SimulationCraft for Warlocks (Cataclysm Edition) - Page 33

daia · 07/02/11, 9:29 PM

The calculations for the Doomguard are incorrect.

It snapshots mastery, but never does anything with it.

From my testing, a fair estimate is increasing the base damage of the attack by 3 per mastery. A more accurate calculation is just the increase in damage from the Mastery. Only Demonology receives a benefit from the mastery.

Also, the cast speed does increase with haste from gear. This is also not reflected in Simcraft.

Meerien · 07/03/11, 12:33 AM

Originally Posted by daia

Summoning the Doomguard during the Demon Soul: Felguard does increase its damage and cast speed, providing a significant increase in DPS.

This is not true. Doomguard does not recieve any benefit from any of the Demon Souls.

I've been trying to create a list of things that do and do not affect Doomguard's damage output, from what I can tell, it snapshots Mastery, Spellpower, Crit % (including raid buffs) and Hit % (I only noticed this last one when I summoned him whilst under Chimeron's Caustic Slime and he missed 70% of his doombolts).

He scales with haste, but only off of the rating on your gear. I would test with Haste rating procs but I don't have a trinket with a haste proc on it to test with (if anybody wants to do that for me it would be great). He's NOT affected by raid haste buffs (Including Heroism), or % damage increase buffs (Engulfing Magic). However he IS capable of receiving helpful buffs, such as Omnotron's Power Generator (so long as he stands in it) or Sinestra's Essence of the Red.

On the note of the Doomguard, his dps contribution to a 3-4 minute fight is so significant that I can't imagine selecting a trinket other than one with a huge Spellpower, Intellect, or Mastery proc specifically just to prop him up.

daia · 07/03/11, 1:03 AM

You're absolutely correct. I retried it and definitely did not get an increase in damage. The increase in casts is also questionable because of how wonky the Doomguard acts on training dummies in general.

Meerien · 07/03/11, 1:53 AM

Originally Posted by daia

You're absolutely correct. I retried it and definitely did not get an increase in damage. The increase in casts is also questionable because of how wonky the Doomguard acts on training dummies in general.

I would love to be able to get better data regarding Doomguard's haste scaling but it is really hard because of how Doomguards always just sort of stand around for 1-4 seconds when you first summon them : /

Jmickey · 07/03/11, 2:44 AM

The following items don't exist in game as heroic items and are included as such in the BiS simulations:

- [Rune of Zeth]
- [Rippling Flamewrath Drape]
- [Crystalline Brimstone Ring]

bmillz1341 · 07/03/11, 3:27 AM

Is there any way to specifically set up a sim for demo using Moonwell chalice at the same time as Meta throughout the fight? I imagine it would be hard to set up a sim for it, but just thinking about it I can't imagine any other trinket in the game being better for Demo. Also considering that the Doomguard is affected by mastery procs I shudder to think about the things that are possible if you were to line up a proc from Heroic Theralion's Mirror with this trinket and a SP potion prior to summoning a Doomguard.

• Zakalwe · 07/03/11, 4:26 AM

Originally Posted by daia

The calculations for the Doomguard are incorrect.

It snapshots mastery, but never does anything with it.

From my testing, a fair estimate is increasing the base damage of the attack by 3 per mastery. A more accurate calculation is just the increase in damage from the Mastery. Only Demonology receives a benefit from the mastery.

Also, the cast speed does increase with haste from gear. This is also not reflected in Simcraft.

We do model him scaling with mastery. Not by "3 base damage per mastery", but by 2% per point of mastery, to match the benefit other pets get. If that is incorrect, it would be helpful to have exact numbers from your testing.

As for cast speed, last time I looked through some WoL parses I saw a pretty consistent time between casts of 3.0, which is also the base cast time of his spell. If he somehow benefits from haste there's something else slowing him down as well. Again, exact numbers from your testing would be helpful.

Originally Posted by bmillz1341

Is there any way to specifically set up a sim for demo using Moonwell chalice at the same time as Meta throughout the fight? I imagine it would be hard to set up a sim for it, but just thinking about it I can't imagine any other trinket in the game being better for Demo. Also considering that the Doomguard is affected by mastery procs I shudder to think about the things that are possible if you were to line up a proc from Heroic Theralion's Mirror with this trinket and a SP potion prior to summoning a Doomguard.

We already do line up meta with Moonwell Chalice in the trinket test in the OP. For Theralion's Mirror, it seems to be a 240 DPS gain to line it up with the Doomguard, so I'll put that in there.

Zantamyr · 07/03/11, 5:11 AM

Why "T12 normal" profile doesn't exist?

Meerien · 07/03/11, 8:58 AM

Originally Posted by Zakalwe

We do model him scaling with mastery. Not by "3 base damage per mastery", but by 2% per point of mastery, to match the benefit other pets get. If that is incorrect, it would be helpful to have exact numbers from your testing.

As for cast speed, last time I looked through some WoL parses I saw a pretty consistent time between casts of 3.0, which is also the base cast time of his spell. If he somehow benefits from haste there's something else slowing him down as well. Again, exact numbers from your testing would be helpful.

World of Logs - Real Time Raid Analysis
For starters, you can see 60.2 active seconds, and 21 Doom Bolts, despite him getting a fair bit of pushback from Lava Spew. Going into the combat log, I see

[19:32:10.202] Doomguard begins to cast Doom Bolt
[19:32:13.019] Doomguard begins to cast Doom Bolt
2.9 second delay between Doom Bolts

[19:32:15.814] Doomguard begins to cast Doom Bolt
[19:32:15.932] Magmaw Lava Spew Doomguard 17743 (R: 4436)
[19:32:17.550] Magmaw Lava Spew Doomguard 18021 (R: 4505)
[19:32:19.032] Doomguard begins to cast Doom Bolt
Two lava spews causing pushback, 3.2 second delay between Doom Bolts

[19:32:25.059] Doomguard begins to cast Doom Bolt
[19:32:27.903] Doomguard begins to cast Doom Bolt
2.84 second delay between Doom Bolts
Etc Etc.

I am fairly sure the reason I see a variable number between 2.8 and 2.9 seconds between him starting new Doombolts is because the Doomguard has a fondness for standing around doing nothing for short periods of time (I have no idea what else could explain the variation). Either way, Something is speeding up those Doom Bolt casts, if it's not my haste present on my gear than I'm at a bit of a loss for what it might be.

• Zakalwe · 07/03/11, 10:43 AM

The combat log is notoriously inaccurate due to various types of lag, so you can't really look at the time between individual entries and draw conclusions from them like that. And the "active time" metric doesn't include the cast time of the first doom bolt.

Try pasting this query set in the log browser:

[{"eventTypes": [6], "sourceNames": ["Doomguard"]}, {"eventTypes": [1], "targetNames": ["Doomguard"]}]

You'll see that the time between the start of his first cast and the start of his last (21st) cast is 60.268 seconds. In this time he was able to fit 20 doombolt casts, so an average cast time of 3.0134 seconds. And I do not believe he is affected by spell pushback at all - you'll see cases like this:

[19:31:36.846] Doomguard begins to cast Doom Bolt
[19:31:40.070] Doomguard begins to cast Doom Bolt

There's no damage incoming to the Doomguard between these two entries, and he still took 3.224 seconds between the two casts. And there are examples of the opposite occurring too:

[19:31:48.916] Doomguard begins to cast Doom Bolt
[19:31:49.849] Magmaw Lava Spew Doomguard 20103 (R: 2234)
[19:31:51.718] Doomguard begins to cast Doom Bolt

Here the time between casts is 2.802 seconds despite incoming damage. So it's pretty clear that the variance you're seeing is just normal combat log inaccuracies.

Tokens · 07/03/11, 5:44 PM

About the Doomguard.

I've said some of these things in previous posts;

1. It's a significant DPS increase removing the 'if.Metamorphosis.down' variable from the Doomguard and giving it higher priority (i.e using it after Immolate or Corruption).
2. Doomguard scales from hasterating on gear, but at our hastelevels we won't really notice it as it has a really clunky AI like other Guardians in-game (has problems deciding between BoA and BoD-targets, latency issues etc). 20- 21 casts as Demo is usually what you'll get out of it, despite the casttime being 2.5s with 20.50% hasterating from gear. I used mine with 4900 haste rating (haste elixir, haste pot, heart of ignacious) and got 3 extra casts of as Destruction on the dummies to test.
3. The Doomguard scaling in SimCraft is wrong and the Doomguard is a lot higher in reality. Demonology locks on Chimaeron will back this up, but I'll try and find some conditions in-game that you can compare with in the simulation.

daia · 07/03/11, 6:08 PM

Just looking at the simcraft results you can see the Doomguard calcs are incorrect.

Affliction 12706 hit, Demo 14774 hit. In my gear, which is inferior to T12 setups, the Doomguard should be getting 34% more damage from the mastery alone, not to mention more SP from DP talent.

Additionally, without any raids buffs, I can get 30k average hits on a training dummy as Demonology by utilizing it when everything procs at the beginning.

Just summoning a Doomguard right now, with no procs or debuffs up as Demo with 34% mastery bonus, the Doomguard hits for 20500~ with a 2.7 sec cast time, which is consistent with what you would expect it to round to with base 13% haste.

Any extra delay you see between casts is probably something to do with the AI being poor like the Imp with chain casting and no Dark Arts.

Meerien · 07/03/11, 6:50 PM

Originally Posted by Zakalwe

The combat log is notoriously inaccurate due to various types of lag, so you can't really look at the time between individual entries and draw conclusions from them like that. And the "active time" metric doesn't include the cast time of the first doom bolt.
.....

Hmm, how disappointing you are then, WoL.

Perhaps I made some wrong observations, either way I really just wanted to show that the doomguard has the capacity to fire faster than 3.0 seconds.

• Zakalwe · 07/03/11, 8:53 PM

We do model the mastery scaling correctly - the fact that his damage for demo is less ahead of affliction than you'd think is that affliction pops him at the beginning when many procs tend to be up, while demo waits until after metamorphosis, ie. when no procs are up. So the demo profile should be changed to not wait so long, but the modeling isn't wrong.

What is disturbing is that you're apparently seeing damage numbers without buffs or procs that are way higher than the sim produces with full raid buffs. This indicates that either he somehow scales much more with mastery than he's supposed to, or the base damage and/or scaling of his spell is off. We grab our values directly from the WoW client spell database, but it's been known to be wrong in the past.

In order to confirm/deny that last theory, I would need logs of the doomguard being summoned with varying levels of spell power (one time naked and one time fully geared should suffice), as well as in different specs. Along with the logs I'd obviously need information about the exact spell power and mastery rating used for each test. I'll eventually get around to doing this myself if no one else does, but it may take a week or more as I'm extremely busy with work right now.

As for his casting speed scaling with haste, logs would help convince me there too, but for now I'm satisfied that his number of casts in the sim results matches what I see in WoL logs quite well.

daia · 07/04/11, 1:53 AM

All tests done with an Orc.

168 SP
0 Mastery

Demo
Min 2092 Avg 2172 Max 2261
Min 4323 Avg 4371 Max 4419

Afflic
Min 1401 Avg 1472 Max 1508
Min 2947 Avg 2968 Max 2989

505 SP
0 Mastery

Demo
Min 2252 Avg 2322 Max 2413

Afflic
Min 1666 Avg 1702 Max 1754
Min 3528 Avg 3528 Max 3528

168 SP
321 mastery rating (+3.58%)

Demo
Min 2256 Avg 2359 Max 2435
Min 4585 Avg 4700 Max 4820

185 SP
Pet out for Demonic Pact

Demo
Min 2116 Avg 2198 Max 2284
Min 4320 Avg 4405 Max 4550

15123 SP
1680 Mastery

Demo

Min 29064 Avg 29176 Max 29307
Min 58107 Avg 58225 Max 58342

10985 SP

Afflic
Min 9153 Avg 9213 Max 9252
Min 18318 Avg 18363 Max 18422

As you can see, in real gear, it scales up to a ridiculous amount. None of the tests have any debuffs in play.

As for the spell cast speed, it's very very easy to see yourself. Just create a Doomguard and look at his cast time. It undoubtedly scales with your own spell haste. The problem is, he doesn't chain cast as well as he should. In a real environment, a 3s cast Doomguard will get way less than 20 casts in. It just so happens that with most people closer to 2.5s cast on theirs, he will get 19-21 casts off on average.