SimulationCraft for Warlocks (Panda Edition)
 

This post automatically updates to reflect the latest results at simulationcraft.org. Click on the charts below for details and scale factors:





The AoE report is from an 8-target simulation with a fight duration of 2-3 minutes.



In the pets/grimoires comparisons, the baseline T14N profile has no talents but is using the highest-DPS pet.







Watch the changelog for updates to the SimC warlock model. Please allow up to an hour for new changes to be reflected in the results.

The warlock module in simc has undergone a LOT of work, due to the large amount of changes and exotic mechanics that are being added to the warlock class in MoP. It's quite possible there are some bugs still, so please look over the results and make sure to let me know if you spot anything that looks wrong.

Two quick things about demo:
- Soul fire's damage seems excessive, might be, that its bugged currently in beta ofcourse, but im hitting nowhere near 100k/cast, around half of it to be honest. Additionally molten core stacks... and seems to not have a limit - i just had 45 stacks of it, so it might be a good idea to keep some stacks to prolong meta, although will be more practical when there's more than 1 target to shadowflame(and imp swarm is off cd) due to the low proc chance.
- Shadowflame stacks currently, so it seems best to "clip" the dot with your 2nd gul'dan charge for ~50% more dot damage - gul'dan > normal shadowflame ticks > gul'dan > double shadowflame ticks, "anty-clip" mechanic works, so the last tick is moved to your 2nd, double damage shadowflame.
- Demonic Slash deals some really pathetic damage currently, but that's more than likely a bug. ;)

When you're talking about soul fire, do you mean soul fire cast outside of meta? In the simulation, soul fire hits for anywhere from 60k to 250k, depending on whether meta is up and whether other buffs are aligning. But I only tested soul fire extensively as destro (before they removed it from destro), so it's possible something has changed or that the crit scaling works differently for demo or something.

The molten core stacking is odd, and may easily be a bug - I'm not going to spend time implementing that until I'm more confident it wouldn't be wasted time.

Shadowflame stacking is interesting, I hadn't realized that. Though from my very limited testing it doesn't look like a two-stack shadowflame is doing anywhere near double damage.

For demonic slash (and really, for most of the other abilities, with the exception of soul fire which I had to test for the crit scaling) I'm currently just trusting the tooltips and spell data.

EDIT: By the way, how would molten core stacks prolong meta?

12551sp(DI, 10xWoU), 950 crit, 991 mastery, coe'd target - 52-54k outside of meta, 63-64 with meta active. And yeah, my bad about the prolonging, tooltip still states "generates 30 demonic fury" while in meta, didn't notice the cost change... :P

Well I'm seeing almost double ticks - 1813 with single dot, 3566 with two stacks - took off legendary/cunning for testing purposes, so the numbers are a bit low.

Edit:
To add to that - 2793 damage from single shadowflame stack, 5359 from two stacks, 7926 from three stacks(tarecgosa's proc).

Yep, looks like I had the wrong base multiplier for metamorphosis - it was reading the wrong spell data entry, resulting in something much higher than the 10% base benefit. You should see updated results in 20 minutes or so, with much lower totals for demo.

As for shadowflame, with almost no gear on I saw something like 347 with single-stack shadowflame and 472 with two stacks. Very odd.

Drain soul is currently 20% and below, down from 25%. And the doom guard also is setup to be an execute under 20% as well.

Also I have a feeling that under 20% when we are using drain soul that using soulburn Soul Fire and Soul Swap will be worth it, but that is just speculation.

Yeah that drain soul threshold was just an action list issue - the spell wasn't actually simulating as doing extra damage until 20% and below. I've fixed the profile now.

The doomguard execute is modeled, but trying to delay the summon isn't optimal given our default fight length which means there's often room for a second doomguard summon at the end.

Turns out the shadowflame stack multiplier is only applied to the spell power scaled part of the damage, which explains why my numbers were so far from being double. I've implemented the mechanic in simc, and the results have updated to reflect that.

Also, testing reveals the corruption spell power coefficient is 0.3 per tick, not the 0.2 indicated in the tooltip. This is also reflected in the sim now.

Demonology suggestions:
-Use the Felguard, he does equal damage to other melee pets but his Felstorm sets him apart by a significant amount.
-Grimoire:Service pet melee damage doesn't seem to scale in this build so it should use the Imp. If you aren't accounting for this bug, the Felguard should be used there too.
-Use glyph of Imp Swarm on cooldown and in sync with Dark Soul. This glyph still allows Demonic Calling to be up when Imp Swarm is not on cooldown. So a fight start should get 2-3 Imps from DC before getting to 1000f and using all other CDs. If gear limits getting from 200 to 1000 in under a minute on the pull then using Imp Swarm immediately could work if you delay all the other CDs a few seconds until it comes back up too.
-Hold Dark Soul and Grimoire of Service until you are in Meta. The first time will be easy since we will get to 1000f and then use CDs, but we will need a trigger for the situation where you are already in Meta and the CDs come up. A trigger that is based on how much fury is left, the cost of Imp Swarm and the fury generated by the Imps and Grimoire.
-I'm not sure why the multiplier of 8 for the last half of line action 8. Assuming we will be casting Soul Fires and generating fury through pet attacks only (other sources could possibly generate, but I'm not assuming) I think it needs to be much higher depending on haste. At 10% haste something like time_to_die*26 and at 25% haste something like *32.
-Doomguard isn't used.

I understand delaying development of some of these things since it's a bit of work and some significant changes are probably still incoming, but this is a start for what is going on in this build.

edit:
Glyph of Shadow Bolt should give a higher uptime on procs.
Glyph of Hand of Gul'dan seems to give a 1 second ICD on HoG instead of 1.5. (I would like to test and research this further, but my few logs have shown this and I think it was on a tooltip as well.)

Delaying cooldowns until you are in meta is a loss - meta is simply not a great enough buff for it to be worth the lower total number of cooldown activations throughout the fight. I've yet to implement glyph of imp swarm, and there's a bug in the felstorm implementation - I'll try both of those once I can. I'll take a look at other possible values for that constant 8. Touch of chaos is used automatically whenever in meta, it doesn't need to be activated through the action list.

But you're right, I completely forgot to use the doomguard in the demo profile. Adding that should bring demo right up there with affliction.

EDIT: I fixed the felstorm bug, but the imp is still very slightly better than the felguard.

After using fire and brimstone+immolate it ticks twice more damage (depends on mastery amount). Fel flame does not reset this damage bonus.
I do not have simcraft for MoP so can you try this trick? Will it be increase in dps using fel flame to maintain immolate after fire and brimstone?

My dummy tests show the Imp as the lowest pet because of his inability to melee and in return not scaling with haste. All pets should be above him and the Felguard even more so with Felstorm. I'll investigate further to see where my discrepancy might be.

Keep in mind is that you're probably testing on a level 85 dummy, which has less armor than level 88 bosses. Raid buffs will also affect the imp differently than the felguard.

When simulating with target_leveL=85 and optimal_raid=0 I do show the felguard as significantly better than the imp.

EDIT: Turns out I had pet energy regen scaling with haste, which I guess it doesn't in-game. That certainly puts the melee pets over the top, even when raid-buffed against a level 88 target.

Is Immolation Aura modeled? It seems that using it (and potentially even still dropping out of Meta) could be a dps increase depending on the difference between Meta phase damage and Caster phase damage. It would be an obvious part of the rotation if the cost wasn't so high, but perhaps there is a situation where is would be beneficial. I don't want to suggest that it would be part of the optimized rotation, but it is a piece of the puzzle that I'd play around with by doing tests like:
Getting to 1000f then using it.
Getting to 500f, using it and a slash, then dropping Meta.
Lining it up with Dark Soul.