The Prot Warrior QDTG

[Kojiyama]

Good point in regard to the miss/expertise thing, although I think typically you would see a lot lower than a 20% miss rate on a BiS Warrior. At the very least most Warriors tend to be softcapped on Expertise, if not having a little bit more and possibly also have a little bit of Hit floating around.

I somewhat adjusted for your 2nd point already. As I mentioned, my adjustment is a 30% chance to increase the cycle length by 3 seconds--or, alternatively, the lack of Devastate/Revenge's 30% chance to reset the cycle.

(e.g. instead of having a 30% chance for it to be 3s, 21% chance for it to be 4.5s, and 49% chance for it to be 6 seconds it becomes a 30% chance for it to be 4.5s and a 70% chance for it to be 6 seconds. That makes the average cycle length 5.55s instead of 4.7844s, or a loss of 0.7656s. I slightly miscalculated this when I typed it put it as 0.9s, so the value is 0.1344s off. I will correct it, but shouldn't alter the results much.)

Also don't know why I typed 51%, think I just copied the sum instead of the correct column in my numbers. (You have a 49% chance of reaching the 3rd GCD, as you point out.)

All those points considered, it probably is still worthwhile to only use Revenge in the 3rd GCD scenario as that will be optimal regardless. Missed opportunity is still going to be in play even with a reduced hit rate, and while the difference would be smaller it would still exist.

Not sure in regard to the Revenge figures. In damage, I could certainly see it, but are you accounting for Shield Slam's bonus threat and 30% threat multiplier? Additionally Devastate sneaks up due to the Glyph and AP threat multiplier.

Edit: Corrected the figure in example 1 (only 88 TPS lower instead of 144 TPS lower than a normal rotation) and slightly adjusted the maximum average waiting time for a 3rd GCD--just a couple 10ths added, though.)

[Delmonte]

MM yeah I don't know what most people run with, my chance to connect is about 80% and I armoried Kungen and he only had 11 expertise (2.75%) but 7.72% hit so that's around 85% chance to connect.

In terms of the numbers I may have made some mistakes, I put all my formulae in my thread and was using (Damage + 770) x 1.3 for shield slam's threat iirc. It's more likely that if I have made a mistake, it will be with revenge as I get SS and dev in similar ratios to you but revenge is higher than devastate as well as being on a par with SS (in my gear in ICC).

Here's my thread, formulae are near the bottom:

http://www.tankspot.com/showthread.p...es-spreadsheet

It could be that we're looking at quite different setups in terms of stats/buffs. Are your numbers with or without improved revenge?

[Kojiyama]

AFAIK, Improved Revenge was shown only to affect the base damage, so it's not a massive increase.

As I maintain Rawr.ProtWarr I code/use that for all my calculations. There is a chance something is not up to date, but last I checked it should all be. Keeping in mind, of course, that gear/talents/glyphs will modify the results quite heavily.

Even with talents and such, though, I've never seen Revenge being calculated as remotely matching Shield Slam's threat. If you are calculating Improved Revenge into the scaling component--or they have actually fixed the talent when I wasn't looking--that would probably explain the difference.

[Delmonte]

Ah ok that's interesting, can you remember the link to where Imp revenge was tested? I can't remember exactly what testing I have for revenge, I'll have a look through my data some time and play around with the talent.

I presume it was verified in raids or instances but can you remember how extensively it was tested? I'm just asking because, to the best of my knowledge, there is a tooltip bug with rend where the improved rend talent is concerned, it only displays the damage with the base being multiplied by the talent but actually, when you test the ability I found that both the base and AP/weap dmg component of damage are in fact multiplied and revenge is somewhat similar in that it has a base and added AP component.

I will recheck rend too in case I made a mistake but I was fairly sure at the time that it was a tooltip bug, there is also a tooltip bug with haste showing reduced dmg on tooltip but in fact not affecting rend damage at all.

Just quickly looking at revenge, the tooltip values only multiply the base as you say but I'm not convinced that the tooltips are correct so I'll look into it.

[Kojiyama]

I don't remember as it was some time ago, but I do recall people being fairly convinced at the time that it was the case.

It should be fairly easy to test again with high AP, as the difference between the two multipliers should be quite evident. If you happen to find it is multiplying the entire value, I would be happy to know that since I always found that functionality a bit inconsistent and buggy.

(I will also test on beta just for the heck of it.)

[Delmonte]

Yeah just trying to play around with it atm, trying to motivate myself to untalent my warriors again ^^. I'll try to put some results up soonish.

[Delmonte]

Just done a quick test in Nexus heroic on a lvl 81 warder. I did all tests without improved d-stance or 1h weapon spec, without and then with 2/2 improved revenge. I forgot I had mongoose on my weap but I don’t think there were any other factors that should have affected the results (I hope!) and mongoose would make negligible difference to AP anyway. I was very careful not to use revenge when the mob had it's barrier thing up but you actually get a full absorb on revenge with it anyway and it shows up in a different row in recount if you do so it wouldn't be shown in the hits row which is all I used for this test.

Armour estimation – using shockwave without t10 (although it worked with t10 to):

I worked out a rough value for the armour by interpolating values I found for lvl 80 and 83 mobs (not sure how accurate the 80 one is), it’s not spot on because I’m pretty sure mob armour vs level is a curve and I used a straight line to interpolate but it seems reasonable nonetheless. I got a value of around 39.7% damage reduced i.e. around 60.3% damage taken. For shockwave it predicted the following:

Zerker dmg – 2190
D – stance dmg – 2081

And I observed these damages:
Zerker – 2193
D – stance – 2083

So it seemed fine for the purposes of this test.



Revenge without imp revenge talent in d-stance:
N.B. I used quite a small number of revenge hits but the predicted values varied so greatly that I felt fairly happy drawing a conclusion with these numbers.

Results:
4845 AP
Predicted tooltip values – 3137.95 – 3499.95
Observed – 3138 – 3500

This predicts damage values of:
Min – 1797, Avg – 1901, Max - 2005

Observed damages (12 revenge hits):
Min – 1854, Avg – 1908, Max – 1982


5395 AP
Predicted tooltip values – 3308.45 – 3670.45
Observed – 3308 – 3670

This predicts damage values of:
Min – 1895, Avg – 1999, Max - 2102

Observed damages (12 revenge hits):
Min – 1932, Avg – 1995, Max – 2076


Using 2/2 Imp revenge:

With 4845 AP

So the 2 predicted ranges for the tooltip are:

If imp revenge only affects the base damage:
4119.55 – 4698.75

If imp revenge affects the total dmg:
5020.72 – 5599.92

Observed tooltip values:
4119 – 4699

So the value displayed on the tooltip is:
(base damage x 1.6) + (AP x 0.31)


The predicted damages are:
If imp revenge only affects the base damage i.e. the tooltip is correct:
Min – 2359, Avg – 2525, Max – 2691

If imp revenge affects total damage:
Min – 2876, Avg – 3041, Max – 3207

Observed damages (25 revenge hits):
Min – 2900, Avg – 3058, Max – 3165


So this lies completely within the range predicted if imp revenge actually multiplies total dmg by 1.6 and the average looks reasonable give the small sample size. It is also completely outside the range predicted if imp revenge affects only base dmg.


Similarly with 2/2 imp revenge and 5395 AP:

Tooltip value: (base damage x 1.6) + (AP x 0.31)

Predicted damages:
Imp revenge affecting base only:
Min – 2457, Avg – 2623, Max – 2789

Imp revenge affecting total dmg:
Min – 3032, Avg – 3198, Max – 3364

Observed (25 revenge hits):
Min – 3042, Avg – 3215, Max – 3361

Which is a similar pattern to the results for 4845. I’d conclude that there is a tooltip bug with improved revenge and that the talent actually affects total dmg. I’m pretty sure the same tooltip bug exists for improved rend. Hopefully I haven’t copied numbers incorrectly, if they look wrong or you’re not sure how I arrived at them then let me know and I’ll check I copied the right ones and the formulae are right but I observed the patterns I described. I admit it's pretty rough and ready, if anyone see any problems with this let me know.

N.B. predicted damages for revenge are base x 0.95 (d-stance) x 0.60288...... (which was my estimated armour modifier tested with shockwave).

[Kojiyama]

I think that is a pretty conclusive test, from the looks of it. It seems that they have fixed the bug (I'm pretty sure I confirmed that it was base-damage only at some point in the past) but not reflected the fix in the tooltip. I will adjust my calculations and Rawr with the correct values.

I would expect the same to apply to Rend, really.

Thanks for doing that testing, it's quite handy info.

In regard to how it affects my previous posts--all it would do is increase the base TPS of the normal rotation (by quite a fair bit) and further decrease the value of using Shockwave at an incorrect time.

Moral of the story: don't press Shockwave at the wrong time.

[Delmonte]

No problem, I tested rend on the target dummy a week or 2 ago iirc and it had a similar tooltip error with imp rend. It may well be that both were bugged then fixed but the tooltips were not updated as you say. There is also a bug where haste reduces tooltip dmg for rend but does not affect actual damage caused.

[Insahnity]

Sorry for falling behind on this, one points from a few days ago that I'd like to chip in.

First off, lemme preface with Agg's own words, repeated several times.

Idk this is why it's a quick and dirty guide just to get people on the right track...

He just can't cover it all in a Q&D guide and nor should he. Bearing that in mind:

Some minor ones:
2. You point out that bothering your 5 stacks of sunder as fast as possible isn´t as mandatory as most people think. I would make it a bit clearer that 5 stacks of sunder do increase your own tps by reducing the targets armor, not only by using the devastate itself.

This is true, but it is REALLY suboptimal for your burst threat at the beginning of a fight to get the sunder stacks up. Like yes it does increase your TPS to have 5 stacks up, but the TPS loss from getting 5 devastates out first before unloading your big guns is too big. Let the sunder stacks get up as you go or make a fury warrior do it or something =P.

Priority is something that *HAS* to be flexible. If you are tanking the trash leading to marrowgar, TClap and Shockwave are priorities, where as SS and Revenge/Devastate are for bosses. Everybody here understands this, but those trying to read it may not, and they shouldn't depend on Q&Ds for learning this (and really, I hope they learned this LONG before stepping into ICC). Also, it's learning to manage threat moves as well as keeping up the mundane but important debuffs like TClap and Demo Shout without slipping on the TPS. Speaking as a healer, a lot of healers can carry your through if you develop bad habits in regular, but you *will* get slaughtered in heroics.

What we are missing is fight dynamics. Are the DPS switching targets? Are you picking up something before DPS get to it? These will determine your order somewhat. If DPS is not on your target on the get-go, then yeah putting up 5xdevastates could outweigh an SS, because by the time you get up your 5 sunders, you still have time to toss on that SS. If you *have* to get snap aggro, then sunder stacks take a back seat. Also, don't forget you have an OT. If it's a warrior, make them the Sunder/Demo Bitch while you are grabbing snap aggro. Druid? Infected wounds pops up quickly, and Demo roar is top on their priority. DK? Icy touch is done, and perhaps ask a DPS warrior to toss the demo shout, it's only one GCD. Paladins are a swing item, they will judge quickly to get the Attack Speed debuff quickly (and if specced correctly, to also get crusade up for 3% crit), but the vindication is subject to RNG. it should go up in 3-ish GCDs, but if you don't see it going up fast, then you will have to Demo shout in the interim, until vindication overwrites this (most warriors spec 15/3/53 and don't have imp demo, whereas proper paladins spec into 2/2 vindication, which will eventually overwrite a vanilla demo when it pops up.).

Blood Princes comes to mind where you can think of devastates over SS. If you are charged with tanking the 2nd prince, everybody will be on the first prince so you could get your sunders up and then wail in on the fellow, whereas the tank on the first prince needs snap aggro.

Finally, snap aggro is academic if you can train your hunters and rogues to MD/TotT. I know some don't have that in 10 mans or can't train people to do it, but if you can, a raid with appreciable melee benefits more from 5x sunders than having the tank grab snap aggro for the sake casters. Again, Q&Ds don't cover this.

[Kojiyama]

I honestly never worry about Sunder stacks as I use Glyph of Devastate which makes it trivial to get a full stack in any resonable amount of time.