoW>LH>HW>Cons
CS>J>AS>HoW>HW>LH>Cons
CS>J>AS>HW>HoW>LH>Cons
CS>J>HW>AS>HoW>LH>Cons
CS>HW>J>AS>HoW>LH>Cons
HW>CS>J>AS>HoW>LH>Cons


Final imported profile: http://pastebin.com/fPZeeE2m

HTML Report: https://www.dropbox.com/s/h2tilpzc7p365 ... .html?dl=0

Relevant Plots:



Commentary
DPS error ~4, TMI error ~37, build 18898 again

You can see from the DPET plot that SW makes Holy Wrath a beast. Admittedly that's averaging in some casts with Final Wrath glyphed, but still... holy crap. It's no wonder that it gets a high billing on the DPS rankings.

Since it's a Holy Power generator now, it also does rather well no matter where you slot it in. There's a slight TMI cost to putting it ahead of CS, but not a lot. Then again, the DPS gain for slotting it ahead of CS isn't that huge either (<1%). Part of that has to do with pushback: HW>CS>J gives you the repeatable pattern:
HW-CS-J-X-CS-X-J-CS-X-X-

Which means you're pushing CS and J back by one GCD each every cycle. In comparison, CS>J>HW delays HW by 1, 2, 0, 2 GCDs each time it's up. This is easier to sketch out by hand:


Ultimately, I decided that the "safe" rotation here was CS>J>HW>AS>HoW>LH>Cons. It was min TMI, and second-best DPS. We'll keep "HW>all" in the back of our head for when we're developing a max-DPS queue for the tiny gain it'll give us, but especially for less advanced tanks I think the CS-J cycle is worth maintaining.

Conclusion:
With Sanctified Wrath, we'll want to use CS>J>HW>AS>HoW>LH>Cons. That means we'll be using a holy_wrath,if=talent.sanctified_wrath.enabled line between J and AS, and another simple "holy_wrath" line at the end after Cons. And of course, our final wrath line for when SW isn't talented.

Obviously have a lot more to post, but I have to make dinner and do things. SS / L100 talents to come. Opening thread for discussion though, so feel free to go nuts, just keep in mind that asking questions like "WHAT ABOUT L100 TALENTS" is dumb since you know they're coming.

Resuming where I left off yesterday, the next thing I looked at is our L45 talents. We'll start with Sacred Shield.

I modified the Actions Header slightly for the L45 talents since now we want to include Eternal Flame. I'm using the same EF refresh lines that we did in 5.x simulations. Obviously it won't affect the Sacred Shield or Selfless Healer sims, but I'm using the same header for all three sims just for consistency.

Default Talents: 0032223
Default Glyphs: alabaster_shield/focused_shield/final_wrath

Actions Header:

Code: Select all

actions.precombat=flask,type=greater_draenic_stamina_flask
actions.precombat+=/food,type=talador_surf_and_turf
actions.precombat+=/blessing_of_kings,if=(!aura.str_agi_int.up)&(aura.mastery.up)
actions.precombat+=/blessing_of_might,if=!aura.mastery.up
actions.precombat+=/seal_of_insight
actions.precombat+=/sacred_shield
actions.precombat+=/snapshot_stats
actions=/auto_attack
actions+=/eternal_flame,if=buff.eternal_flame.remains<2&buff.bastion_of_glory.react>2&(holy_power>=3|buff.divine_purpose.react|buff.bastion_of_power.react)
actions+=/eternal_flame,if=buff.bastion_of_power.react&buff.bastion_of_glory.react>=5
actions+=/shield_of_the_righteous,if=holy_power>=5|buff.divine_purpose.react|incoming_damage_1500ms>=health.max*0.3


Actions Footer:

Code: Select all

actions+=/sacred_shield


Rotation Configurations:

Code: Select all

CS>J>HW>AS>HoW>LH>Cons talents=0002223
CS>J>HW>AS>HoW>LH>Cons name=Cons>SS
CS>J>HW>AS>HoW>LH>SS+DR<2>Cons name=SS2>Cons
CS>J>HW>AS>HoW>SS+DR<2>LH>Cons name=SS2>LH
CS>J>HW>AS>SS+DR<2>HoW>LH>Cons name=SS2>HoW
CS>J>HW>SS+DR<2>AS>HoW>LH>Cons name=SS2>AS
CS>J>SS+DR<2>HW>AS>HoW>LH>Cons name=SS2>HW(SW)
CS>SS+DR<2>J>HW>AS>HoW>LH>Cons name=SS2>J
SS+DR<2>CS>J>HW>AS>HoW>LH>Cons name=SS2>CS


Final imported profile: http://pastebin.com/1nS8iwNj

HTML Report: https://www.dropbox.com/s/b4jzvjofj91w8 ... .html?dl=0

Relevant Plots:


Commentary
DPS error ~4, TMI error ~32, build 18912

We start this one out with our basic rotation (with Sanctified Wrath taken, so Holy Wrath has an elevated priority) and with no L45 talent selected, giving us our "CS>J>HW>AS>HoW>LH>Cons" parse. This is obviously the top DPS parse and the worst TMI parse. Adding the sacred shield talent and a simple "/actions+=sacred_shield" line to the profile nets us about 10k TMI at a cost of a little DPS. The DPS cost can be explained by the fact that if you use SS in what was otherwise going to be an empty GCD, you cannot immediately react to a Grand Crusader proc should one occur.

We can do a little better by adding a line to refresh SS if there's only a short duration left. The line I'm using here is
"actions+=/sacred_shield,if=target.dot.sacred_shield.remains<2"
which refreshes SS if there's less than 2 seconds left on the driver buff. You can see that there isn't a lot of variation here though. Partly that's because we're generally refreshing via the fallback option, so we don't often have to invoke the emergency refresh line. I chose to use "SS2>Cons" going forward, since it's a reasonably high-DPS rotation and there isn't much benefit to be had by pushing it up higher in the queue.

That said, I want to go back and reconsider some of this once we have a fuller rotation. Maybe with more used GCDs, we'll want to push that emergency line up right behind holy power generators to make sure we're keeping SS up. There's also room to play with the timing here. Two seconds was sort of arbitrary, and you can imagine that it might perform well slotted behind HPGs with a slightly earlier refresh. Recall that the last tick is a partial one with the new HoT/DoT system, so there may be incentive to refresh as much as 6 seconds early since we don't lose that time.

Conclusion: for now, we'll add an SS refresh line right ahead of Cons, cognizant of the fact that we may modify that line later when we sim it in more detail.

Selfless Healer is one we don't think about much, but it's worth looking at nonetheless. We'll add a
"actions+=/flash_of_light,if=buff.selfless_healer.stack>=3"
line to the APL and see where it fits in. This will at least give us a representation of the stochastic effect it would give you if you just used it whenever it was available, similar to the passive self-healing of EF or SS. In theory, we may do better with a conditional that responds to damage, though we didn't really have much luck with those when trying to do that with EF. But again, that's kind of advanced tweaking, so we'll come back to that later once we have a solid base APL to work from.

The only changes from the last sim are the default talents and (obviously) the rotation shorthands:

Default Talents: 0012223

Rotation Configurations:

Code: Select all

CS>J>HW>AS>HoW>LH>Cons>FoL+SH3 name=CS>J>HW>AS>HoW>LH>Cons>SH
CS>J>HW>AS>HoW>LH>FoL+SH3>Cons name=CS>J>HW>AS>HoW>LH>SH>Cons
CS>J>HW>AS>HoW>FoL+SH3>LH>Cons name=CS>J>HW>AS>HoW>SH>LH>Cons
CS>J>HW>AS>FoL+SH3>HoW>LH>Cons name=CS>J>HW>AS>SH>HoW>LH>Cons
CS>J>HW>FoL+SH3>AS>HoW>LH>Cons name=CS>J>HW>SH>AS>HoW>LH>Cons
CS>J>FoL+SH3>HW>AS>HoW>LH>Cons name=CS>J>SH>HW>AS>HoW>LH>Cons
CS>FoL+SH3>J>HW>AS>HoW>LH>Cons name=CS>SH>J>HW>AS>HoW>LH>Cons
FoL+SH3>CS>J>HW>AS>HoW>LH>Cons name=SH>CS>J>HW>AS>HoW>LH>Cons


Final imported profile: http://pastebin.com/9taHcCui

HTML Report: https://www.dropbox.com/s/5wvurc0d3x5sy ... .html?dl=0

Relevant Plots:


Commentary
I forgot to include a baseline in this sim, but you can scroll up to see what it was in the last sim: about 166k TMI and 17619 DPS. SH has almost no effect on TMI in this simple scenario, and causes a somewhat inexplicable DPS loss. I haven't had time to look into exactly what's causing that, but I plan on re-simming this with the baseline added anyway, so we'll look into it later. For now we'll slot SH in at the end of our queue and just make a mental note that we need to come back to this one. When we do, we'll also look at various conditionals that might improve its effectiveness over the stochastic model.

I didn't run any sims to figure out how to optimize EF, by the way. For now I'm just assuming that our implementation from 5.x is the best use of it. Again, that's something we can revisit as we try to refine the model.