SimulationCraft for Warlocks (Panda Edition) - Page 21

Annybaal · 08/18/12, 4:07 PM

Deleted

• Zakalwe · 08/18/12, 4:17 PM

Originally Posted by plopinou

In the "Stats details: corruption" unfolded section we have "Direct Results : 626.28". From what I understand, this is the MG + DS average corr ticks count from all the iterations done by the sim.
In theory it should be equivalent to the number of time the MG and DS ticked.
But in this same table, MG + DS averages out to 577,5.

What did I miss ?

You can also see that on 2500 iterations, the average direct crit corr damage should be twice the hit, but it's actually a lot more than that in the table.

The same thing are present for UA and agony.

That part is a reporting issue - it's counting the regular dot applications as direct results with 0 damage.

plopinou · 08/18/12, 4:24 PM

Ok, that makes sense.
The only remaining issues I see is the haunt spell damage, and the nightfall in-game average proc.
When the servers are back up I'm gonna do some more 15mn reports just to make sure I wasn't simply extremely not lucky.

• Zakalwe · 08/18/12, 5:10 PM

It's certainly possible they reduced it to 0.4% per tick - that would cause an average over time of almost exactly 5%, so it would make more sense given the tooltip. And 5.0% is within your 95% confidence interval.

Solipsist · 08/18/12, 10:00 PM

EDIT: nevermind, I just answered my own concern. Sorry about that.

whi · 08/19/12, 5:49 AM

Anyone seeing a reduction in damage of certain demo spells in current beta build? Or the bug that GC talked about yesterday? Hotfixes? As there was nothing about any relevant changes in yesterday's datamining.

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

ZumoO · 08/19/12, 1:28 PM

(The WoL has been changed somehow 30 min after I uploaded it, I'm fixing the numbers O_O)
(Done)

I have tried to do the same check as plopinou, but for Destruction.
I have the same stuff and final stats as in the dummy simulation, except -2 Int (orc).
I respected the rotation of the simulation, mainly meaning no pooling of embers to use them with cooldowns.

WoL link : World of Logs - Real Time Raid Analysis

Simulation / Ingame

Immolate initial hit
hits : 10207 / 10218
crits : 21977 / 21937
execs : 53 / 62
I can't explain why I had to cast it 9 more times. I usually refreshed it with 7-8 sec left, bumping it to 23-24 sec left.

Immolate ticks
hits : 10191 / 10300
crits : 21948 / 21487
execs : 355 / 364

Incinerate
hits : 38290 / 38129
crits : 81910 / 83072
execs : 420 / 413

Conflagrate
hits : 41307 / 41368
crits : 89222 / 89693
execs : 81.7 / 83

Chaos Bolt
crits : 197025 / 194916
execs : 46.9 / 45

Shadowburn (I did only one crit, not accurate at all)
hits : 128638 / 116577
crits : 272221 / 323065
execs : 13 / 14

Total without pets
Simulation : 40803
Ingame : 39971

Taking in account lag and poor luck with crits, the simulation's damage seem really pretty accurate.
Still, I don't know why I casted 9 more Immolates.

As I am an orc, pets damage are less accurate, but it is not a great deal. I will summarize them anyway :

Doomguard : Doombolt
hits : 21673 / 22562
crits : 47899 / 47449
execs : 36 / 30
Same remark as plopinou, the Doomguard casted 15 bolts each time, not 18.

Main Succubus : Melee (Creatures => Elerna in the WoL)
hits : 4329 / 5692
crits : 9279 / 11777
execs hits : 301 / 214
execs crits : 113 / 75
execs glances : 135 / 91
??

Main Succubus : Lash of Pain
hits : 21442 / 21962
crits : 46136 / 46380
execs hits : 134 / 136
execs crits : 34 / 46

Service Succubus : Melee
hits : 5168 / 7024
crits : 10589 / 13972
execs hits : 30 / 29
execs crits : 42 / 17
execs glances : 24 / 29
Very strange results, both from damage and number of executions. The 2% from Orc can't explain the gap. Executions count seem even more weird since my Service Succubus didn't do any Whiplash.

Service Succubus : Lash of Pain
hits : 26293 / 26602 (pretty accurate without the 2% from Orc)
crits : 53250 / 53248
execs hits : 23 / 24
execs crits : 18 / 17
Seems pretty good for LoP.

Pets summary
Succubus : 7520 / 7677 (+2%)
Service succubus : 2481 / 2175
Doomguard : 1017 / 984

There seems to be some discrepancy for pets results between the sim and ingame, or something is happening.
Executions counts are pretty weird might be a report problem ?
Considering I'm an orc ingame they should hit a bit harder, but their melee hit for much more in reality, but they do much less of them.

plopinou · 08/19/12, 3:53 PM

5 logs from today (build 16004) :
World of Logs - Real Time Raid Analysis
World of Logs - Real Time Raid Analysis
World of Logs - Real Time Raid Analysis
World of Logs - Real Time Raid Analysis
World of Logs - Real Time Raid Analysis

5783 corruption ticks (MG's + DS's + normal ones)
285 nightfall proc, or a 4.92% average
So yes, it's probably now at a 0.4% stack-able chance to proc.

Also, 10 doomguard were used, all at full time dps, and there were in grand total 153 dom bolt, or 15.3 per summon.
I have ~300 ping on the beta, so maybe it has some influence.

Also, haunt bonus damage is still at 25% on the beta, with the tooltip saying 20%.

• Zakalwe · 08/19/12, 4:25 PM

I'll go ahead and update nightfall to 0.4% per tick in the sim.

I have no idea why your doomguard is casting so few bolts - it would mean it spends 4 seconds per doom bolt instead of the expected 2.7 from the haste you have. I've heard some reports that pets are behaving erratically on the raid dummies - did you observe anything like that?

In my independent tests conducted at level 85 I've had the doomguard get off anywhere from 18 to 21 bolts depending on my haste. He's certainly lagging between casts a bit, but rarely exceeding a full second.

netsach · 08/20/12, 2:52 AM

Originally Posted by Zakalwe

I have no idea why your doomguard is casting so few bolts - it would mean it spends 4 seconds per doom bolt instead of the expected 2.7 from the haste you have. I've heard some reports that pets are behaving erratically on the raid dummies - did you observe anything like that?

It's not due to test dummies (as doomguards are rangesd caster and thus don't suffer from the "start/stop/move erratic behaviour). But in fact, the problem is that the central zone in Shattrath pulses "special buffs" to players periodically, and pulses "special debuffs" periodically to creatures.

And sadly, all our pets (permanent pets, wild imps, and guardians), are considered creatures in that case, and do get the following debuffs while in shattrath (edit : i'll post a bug on the eu beta forums to report this) :
1) Reduced cast time by 30% debuff
2) Reduced physical damage dealt by 10%
3) Increase spell damge taken by 5%
4) Increase physical damage taken by 4%
5) Reduce Armor by 12%

Thus 1) explains the low number of DG casts, and 2) might explains the difference between pets and sim (i haven't checked)

whi · 08/20/12, 4:32 AM

Did some sanity checks for spell coefficients after seeing a drop in damage of some of demo's spells after last build, here's the list(sim ones taken from 'results' page/current beta; per tick for dots):

corruption: 0.2/0.2
hand_of_guldan: can't find one/0.5, ~535 base
melee: 0.1/0.0831, ~89 base
shadow_bolt: 1.447/1.2, ~1283 base
shadowflame: 0.137/0.137
soul_fire: 0.85(?)/1.4, ~1491 base --- 0.7, ~745 base
touch_of_chaos: 0.79/0.666, ~708 base
doom: 1.25(?)/1, ~1068 baste

Edit: fixed soul fire coefficient.

• Zakalwe · 08/20/12, 4:44 AM

Doom was 1.25, yes, prior to these hidden nerfs. But for the soul fire coefficient, are you taking into account that it's all crits, so it's all double damage? If you weren't, I guess the coefficient was nerfed from 0.85 to 0.7? That would make a lot more sense than it being buffed by that much.

whi · 08/20/12, 4:53 AM

Well that solves that mystery, I didn't.

Unknowndiv · 08/20/12, 7:10 AM

Not sure this worth mentioning, but I was interested in glyph of shadow bolt and demonic fury gains.
As you know when shadow bolt *hits* the target player gains 25 demonic fury. With glyph of SB there are 3 *hits*, so how does it work?
I've tested it without gear to get some *misses* on the target (I know we should have hit cap, still..).
Situation 1 : all 3 SB hit the target - player gains 25 demonic fury.
Situation 2 : 1st SB *misses* the target - player does not gain any fury. Player does not do any damage at all! If 1st SB misses, there will be no other SBs. Just one lonely little skull heading towards the target. Maybe that's all 3 SBs *miss* the target, combat log shows only 1 line. And I did not see any log with "*miss*+*hit*/*miss*+*miss*/*hit*" lines.
Situation 3 : 2nd SB *misses* the target - player does not gain any fury, but do some damage, atleast 1/3 (1st SB) of unglyphed SB damage.
Situation 4 : 3rd SB *misses* the target - player gains 25 demonic fury.

By "1st, 2nd and 3rd" SBs I mean order in combatlog. These shadow bolts do not hit simultaneously.

realCool · 08/20/12, 7:30 AM

While looking at the destruction priority list for the aoe 8 targets simulation I found the following error in the sample sequence:

79ANOTTP

7, 9, A and N are lines for the activation of CD's so thats ok. O is Rain of Fire that's also ok. But the next is T which is cycling of immolate and only after that comes P which actives fire and brimstone. As we are starting the fight with 1 ember should not P be directly after O instead of cycling immolate first manually?