[TC] Mage Bomb Snapshotting - Arcane Edition

[Komma]

I made a thread last month about snapshotting as a Frost mage as a single target fight. Since then, there's been a number of mages claiming on forums that the results were valid only for Frost, not Arcane. I've also received a number of questions asking if something similar has been done for Arcane. It's a lot of work to set up, troubleshoot and verify Snapshotting sims. It also takes a very long time to run the high number of iteration required to get accurate results.

Once again we start with our questions:

"Does snapshotting Mage Bombs on 450 second encounters increase the DPS of a 588 itemlevel BiS Arcane Mage?"

Part 1: Setup
For the purposes of this test, I used the default 4T16H profile. You can find it under "4-set BiS list" in Voltaa's compendium: http://altered-time.com/forum/viewtopic ... nchor-Gear" target="_blank

Fight duration is set to 450 seconds by default. I don't think duration has too much to do with whether snapshotting will give a bonus or not.

Simulations were performed on a custom version of SimC 548-7, modified to allow 4th LB casts.

One problem people complained about last time is that I focused only on single target. Using more advanced tools than last time (see next section), I've decided to expand this test to 3 targets this time. With Living Bomb, this required some code hacks, since the current Mage module does not like casting a 4th LB when 3 are already active. There is also one other issue: SimC does not allow separation of enemies right now, so LB explosions will always hit all targets. With NT this is of course desired behavior, but with LB it's not perfectly accurate. For our purposes though, it shouldn't affect conclusions on snapshotting.


Snapshotting Methodology
After how the last thread turned out, many folks asked if I could simply use a method more translatable to AffDots, since that is how most folks do their snapshotting. Turkeyburger has been kind enough to provide me with a WeakAura string that does roughly the same thing. I've been able to reverse engineer it to provide behavior similar to AffDots - computing a relative % strength between your castable Mage Bomb against the actively ticking one on the target.

I decided to make 5 different cases - Snapshotting at 120%, 140%, 160%, 180%, and 200%. There are also additional checks: making sure that the target will live for a number of ticks, or waiting until the current bomb has a limited number of ticks remaining before refreshing it. Here's the code used for LB:

Code: Select all

# Snapshotting test block # Case 1 : WA 120 # actions.single_target+=/living_bomb,cycle_targets=1,if=talent.living_bomb.enabled&ticks_remain<=2&tick_damage*(100+crit_pct_current)>=dot.living_bomb.tick_dmg*(100+dot.living_bomb.crit_pct)*(100+dot.living_bomb.haste_pct)*spell_haste*0.012&target.time_to_die>tick_time*3 # Case 2 : WA 140 # actions.single_target+=/living_bomb,cycle_targets=1,if=talent.living_bomb.enabled&ticks_remain<=2&tick_damage*(100+crit_pct_current)>=dot.living_bomb.tick_dmg*(100+dot.living_bomb.crit_pct)*(100+dot.living_bomb.haste_pct)*spell_haste*0.014&target.time_to_die>tick_time*3 # Case 3 : WA 160 # actions.single_target+=/living_bomb,cycle_targets=1,if=talent.living_bomb.enabled&ticks_remain<=2&tick_damage*(100+crit_pct_current)>=dot.living_bomb.tick_dmg*(100+dot.living_bomb.crit_pct)*(100+dot.living_bomb.haste_pct)*spell_haste*0.016&target.time_to_die>tick_time*3 # Case 4 : WA 180 # actions.single_target+=/living_bomb,cycle_targets=1,if=talent.living_bomb.enabled&ticks_remain<=2&tick_damage*(100+crit_pct_current)>=dot.living_bomb.tick_dmg*(100+dot.living_bomb.crit_pct)*(100+dot.living_bomb.haste_pct)*spell_haste*0.018&target.time_to_die>tick_time*3 # Case 5 : WA 200 # actions.single_target+=/living_bomb,cycle_targets=1,if=talent.living_bomb.enabled&ticks_remain<=2&tick_damage*(100+crit_pct_current)>=dot.living_bomb.tick_dmg*(100+dot.living_bomb.crit_pct)*(100+dot.living_bomb.haste_pct)*spell_haste*0.02&target.time_to_die>tick_time*3 # Snapshotting test block ends

And here's the code used for NT:

Code: Select all

# Snapshotting test block # Case 1 : WA 120 # actions.single_target+=/nether_tempest,cycle_targets=1,if=talent.nether_tempest.enabled&action.nether_tempest.tick_damage*(100+action.nether_tempest.crit_pct)>=dot.nether_tempest.tick_dmg*(100+dot.nether_tempest.crit_pct)*(100+dot.nether_tempest.haste_pct)*spell_haste*0.012&target.time_to_die>6&ticks_remain<=6 # Case 2 : WA 140 # actions.single_target+=/nether_tempest,cycle_targets=1,if=talent.nether_tempest.enabled&action.nether_tempest.tick_damage*(100+action.nether_tempest.crit_pct)>=dot.nether_tempest.tick_dmg*(100+dot.nether_tempest.crit_pct)*(100+dot.nether_tempest.haste_pct)*spell_haste*0.014&target.time_to_die>6&ticks_remain<=6 # Case 3 : WA 160 # actions.single_target+=/nether_tempest,cycle_targets=1,if=talent.nether_tempest.enabled&action.nether_tempest.tick_damage*(100+action.nether_tempest.crit_pct)>=dot.nether_tempest.tick_dmg*(100+dot.nether_tempest.crit_pct)*(100+dot.nether_tempest.haste_pct)*spell_haste*0.016&target.time_to_die>6&ticks_remain<=6 # Case 4 : WA 180 # actions.single_target+=/nether_tempest,cycle_targets=1,if=talent.nether_tempest.enabled&action.nether_tempest.tick_damage*(100+action.nether_tempest.crit_pct)>=dot.nether_tempest.tick_dmg*(100+dot.nether_tempest.crit_pct)*(100+dot.nether_tempest.haste_pct)*spell_haste*0.018&target.time_to_die>6&ticks_remain<=6 # Case 5 : WA 200 # actions.single_target+=/nether_tempest,cycle_targets=1,if=talent.nether_tempest.enabled&action.nether_tempest.tick_damage*(100+action.nether_tempest.crit_pct)>=dot.nether_tempest.tick_dmg*(100+dot.nether_tempest.crit_pct)*(100+dot.nether_tempest.haste_pct)*spell_haste*0.02&target.time_to_die>6&ticks_remain<=6 # Snapshotting test block ends

These look a bit more sophisticated than last time, and when I have time maybe I'll also run them for Frost.

Part 3: When to snapshot?
With Frost, there was some consensus that snapshotting is better done when you have nothing else to do. With Arcane it's a bit less obvious - should you actively snapshot when you have Missile procs available? What about 2T16 stacks, or Arcane Charges?

I settled on 3 variations, each of which will be suffixed in the results. The "Regular" snapshotting will be done when there is nothing better to do than cast Arcane Blast to build charges. The "Aggressive" profile, suffixed with "_AGG", chooses to prioritize snapshot over the regular Arcane spells (AB/AM/ABar). The "Extra Aggressive" profile, suffixed with "_AGGX", is the "Aggressive" profile coupled with ignoring current remaining bomb duration.

Part 4: Simulation results
Single Target Living Bomb:

450s duration, 2000K iterations, DPS error = 26.30
(Komma's comment: 2 million iterations later and the differences are still smaller than the error bar...)

3 Target Living Bomb:

450s duration, 250K iterations, DPS error = 106.48

3 Target Nether Tempest:

450s duration, 250K iterations, DPS error = 134.34
(Komma's first reaction to this chart: http://i.imgur.com/XsD8Mia.jpg" target="_blank)

Part 5: Conclusions
The charts pretty much tell their own story.

Regarding LB:
1. Snapshotting with LB on single target is a complete waste of time.
2. Snapshotting with LB on multiple targets is also not worth it.
3. The higher you set your snapshotting threshold, the better - mostly because the less snapshotting you do.
4. Getting more aggressive with LB snapshotting only makes things worse.
5. The ONLY exception is if you aggressively snapshot massive 200%+ multipliers, on multiple target fights.
6. NEVER ignore current LB's remaining duration when snapshotting.

Regarding NT on multiple targets:
1. Snapshotting with NT on multiple targets is definitely worth it.
2. Snapshotting is good even with a single trinket/meta proc advantage.
3. Snapshotting with NT on multiple targets should be prioritized over regular spells.
NOTE: I didn't run extra aggressive profiles for NT, so snapshots were only done on NTs with limited duration remaining.

As usual, I include a TL;DR note as an answer to the opening question.

"Snapshotting with Living Bomb as an Arcane Mage IS NOT a DPS gain, regardless of single or multi target."
"Snapshotting with Nether Tempest as an Arcane Mage IS a DPS gain on multi target fights."

[Chev]

I'm a while off being BIS but cheers for this. It is a very interesting read.

[Wilderness]

Thank you Komma, that was indeed very interesting. Guess I'm going to have start using that the WA that turkey posted that shows the %.

[Hendrikto]

I just released my fix to the AffDots Mage module, maybe you want to check it out, now that you know how and when to snapshot: http://altered-time.com/forum/viewtopic.php?f=3&t=376" target="_blank

Btw: Thanks for all the quality TC, Komma

[Trustbucket]

I would think that for NT, if the increase in percentage in strength is greater than the percentage left on the duration then it would be worth it to snapshot when the proc is about to end. Obviously its not that simple because of haste effects and the GCD time but I think it would still be worth it to snapshot with ~160% increase which is the percentage increase I get with two procs of trinkets or meta under certain circumstances.

The pull is an interesting time and I'm not sure whether or not it is a gain, but what I do is living bomb after precasting flamestrike or AB along with TW then when all of my trinkets proc I use arcane power and put up my mage bomb then go into AT. This way it saves a global inside AT, has a huge mage bomb while I am spamming missiles and AB inside AT, and the procs will usually still be up when the bomb is about to expire or explode because I reapplied it right before AT. This is a case where it seems worth it, too.

[Dutchmagoz]

Very interesting read. Great work as usual.

[Komma]

I would think that for NT, if the increase in percentage in strength is greater than the percentage left on the duration then it would be worth it to snapshot when the proc is about to end. Obviously its not that simple because of haste effects and the GCD time but I think it would still be worth it to snapshot with ~160% increase which is the percentage increase I get with two procs of trinkets or meta under certain circumstances.

The pull is an interesting time and I'm not sure whether or not it is a gain, but what I do is living bomb after precasting flamestrike or AB along with TW then when all of my trinkets proc I use arcane power and put up my mage bomb then go into AT. This way it saves a global inside AT, has a huge mage bomb while I am spamming missiles and AB inside AT, and the procs will usually still be up when the bomb is about to expire or explode because I reapplied it right before AT. This is a case where it seems worth it, too.

I struggled for a while with the amount of time/ticks left before letting is snapshot. There really isn't any documented research on timing or anything - as mentioned before, so much of it is just "hand-waving" and "sounds good" snapshotting strategy. It's not as simple as just calculating the DPET, subtracting the damage remaining on the ticking NT, and comparing it against other spells either; one of many issues is that every time you do an early refresh, you effectively reduce the DPET of the previous NT casted, and this affects the "correctness" of the previous refresh/snapshot.

I haven't thought of precasting flamestrike on multitarget fights. I should probably add that to fire spec in WoD sims.

The "apply bombs before and after AT" is more or less automatically handled by the snapshotting logic which takes into account all buffs and effects (since it directly compares tick damage, crit and haste). You could argue that the "refresh before buff expiration" logic is missing, but trying to incorporate that in combination with the multiplier logic seems rather difficult.

[Trustbucket]

I haven't thought of precasting flamestrike on multitarget fights. I should probably add that to fire spec in WoD sims.

Well its an increase now for arcane for fights that start with at least 3 targets like protectors and dark shamans. I'm not sure if it would be an increase for fire in WoD since I would think it would be more important to get up the pyroblast debuff along with an ignite that can be spread to other targets if it is a multi-target situation. Guess this is something that can be easily simmed though.

[trololololololol]

I've been replacing the precasted pyro with flamestrike in the opener for the last 3 months or so. Using common sense maths it shouldn't be a dps increase considering it does around 600k damage on a fight like protectors while precasting a pyro would mean a pyro dot + the ignite of the precasted pyro ticking (+ the subsequent fireball if no crit) before building combustion. If you do flamestrike you wont have pyro dot up on all 3 targets pre first combust spread. I do it however to increase consistancy in my pulls as flamestrike has a massive chance to proc immerseus trinket.

[Trustbucket]

I've been replacing the precasted pyro with flamestrike in the opener for the last 3 months or so. Using common sense maths it shouldn't be a dps increase considering it does around 600k damage on a fight like protectors while precasting a pyro would mean a pyro dot + the ignite of the precasted pyro ticking (+ the subsequent fireball if no crit) before building combustion. If you do flamestrike you wont have pyro dot up on all 3 targets pre first combust spread. I do it however to increase consistancy in my pulls as flamestrike has a massive chance to proc immerseus trinket.

It doesn't have to do with only the damage of the precasted pyro vs flamestrike but how it affects your opener because that is so much more important than a pre cast. Arcane it is better than arcane blast because it does more damage on that type of fight and procs your trinkets much more consistently while benefitting arcane's alter consistently because PBoI procs much less consistent than fire's. I highly doubt the damage of flamestrike would be greater than the precasted pyro hit, pyro dot on 3 targets and ignite (spread to three targets if you're smart) of that pyro.

No quantitative data to prove that the precasted pyro would proc trinkets more consistently with the pyro dot on 3 targets and the ignite but I can't remember PBoI not proccing when I need it as fire. Precasted flamestrike would 100% have a higher chance to proc Immerseus on pull but not in total, but as fire you don't need the proc for the first couple of seconds of the fight since you aren't doing anything important.

[trololololololol]

That's what i said - it doesn't do more damage but it adds consistancy ^^ I did forget to say that i only do this for fights with adds on the pull ie. Nazgrim, Dark shams, protectors and klaxxi.

[Trustbucket]

That's what i said - it doesn't do more damage but it adds consistancy ^^ I did forget to say that i only do this for fights with adds on the pull ie. Nazgrim, Dark shams, protectors and klaxxi.

But what I was saying is that it doesn't add consistency and loses damage. Anyway on Nazgrim arcane explosion on the adds will always proc trinkets, if thats what you are going for.