Berthold

Using my own experience from warrior and DK tanking, as well as data from wws logs, 1.9 s (little bit higher than GCD) seems like a good estimation for non auto attacks that can cause parryhaste. I also tested with somewhat higher and smaller values, but it did not make any significant difference.

Concerning the class comparison, you are absolutely right that paladins will have no hard time dealing with parry haste, because they do not attack (physically) that often.

 

dukes

In terms of a frost DK:
- Icy Touch which is a spell;
- Howling Blast which is also a spell;
- Frost Strike which is unparryable;
- Rune Strike which converts mainhand hits to be unparryable;
- Pestilence which is a spell (although not used on single targets, obviously).

You're saying that warrior / DK tanks have a comparable number of attacks that are parryable? I'm not quite sure where you get that from, especially considering the difference in weapons (one using a slow 2h, one using a fast 1h) and the fact that Heroic Strike can still be parried while Rune Strike cannot.

Could you provide rates of parryable attacks that you have used for Warriors / DK's as I would be interested in seeing what your assumptions were in producing those graphs.

 

Berthold

Although I did not intend to compare the difference in classes, you (dukes) are absolutely right, that there are differences in the amount of special attacks for different classes.

Ok.. let's get some numbers.

Frost DK (20 seconds); (3.6 mace)
IT,PS,BS,BS,OB,IT,PS,OB,OB, (and some irrelevant FS) and let's say four (2.0 boss swing timer, 40% combined doge/parry of the tank) additional Rune Strikes (reducing the number of parryable attacks). -> 7 special attacks, 2 white attacks.

Warrior (20 seconds) (1.5 mace);
I am not sure about sunder armor, but every other attack should be parryable. Lets assume it is. This results in:
13 parryable special attacks, 13 auto attack / heroic strike.

Ok. kinda different :-)
Especially because DW Frost Tanking uses HB instead of OB, the DW tank should produce 3 less parryable special attacks, but a lot more white attacks. Even with two Broken Promises we have 16-4 = 12 parryable attacks.

So we have
Warrior 26 parryable attacks
DK 2h 9 parryable attacks
DK DW 16 parryable attacks.
I am still sure that DW will be worse than 2H tanking (including the benefits of two tanking weapons), but I will change the simulator as soon as I got spare time.
On the warrior / class comparison side, I assume that Blizzard did a good job and blocking outweighs the higher parrygib chance. Maybe I will research on that detail a bit more some time. But for really meaningful data, I would have to include Shield Block/Wall, Ice-Bound Fortitude, etc. So this is not very likely.


Thanks again for pointing that out dukes.

 

Berthold

Ok. updated it:




conclusions:
still DW is a bad idea.
Warrior Tanking (ignoring block) performs worse.

 

Berthold

For anyone (=C programmer) interested in playing around with the simulator, here ya go.

http://www.iseurope.de/theheap/wow/T...ParryHaste.zip

its wrapped in an Objective-C Project, but the code itself is pure C - it should compile with any gcc.

hf

 

• GravityDK

Thanks Berthold, great contribution.

Could you, or anyone with the smarts to understand the code, answer these questions about the simulator:
- how does it determine 'wipe probability', does it assign weightings to probability of occurances such as when the healers cannot catch up with the incoming damage because it all came within one second, or assume the healers produce only a constant HPS and therefore any spiky breaches of incoming damage above that threshold produce a wipe probability?
- does it account for the reduced likelihood of dodging a parrygib due to the extra 1% or so avoidance that DW tank has thanks to the 1H itemisation?

I presume this statement are true:
- it does not account for reduced Death Strike self-healing, and probably doesn't account for DS at all

 

Berthold

As mentioned in my first post, it does no statistical probability calculations, but more of a "brute" force approach.
It simulates a real boss fight in 0.1 second steps:
1)Weaponswings are updated
2)if swingtime is over, the hit(MT and boss) is calculated (hit/parry/avoid). If hit-> deal damage, if parry -> speed up the others next attack
3) if hitpoints of MT are 0 or below-> increase number of wipes by 1.

4)every n-th 0.1 second step a heal is cast (depending on kind of heal). Increase current MT HP. Has a 10% chance, that the heal is not cast due to brainlag, out of range, AE stun/fear effects etc.
5) go to 1

This boss "try" is done until the MT is dead(increase number of wipes by 1) or 6 minutes are over(increase number of kills).
We now repeat this 10000 times to make sure we got a statistically stable number of wipes and kills.
The overall "wipe possibility" is then
#wipes/ (#wipes+#kills)


yep

yep

 

T1mm4h

To be honest, thought I never saw the original thread/quote, but think of it this way: Parry-haste used to or is a regular mechanic on WoW, thus instead of setting the flag as on for each individual mob they want with it (as in "X mob benefits from parry-haste"), they just flag mobs/bosses they want to NOT have the parry-haste e.g. X mob is flagged as not having parry-haste.

Hopefully that clarifies it a bit. No sense in Blizzard wasting time setting flags or even scripting something as flagged for a particular property when they can just put it in globally and divvy out what they don't want using it.

Also, I believe player parry-haste is still working because quartz seemed to reset when I've noticed it before. I haven't parsed anything however, is anyone able to verify this?

 

◊ Quigon

If you're really concerned with the parry haste flag and how it is implemented, there was another blue post concerning it in the days of the Mother Shahraz fiascos - when it was of large concern to the tanking community. If you hunt that down perhaps it would give more detail (and perhaps nothing at all) - but the idea was they would create such a flag, and I believe Tigole mentioned that the flag would disable parry haste for that particular encounter. This would imply that the haste was still native.

 

humbe

I did some math on how much extra damage I expect to take due to parry-haste effect in order to weight how good expertise was for druid mitigation.

The calculations in detail are here: World of Warcraft - Evaluating druid tank gear

I hope I haven't miscalculated anything, but as it looks to me, like even with no expertise (but 10 expertise talent), a druid will only take around 4.3% extra damage on average, given 16% parry chance for boss. Druids don't hit that often though, so might be more for other tanks. Especially a duel wielding one.

I doesn't sound like that bad burst damage to me. There still has to be 60% of base swingtime between hits. A streak of non-dodged attacks sounds like a much bigger burst. Anyways, unless I got something wrong, I don't see expertise as a strong mitigation stat for druids.

 

Elimbras

Take care that if you want to compare the survavibility of tanks, you should also take into account talents. With the same available gear, not all classes have the same stats.
Take also care that somes classes (read warriors) have more (and good) panic buttons for tanking. Since we are interested in the occurence of rare events, this can have an impact.

 

Kethas

So when guessing which bosses Blizzard would disable parry haste for - surely they'd throw a bone to the folks working on the hardest encounter ingame, where the "main" boss's melee damage isn't even a major part of the fight, right?

Wow Web Stats

In particular, try 10, filtering for actions performed by Sartharion:

So. A 0.50 second swing timer. Interesting. Let's see who attacked in that time (excerpt):

Conclusion: in Sarth10+3 (and, while not confirmed, presumably in Sarth25), Sartharion has parry haste enabled.