kalbear

Correct. But it's the first few seconds that might be problematic on snap aggro, which is why I suggested it.

Feral tanks already have the highest overall physical mitigation and close to the highest spell mitigation. They don't need to catch up to anyone currently.

But it doesn't matter, because PvP gear doesn't work like that; it only is supposed to affect damage from players. So aside from the odd mind control situations, it really isn't an issue.

 

Oltsu

The thing is, full nuke straight from the pull is completely valid with a paladin tank, and atleast with whatever-spec-was-fotm-in-naxx death knights. Be it an aoe pack or a single mob, there's no reason to wait for anything. Of the top of my head, the threat figures seen from a paladin over the first 2-3 seconds of a pull seem to ranging from 2 to 4 times the threat I can push out. As I mostly run with a paladin tank, this more or less results in a dead shaman or two if my target happens to be the first one to kill. (Not counting rogues, getting those killed is one of the few perks in tanking.) It's annoying for me not being able to tank smoothly, and for the rest of the raid not being able to faceroll freely on trash. You can argue that the dps should learn to control their damage, but trash and farm bosses are tedious enough as it is without an angry bear yelling on Ventrilo about aggro management. Though at times entertaining, I'd rather I didn't have to do it all the time.

As a side note, has anyone tried logging threat values, or are there any spesific addons available suitable for it? It'd be interesting to see some graphs between different tanking classes on identical pulls.

 

Helistar

This is the big problem IMHO. Blizzard is certainly altering the damage/heal/timestamps to ensure robustness against lag, probably "waiting" some time before declaring some damage really inflicted (or a tank really dead), in case a heal with a timestamp in the past arrives at the server. So there's probably some quite complex heuristic which plays fast and furious with the healing to provide a picture which is as consistent as possible while correcting as much as possible for any lag, and this weird heal events may be a result of the heuristic. Without knowing exactly how it works it's near to impossible to "reverse engineer" the events.

I've been looking at using an addon to store the tank's health value directly. The advantage: I get an actual number and not just the variations, so keeping in sync is automatic. The disadvantage: I lose the information on the origin of the events, I can see the health going up or down, but there's no (easy) way of knowing the cause for it. It also risks being very user-unfriendly since the addons cannot use the file IO of lua, so the only possibility is to store everything in a table and have it saved as a persistent variable at logout time. I don't know if this is feasible or it runs into memory/lua constraints issues. Considering the amount of data auctioneer can keep, I guess it should work. I'm learning about addons and during the holidays I'll try to see what I can come up with.

 

Latas

Doesn't grim reaper allow you to review the health of someone directly in addition to the changes and what caused them?

 

Pleco80

I have a question regarding the Improved Mangle talent, and theat rotation.

I used to have 3/3 Improved Mangle, as it seemed great for threat. Then I read up, look at simulation tools, etc, and tried 0/3 Imrpoved Mangle. It has worked out for me pretty well.

I do however find that often I have to wait for my Mangle CD and waste a small amount of time on that. This seems to happen more often when I use FFF between my Mangles. (which is almost always, since their CDs are both 6 sec). It's almost like FFF uses up less than the 1.5 sec GCD? Probably because it is affected by Haste where my melee GCDs are not. (not that I stack haste) Does anyone else find they are waiting a fraction of a second for Mangle?

Upon noticing my wasted time in my threat rotation, I am debating trying 1/3 Improved Mangle, in order to eliminate the time I am waiting for Mangle CD. Any opinions on this? As I consider it, I think it would probably be a waste, since then Mange and FFF would no longer sync up, which gets back into the potential problems with 3/3 improved mangle.

(BTW, if anyone is looking at my spec, I am currently in a higher threat spec, not a general tanking spec, I know it is not the optimal general tank spec)

 

kalbear

FF has a 1.0 GCD.

The primary problem is as you said - while you'd not waste that .5 seconds some of the time, you'd still be waiting for FF to come off CD.

 

Supergreat

Can someone please correct me if I am wrong, but I thought the general school of thought on Improved Mangle for bears was to take all or nothing, as taking partial points in it does not allow you to overcome the global cooldown (thus making them wasted talent points)

I used to use 3/3 in Imp Mangle but now use the points elsewhere. With 0/3 in Imp Mangle I find there are less conflicts with FFF and mangle being ready at the same time.

 

kalbear

That's basically correct. The 'advantage' - if you want to call it that, is that FF is a 1.0 second GCD, which means that if you had 1/3 mangle, you could do something like this without any wasted seconds:
0.0 Mangle
1.5 FF
2.5 Lacerate
4.0 Lacerate
5.5 Mangle

The problem is that you still have to wait 6 seconds for the CD for FF to come off, so you'll end up wasting the time anyway.

Imp mangle sounds like it's a really great deal, and in low-rage situations it is. But Maul simply is far too dominant as far as bear threat goes, so even strong threat moves aren't particularly important comparatively.

 

Mooka

Hello all! I just had a quick question, I've searched google and also tried to find it on this forum but the link someone posted eludes me! All I am looking for is the link to the site where druids are ranked based on their gear taken from the armory.

I wanted to compare/contrast my tanking gear to some of the top geared druids. (I know, my stupid blue trinket makes me look 'nubbish'... dang [Heart of Iron] not dropping! =/

Thanks in advance all!

 

Labrynth

WoW Heroes - World of Warcraft PvE character info & ratings is what you are looking for most likely.

 

Boevis

That site is utterly horrible, creating a ranking based on ilevel and the skill level required for you enchants is just a stupid way of comparing gear, especially for druids who are very likely to use lower level idols, weapons, and enchants simply because of poor itemization or unlucky drops. It also limits you to 1v1 comparisons. It would be very cool to see a site that basically imported every druid with a Feral Spec to Rawr in order to convert to bearform accurately and allowed you to browse everyone's unbuffed HP, Armor, Avoidance, Hit, Crit, Expertise, and AP. I remember something like that being part of CTProfiles (or whatever it was called) a 3rd party character listing that was widely used pre-armory.

 

goodolarchie

Do any of you bear tanks have parses from Ulduar fights where you've hit 58-60% effective miss rate? I don't care which fight, I'd just like to compare a few things.

 

Kestre

I generally use Berserk as soon as I get hit by the boss (ie whenever I have enough rage). I have maul macroed to mangle - which works fine in any non-rage starved case. Berserk seems to work very well for aggro in the beginning of a boss fight.

 

Delorn

On my druid I've been having a problem of not being able to build up rage after a few minutes in a boss fight and a loss of aggro where adds will jump off me and run to the dps. and other times the dps stays on me and the boss but I still cant build up any rage to attack. i always have to wait for enrage to cd. and that feels like an hour when in a fight. has anyone else had this problem?

 

Inaiwae

One thing worth mentioning is that imp. mangle increases uptime of [Idol of the Corruptor], since in bear the proc is not 100%.

 

Kestre

Delorn, I can't see your bear's armory so make sure you have Ferocity, Natural Reaction, and Primal Fury. I often have a bit of rage difficulty if I'm tanking a 5 man instance in Ulduar gear: at that point I use a mix of cat and bear gear allowing for high stamina but putting in my better damage pieces.

 

Helistar

Time for a little update on my simulations: from the statistics on the combat logs I've cooked up a healing model which is somewhat more realistic than the "perfect reactive healing" which I was using. The approach is still very rough, without any specific combat details, but has the advantage of breaking up the heal in smaller heals which are distributed non-uniformly.
Roughly, 40% of the healing occurs between 0.7 and 1.7 seconds after the damage is received, while 60% occurs between 1.7 and 6 seconds. The heals are random with an average around 6k and there's systematically 10% overheal. The resulting health/dwell time graph does not show the "staircase" effect which was present on the previous ones and resembles the "real" ones obtained from the log analyses.
It's interesting the shape:
- linear at high HP
- exponential at low HP
This would indicate that the area where healing is effective is the high part, with the low part of the distribution being due to vanishingly small events (multiple failure in avoidance/parry-haste events).

Example:


Using this model for healing, I've tried to determine a set of coefficients relative to "instagibbing". The approach is the following: during my long simulation, I record the number of deaths and in particular the number of "spike deaths", which are defined as follows:
- at time T0 the tank is beyond 70% health
- the tank is dead less then 5 seconds after T0.
By playing with the values for the stats I can calculate the weights affecting the relative number of "spike deaths". Very unsurprisingly, the coefficients are (bigger = bigger reduction in spike deaths):

Strength: 0.0574067
Agility: 0.173634
Stamina: 0.386456
Armor (Leather): 0.223255
Dodge Rating: 0.0711512
Defense Rating: 0.0691489
Hit Rating: -0.00826767
Expertise Rating: 0.306579
Haste Rating: -0.0457746
Critical Rating: 0.0160051
Armor Penetration Rating: 0.00129082
Attack Power: -0.00233191

(The simulation is done with my current stuff, which is a mix of Naxx25 and Ulduar10/25, I have no Hard mode stuff, so I'm far from BiS. Weapon is Origin of Nightmares + Mongoose, the boss has only physical attack, no special mechanic is simulated.... unrealistic? maybe, but the alternative is one separate simulation for each boss).

Comment: the tps stats are basically irrelevant, it's not like SD can save you. The reliable stats are (in this order): Stamina, Expertise, Armor, Agility. As I said this is very unsurprising: avoidance is of little relevance, since spike deaths are exactly cases where avoidance fails and survival depends on the "reliable" stats. Note that I'm still using 3.1 stats, so agi/dodgeR will fall even lower. Expertise is almost as good as stamina, since clearly parry-haste event are quite relevant to instagibbing, In the simulation healers are never interrupted/silenced, but it's still possible to be instagibbed, pauses in healing will only make things worse (and instagibbing only dependent on the tank stats).
As a result I'm now using for my "survival weights" a TTL approach (averaged boss DPS, like in Toskks') where Stamina/Expertise/Armor have an extra bonus to provide more protection against instagib.

 

Duilliath

I can understand Haste negatively impacting survivability... but hit and attack power?

An thenn tehy wuz al ded. Srsly. (Exodus 1)

 

Doxology

Maybe it's because if you never hit the boss, the boss can never parry your attacks (of course the benefits of hit in terms of threat greatly outweigh the incredibly small survival loss). As for attack power, I have no idea; shouldn't survivability go up because of the stronger Savage Defense?

 

kbranch

The values for AP, armor pen, and possibly hit are small enough that they're likely just the error margin of the simulator. The RNG is still somewhat random, even with an extremely high number of fights.

 

eMagdAeH

I think that is accurate, but it also depends on the proc occurring. If you're on an ICD time where SD cannot proc it does nothing against instagibs. This particularly matters because you cannot control when you want SD to proc and when you don't. As such, gearing specifically for higher SD isn't a bad idea per se, but it's certainly not an effective way to survive. What this translates into is simply that gearing for high AP will reduce the stat allotment you have to put towards Stam, Exp, and Armor. The secondary part of this is the AP we get via Agility, so seeking out a piece of gear for the AP exclusively compounds the negative effect of using stat allotment for AP over another stat. The number is so small it's nearly insignificant, but it is possible that it has a negative weight for the reasons I listed above. This is all speculation on my part, though.

I suppose it could also be SD was not taken into account when running this simulator due to its unpredictable nature, and thus not a reliable means of preventing instagibs.

 

Mijae

I've never heard of an ICD on SD, though it's irrelevant. There is absolutely no case ever that AP or hit could possibly cause negative survivability. Miss, dodge, and parry are all mutually exclusive. Even for attacks that are on a 2 roll system, the only portion that is on the second roll is whether or not the attack is a crit. So, the only mechanic hit rating can affect is the procrate of SD (and it can only increase it). The only mechanic AP can affect is the absorb amount of the SD procs. Also, agility does not give any AP in bear form.

If SD was not a factor at all, then hit and AP would have zero value. Similarly ArP should have zero value (unless you're counting killing the boss faster as reducing the chance for a gib).

Feral DruidStats spreadsheet
v3.2.0.0.xls - updated 8/4/2009

 

eMagdAeH

Yep, my mistake on the ICD, it's just set to the GCD of attacks which amounts to next to nothing. My only other thought then is that the simulator is "saying" the stat allotment used for this is so far behind the top 4 tanking stats it's giving them a negative value? That's a pretty far stretch, but it's all I've got that may explain it.

 

nightcrowler

@Helistar:

Your simulation is pretty interesting and I highly support them. From my simulation I've found a pretty similar rating and I really like them. For istance I always geared with stam>expertise>armor>agility. Preventing instant death is usually better than decreasing average damage taken.

You should also consider another factor. Probably expertise is also better than stamina and I'm going to explain that:
suppose that you have reached an HP value where the boss can't 2-3 shoot you but it can 3-4 shoot you, usually you'll end up needing another 10k HP to reach the "next step" in that case adding expertise till the cap is the only thing that can save you from instant gibbs.

 

Helistar

This is the explanation, I forgot to mention the fact. I left them in because they give you an idea of the error on the coefficients (it's quite small). Negative haste is probably true (it's negative for TTL as well and around 10-20x the value for hit/AP), but for the other two it's very likely numerical error. Note that I gave you the value of the derivatives, without mentioning the "hard" values, which are enormous for stamina/expertise/armor. For example the 0.3 for expertise rating means that comparing two simulations which are 100 expertise rating apart, there's a 30% reduction in the number of spike deaths.

@nightcrawler: what you say is true and was extremely visible in my old simulations (the one with a "staircase effect"). In that case, stamina only plays a role if it moves you from one step to the next (i.e. gives you the ability to survive one more shot), otherwise it's irrevelant. For fights where healing must stop for one reason or another the effect will then be exactly what you say: gamming for more stamina is useless unless it allows you to survive one more shot.

@Latas: I had a look ar grimreaper and the approach is interesting, if I understand it right (I'm not familiar with addon programming), it uses both unfiltered events to see the damage and health change events to correctly track the amount, if I can manage to reproduce the code and store the log in a saved variable it'll be possible to improve the graphs, even if saving requires a logout or /reloadui.