Nezralix

I counted up the talents that have rage gain or efficiency as the sole primary purpose of the talent:

Arms:
Improved Heroic Strike - 3 points
Improved Charge - 2 points
Tactical Mastery - 3 points
Anger Management - 1 point
Endless Rage - 1 point

(Arms Total: 10 points)

Fury:
Unbridled Wrath - 5 points
Improved Execute - 2 points
Improved Berserker Rage - 2 points
Unending Fury - 5 points

(Fury Total: 14 points)

Prot:
Improved Bloodrage - 2 points
Puncture - 3 points
Focused Rage - 3 points

(Prot Total: 8 points)


Here's another set of talents with rage gain or efficiency as a primary contributor to the talent, if not the whole purpose:

Second Wind - 2 points
Intensify Rage - 3 points
Shield Specialization - 5 points
Sword and Board - 3 points
(13 points total)

So you're looking at 32 points worth of talents that do nothing but assist rage gain, or 45 if you include talents that are strongly influenced by rage gain.

To me, looking at this and comparing it to other melee-oriented classes, this is a ton of point investment focused on propping up a somewhat irritating resource system. Markedly different from energy in that there's frequently no benefit from having these talents (i.e. those cases where you have plenty of rage). And also somewhat awkward in that, even with all of these talent points available, you *still* very often end up rage-starved.

I wish they'd take a good look at rage to figure out whether the current implementation is something that players enjoy.

 

Qed

I'm not saying that avoidance doesn't scale 'very well' or 'very fast' or even 'super-linearly'. I'm saying that exponential is a very well defined word, and it's definition isn't 'very well' or 'very fast' or 'super-linear' or even 'to the power [2,3,4,5,6,etc.]'. The only definition of exponential is the mathematical one. If you use exponential as a synonym for 'very fast', you're confusing the term, and needlessly so.

I recognize that most people don't make this distinction which is why this is a nitpick, but it's grating like any other error in grammar.

In a sense, avoidance scaling is 'better' than exponential, because you get to infinite effective life in a finite amount of avoidance, where an exponentially scaling would require an infinite amount of avoidance to do so.

 

mistersix

It's also not fun at all to run an instances well past their expiration date to get that last fun item for your threat set or that super cool item for your avoidance set. Specialty items like that are part of what slows guilds down in terms of progressing through the content faster. I think this aspect of gear homogenization is intended to help drive their stated goal of wanting more players to see more content.

I think that's right. I also think that (and the nerf to bear armor) is where the devs were going with making warrs (and ostensibly paladins) more desirable in terms of mitigation.

 

Gruntle

You should perhaps have read my original post. I never claimed that. The mistake I did was to use hasted speed instead of the base speed of the weapon when getting the added swing damage from the AP dps. This is apparently not how it works.

 

Liar

Exactly what items are you referring to? The only item I can think of would Moroes but that's because early trinket itemization was fucked up in general. They learned their lesson and what we got is a perfect system with MgT's trinket selection and availability. I don't know about you, but just going for 1 item per slot is going to be extremely boring longterm. And that's including offspec.

Unexpected TankPoints error Oh, really?

 

Tankietka

I wonder why almost no one mentions rotation using Taste for Blood mechanic?

For example Overpower/Rend/Slams one, which would work as follow:
1. Overpower whenever possible
2. Rend if not on target
3. Slam spam as long as rage allows

Anyone tested it on PTR?
According to results from my simulator, it outperforms both MS/WW/Slams and Execute/Slams.
I would like to see some real world tests as simulator is still not finished completely so maybe results are not correct.

Below are my sample results.

AP: ~1650
Crit: ~35% with axe spec (32% for OP/Rend/Slams - battle stance)
Weapon: speed 3.6, avg dmg 427
Build is 50+ in arms with all needed goodies
Simulated fight length: 3000000s
Abilities are those from level 70

MS/WW/Slams rotation (MS and WW as priority, Slam spam if both on cooldown or not enough rage for them):
DPS: 628

Swing 35.74% of damage
Mortal Strike 13.78% of damage
Deep Wounds tick 9.17% of damage
Whirlwind 8.07% of damage
Slam 33.24% of damage


OP/Rend/Slams rotation:
DPS: 858

Rend tick 8.21% of damage
Swing 24.24% of damage
Overpower 19.78% of damage
Deep Wounds tick 8.45% of damage
Slam 39.32% of damage

 

moxy

@QED: Ah, you're right. I had y=x^n confused with y=n^x (look! I've got an 'exponent' too!) Nice catch. I didn't mean to play fast and loose with my terms, it's just been a very long time since calculus.

I'm very curious to know what the formula is they're referring to as "diminishing returns". Seems like the most logical approach would be something where each point of avoidance rating reduces the chance of an incoming hit by the same amount, so that it could never reach 100%, and so that the relative value of avoidance stays roughly the same throughout progression. But the fact that dodge and parry are separate stats argues against this, since the formula would need to work on the combined value of dodge, parry, miss to know what the chance of incoming damage is. Also the way the tooltip is worded seems to indicate that its just going to be some discrete reduction placed over top of the old simple formula, where each point past a certain threshold, or perhaps several thresholds will lower the conversion rating, e.g. after 400 dodge rating, the conversion of dodge from rating would drop from 1:39 to 1:44 for each point over 400. In testing I've done over a few hundred hits, I can't see any difference. Either it's too small to pick up without a much larger sample size, or it's NYI, or it simply doesn't kick in until much higher avoidance ratings. (I'm using mostly the "new and improved" T6 with avoidance gems.) Has someone done some more extensive testing?

Oh never mind, I see they now say the hotfix for this is not working correctly.

 

Hypatia

The correct formula would be something like A = 1 - 0.5^(a/n), where a is something like "avoidance rating", and n is how much you want to give 1/2 avoidance. Each n worth added to a gives an additional 50% reduction from the previous value of a. But how to balance that across multiple different sources of avoidance is a harder question.

 

Lithose

Its been about 10 years since calculus, and I know its not exponential in the strict sense of the word, but most people here will relate the term exponential for anything where a finite addition of one variable will or can produce an infinite amount of another. Granted exponential can't be infinite (At least not in the calc I completed.) but it is close enough that the relative use of the term is pretty much understood by the majority.

Which, leaves the point, proper definition use aside, the super linear growth of said stat was its sole advantage of it over armor. Is there a way it can scale in a manner more in accordance with how armor scales without being completely inferior due to small data sets used in tanking?

The only way I can think of, is to have fights last a long time and have mana be an issue (Which they want to do). This way avoidance and armor becomes "saving" stats, while stamina becomes the "anti burst" stat...By making damage/time a bigger issue, it could increase the value of avoidance "gemming" even while it has a DR, as we can't increase armor (Blizzard is specifically staying away from bonus armor itemization.).

Perhaps this is blizzards goal with the regen changes, or one of them. They want avoidance to be valuable not as a means of trivializing all damage, but as a means of allowing mana to last through the length of a fight (Hard thing to balance, imo.).

 

mistersix

This could be accurate if mana regen was getting better across the board but the recent page has nerfed some classes pretty significantly in that regard.

 

Chirality

I tend to agree with you. Rage is exceedingly frustrating at times. Missing pummels because I have 8 or 9 rage and losing games? Not fun. Blowing 100 rage to stance->reflect something and being helpless for 10 seconds? Also not fun.

One thing I would like to see is a significant buff to Anger Management. It should really be around 1 rage per second, not per three seconds.

And the way Power Word: Shield completely rage-starves is completely fucked up.

 

Shha

Well Im not going to look deep into your specific numbers. Let me explain though why overpower seems like meh mechanic at the moment. Ill use numbers for lvl 70 for now so lets go.

Raid buffed at 70 as arms i have around 55% crit. Maybe more. Im at 41% unbuffed (socketed for crit for 2h), and raid buffed I expect to have in realm of 3.5k ap. Those are rough values, but my unbuffed AP is around 2400 (thats with shout though). Str totem, some buffs, 10% ap from buffs.

With Torch of the Damned (which seems to pull ahead other weapons atm due to speed - although my testing isnt conclusive yet)

Slam damage= (130.8+3500/14)*3.8+140=1587
Rend Damage= ((130.8+3500/14)*3.8*0.3+255)*1.5*1.3=1344
Overpower Damage = 130.8*3.8+3500/14*3.4=1347

At very 1st glance you can see that overpower and Rend damage have less damage then Slam. In fact they dont even crit much more - yes overpower crits 100% with my stats. But rend crit is at 0....

Lets see

Slam modified damage =1697*(0.45+0.55*2.27)=2595.
Overpower modified damage = 1344*2.27=3050.

As you can see overpower is barely better for me then slam - and thats counting the crit.
Over 30 seconds we will use rend 2 times, and it will tick 10 times - giving on average 3 overpowers.

3xoverpower+2xrend vs 5 slams. 3x3050+2x1344=11840vs 13475.

As you can see while overpower/rend costs less rage, it produces less damage. AND it forces you to stay in battlestance (seems better then dancing even with new TM).

Hasted speed in raid = 3.8*(1/1.2)*(1/1.06)*(1/1.03)*(1.1)=2.64
White damage over 2.5 seconds= 2.5/2.64 of a swing
Swing damage = [0.25*0.76+0.55*2.06+0.2]*1447=2203.
2203*2.5/2.64=2086
Overall damage is actually higher for "overpower rotation" by 13926 vs 13475. Lets add 20% more damage for this calculation to rend - to compensate for armor 1344x40%=14463. Overal damage gain for using oevrpower is a bit under 1000.

But lets take a faster weapon into account. Lets see Apolyon here. Cutting on the text and staight to calculations

Slam damage= (148.7+3500/14)*3.4+140=1495
Rend Damage= ((148.7+3500/14)*3.4*0.3+255)*1.5*1.3=1290
Overpower Damage = 148.7*3.4+3500/14*3.4=1355

Slam modified damage =1495*(0.45+0.55*2.27)=2539.
Overpower modified damage = 1355*2.27=3076

5xSlam = 12696
3x OP +2x Rend= 11808

White damage lost = 2184 (saving the math here).
40% of rend = 516

Overall 14508 vs 12696 - around 1800 damage gain.

So from 1000-1800 damage gain per 30 seconds, as a function of weapon speed (faster weapon gain more). thats 30-60dps - with a well over 2000 dps build.
Cost is simple - staying in battle stance and losing 3% crit. Its hard to really measure the impact of crit here. The simplest way of multiplying the damage by (0.45+0.55x2.2)/(0.42+0.58+2.2) (approximated the effect of impale for 2.2 multiplier - just trying to judge the general value here) - gives around 2.2% damage increase. There is however side effects like WW use and increase uptime of wrecking crew/deep wounds/better rage generation). All in all, it seems that difference for using overpower is in realm of maybe 20dps if you have fast weapon, or losing 20dps with slow one.
Now after this edit its not that obvious which is better. But frankly this calculation assumes absolutely perfect and non laggy environment. If you clip a rend tick by reapplying it 10 gcd later but with slightly differnt latency - you might lose all the benefit you gained. If you delay MS for a gcd to use overpower - the same happens. Finally its just like is said in post below - will you really stare at rend timer for a POSSIBLE, and not even sure, and very small dps increase?

Problem with overpower is simple - it does crit a lot yea, but its base damage is worst of our all attacks (beside WW). It gets even worse when the weapon is slow since OP is normalized and Slam is not.

To make it appealing for arms warriors they should either bring back the "accidental" bug with rend lighting up to 3 overpowers - look at numbers it barely outdpses slam so its far from OP like people said, or let us rend and overpower in zerker. Right now the OP talents are wasted in my opinion.

Same goes for SD execute. Right now it can be useful but its way too situational with a 2hander. The risk of executing to 0 rage, then getting a glancing blow and losing a slam or so in the process is just too big. There are specific situations when SD is useful - but given how the proc isnt always up, and the benefits are usually small, its to the point where its better to just ignore it.

 

Liar

Maybe I am wrong, but you didn't include boss armor values in your math. Since Rend is not affected by armor at all, OP and TfB could pull ahead. Especially considering how crappy the new Armor Penetration Rating is.

Unexpected TankPoints error Oh, really?

 

Rustik

For the Avoidance discussion, it's currently the topic of choice over at the "Combat Ratings at 80" thread. It's just starting though, so I'm keeping an eye on it for future reference.

On an unrelated note, I was poking around the PTR yesterday, and Shockwave and Warbringer are full of win.
My unbuffed block value and AP doubled from live. I lost about 3% parry, and 2% miss, but I actually gained 1% dodge. Tanking is very entertaining, now.

Though I'm not sure if t-clap is still generating more threat than it should or not. is it still 363% in def?

 

Herb

Also, Slam adds only 140 additional damage at 70 (roughly 1000 damage less for 5 x Slam), and you'd lose 2.5 seconds worth autoattack with Slam (another 1000 damage less, even w/o haste). In other words: for any real life situation (with haste & armor) the Rend/OP cycle is, in fact, more damage.

Why are you comparing with "5 x Slam" to start with? If that's the expected rage you can spend per 30 seconds (75), then you'd have rage for nearly 3 x Slam left on top of Rend/OP (35), which easily is 60% more yellow damage over said period.

 

Graul

From personal experience, the other issue with ToB is that Rend does not have a 100% uptime. You can nearly get that if you stare at it the entire time it's up, but this just ends up going right back to the whole unfun aspect of staring at a swing timer. I've found it much easier to keep consistently higher DPS just staying in Berserker and alternating Slam with lowest rage Sudden Death Executes. Overpower is definately more damage per rage, especially when you factor in the swing time cut out of Slam, but it's not consistently more DPS. And this was just on a dummy, there should be no problem with rage on a raid.

 

Moomore

Is it not interesting then to think about a rend OP and SD execute rotation while DW 1 handers? Since OP is normalized.

Wouldnt this give you more rage making the executes bigger and more wirthwile? Aswell as making sure you crit more often making the SD proc more often.

I'm not sure if its allowed to link to different websites and vids here? Because I have seen a pretty interesting movie regarding the above.

 

Shha

I edited post above to take acount for the replies. Indeed it had some mistakes however:

a) Rage is not an issue. You HAVE rage to use on slam/ms/ww on cd, you really dont for anything else. Heroic strike is too much of a spike loss to even consider. SD is only useful in certain conditions. Either way - if you even consider spending this extra rage for executes its 40x21 damage before crit and armor reduction - but you get less rage (3% less crit in this scenario is around 2% less rage - over 30 sec it adds to an substantial number), and less crit on execute.

b) There is something like "maintenance cost". If 2 styles of play provide same results (as in this case - the effect varies depending on speed of weapon) - pick easier one. Seriously - remember that this calculation assumes you use rend and overpower at perfect times, with no hesitation, delay, human error at all. Frankly if you make a /castsequence MS,Slam,Slam,Slam and just spam the button, you will end up with more damage, I can basically guarantee it.

 

Tankietka

If I understand you correctly, you are comparing OP/Rend to Slam?
In my opinion OP/Rend does not replace Slam in rotation but (in case of the simplest one - OP/Rend/Slams) MS and WW. So we should rather compare DPS of OP/Rend to MS/WW.

With your stats we would have the following:

Slam damage= (130.8+3500/14)*3.8+140 = 1587
Rend Damage= ((130.8+3500/14)*3.8*0.3+255)*1.5*1.3 = 1344
Overpower Damage = 130.8*3.8+3500/14*3.4 = 1347
MS Damage = 130.8*3.8+3500/14*3.4+210 = 497+850+210 = 1557
WW Damage = 130.8*3.8+3500/14*3.4 = 497+850 = 1347

Accounting crit:
Overpower Damage = 1344*2.27 = 3050
MS Damage = 1557*(0.45+0.55*2.27) = 2644
WW Damage = 1347*(0.45+0.55*2.27) = 2288

So over 30s we would have either
1. 2xRend+3xOP = 2x1344 + 3x3050 = 2688+9150 = 11838 damage for 2x10+3x5 = 35 rage
2. 6xMS+3WW = 6*2644 + 3x2288 = 15864+6864 = 22728 damage for 6x30+3x25 = 255 rage

Damage does not look good for OP/Rend, but the point is that rage spending is very low, and we will use that additional rage for more Slam spam.
So every 30s using OP/Rend we will have 255-35=220 more rage to use on slams, which means 220/15=~14 slams.

Slam Damage = 1697*(0.45+0.55*2.27) = 2595

Accounting these slams we have:
1. 11838 + 14*2595 = 11838+36330 = 48168
2. 22728

Of course we are loosing white damage:

Swing damage = [0.25*0.76+0.55*2.06+0.2]*1447=2203

14*0.5=7s worth of swing which is 7/3.8*2203 = 4058 damage lost due to additional slam spam

Finally:
1. 48168-4058 = 44110
2. 22728

Now it looks a bit better but, as you mentioned, reapplying Rend is not always perfect, which additionally lowers number for OP/Rend/Slams rotation. Of course due to less number of swings we have a bit less rage in OP/Rend rotation but still it looks like it is not needed to outperform MS/WW/Slams, even with 3% crit less due to Battle stance.


I've put your stats into simulator. With such high crit values Rend and OP damage % went down, but now slams are clearly the top damaging ability.
Also interesting is that uptime of arms buffs is better with OP/Rend/Slams.

OP/Rend/Slams
DPS: 1955.77

Rend tick 5.06% of damage
Swing 20.03% of damage
Overpower 17.22% of damage
Deep Wounds tick 7.18% of damage
Slam 50.5% of damage

Wrecking Crew: 99.99%
Deep Wounds: 99.18%
Blood Frenzy: 99.98%
Trauma: 100.0%
Rend: 94.67%


MS/WW/Slams
DPS: 1613.87

Swing 26.87% of damage
Mortal Strike 19.1% of damage
Deep Wounds tick 8.55% of damage
Whirlwind 8.53% of damage
Slam 36.95% of damage

Blood Frenzy: 92.09%
Wrecking Crew: 99.19%
Deep Wounds: 92.09%
Trauma: 99.73%

 

PSGarak

That is an invalid assumption. A cycle will use MS and WW on cooldown, and the question is whether to prioritize slam as the filler, or to try to fit rend + OP into that filler as well, and comparing the opportunity cost of the slams being replaced. Rend and OP will not be replacing MS and WW, because they don't collide. You can save your TfB procs for the up-to-three seconds it takes to perform MS+WW, with only a 30% chance of a wasted proc in the worst case. The original comparison is more valid, since you're comparing cycles and he was comparing fillers.

You're also assuming a rage-limited environment, which by all accounts is not the case. As a result of looking at rage cost for a thirty-second MS+WW cycle, you're comparing it to a cycle with 19 attacks, for which there is not enough global cooldowns.

The sim results are interesting, but inconclusive. You need to either compare an OP/rend/slam cycle to slam-spam or, preferably, a MS+WW+Slam cycle to a MS+WW+OP+Rend+Slam cycle.

http://sig.gamerdna.com/quizzes/INFL...erks/Garak.png

 

Tankietka

Ah, I see now, my bad.

Yeah, calculations can be off here, but in simulation rage used is rage generated. How often does rage unlimited scenario happen in real fights?

I have slam-spam rotation implemented already, below are results. As mentioned above - rage used is rage generated.
As for MS+WW+OP+Rend+Slam I think WW would not be worth using due to stance dance needed, but I will check MS+OP+Rend+Slams in free time.

Slam spam (with Shaa's stats)
DPS: 1707.5

Swing 20.97% of damage
Deep Wounds tick 7.92% of damage
Slam 71.11% of damage

Blood Frenzy: 94.29%
Wrecking Crew: 99.42%
Deep Wounds: 94.29%
Trauma: 99.8%

 

Shha

As 2h arms - every fight. As long as you dont touch your execute button, you can use a skill every gcd. That doesnt even take extensive buffs.

Your calculation (if they use 3500 ap number as base), are off at least as far as deep wounds go. With 2400 ap testing setup my average deep wound tick was ~280. Thats 280 dps from deep wounds alone basically - your calculation shows 120 or so.

As far as cycles go - I dont think WW is necessary. However this requires some "math".

3500 AP + Torch 55% crit (2.64 modified speed)

WW damage = 130.8x3.8+250x3.4=1347
Slam Damage = (130.8+250)x3.8+140= 1587
White damage loss= (130.8+250)*3.8*0.5/2.64=274

3500 AP + Cataclysms Edge 55% crit (2.43 modified speed)
WW damage = 138.7x3.5+250x3.4=1335
Slam Damage = 1500
White damage loss =(138.7+250)*3.5*0.5/2.43=279

Modifier for yellow attacks = 0.45+0.55*2.27=1.6985=~1.7
Modifier for white attacks = 0.25x0.76+0.55*2.06+0.2=1.523=~1.52

Torch
1587-1347=240
240*1.7=408 more yellow damage and 274*1.52=416 less white damage.

Edge

165*1.7=280.5 more yellow damage 424 less white damage.

Damage difference between WW and Slam is very minimal .

While actually seems that WW is still beneficial , it does change my opinion on "perfect rotation". Seems like you stronly need to prioritize:

a) Using every GCD
b) Using MS over WW, rather then maximising the amount of instants.

Therefore the 31.5 sec rotation i posted above that gives us maximum number of instants/time is incorrect. A more simple

MS,WW,Slam,Slam,MS,Slam,Slam,Slam with MS every 6 and WW every 12 seconds is more beneficial. Also WW really "doesnt matter" much - using slam instead gives you from 0.66 to 12 dps loss based on weapon. As in - forgetting WW in the rotation wont really have big impact on your dps.

 

Tankietka

Yes, but I was looking at this from a bit different perspective - how many slams/other abilities I can use between swings?

Seems like I will have to do some more testing
1. MS/OP/Rend/Slams cycle
2. Check all rotations in situation when each swing gives 100 rage

Hmm, strange. Currently I'm calculating Deep Wounds as follows

Swing damage = 1447

Deep Wounds damage = 1447 * 0.48 * 1.02 (Blood Frenzy) * 1.06 (2h spec) * 1.3 (Trauma) * 1.1 (Wrecking Crew) = 1074 over 6s = 179 damage per tick

Also the above is the maximum possible value, sometimes it can happen some buffs will not be there, lowering final average tick + sometimes tick will be clipped by close consecutive crits - as far as I know that's how it currently works on PTR?

Some buffs I missed or bad calculations? Please note that currently raid buffs are not implemented in sim, maybe that's the difference you see. Actually I'm starting to think that the lack of those (especially those that give haste) give better results to Slam spam and OP/Rend/Slams rotation. Will have to implement those asap.

 

Grubsnik

Just a quick addendum to the entire slam discussion, doesnt the 1.5/0.5 sec cast time get modified by haste like everything else?

I've never tested it, since it was negligble for live, where your reaction time was more important that the ~.1-.2 less seconds of cast time you might be getting. But it will matter in 3.02, where you can spam it.

Should be easy to test too. Just grab a shaman, get a Lust/heroism, do an unimproved slam and note down if you get it in 1.35 or 1.5. Would be nice to clear up if Slam is getting penalized through haste and lack of scaling.

 

cpkfolief

One big point I would like to make though is that through RNG mechanics inherent in the game it is completely possible for a Slam/WW/MS rotation to have Deep Wounds fall off, even into mid-range gear levels (obviously a raid DPS loss). A Rend rotation is basically guaranteeing Blood Frenzy uptime. Keep in mind that without Rend the window for refreshing Deep Wounds is six seconds, which is considerably small at lower gear levels.

Another thing I just realized is that lower expertise levels one will see more dodges, so being spec'd into Imp. Overpower and Unrelenting Assault is giving a very nice DPS save, with Overpower's cost being only five rage.

Let's look at it from this point of view. Rend/Slam/OP rotation seems to give advantages as low gear levels, whereas the Slam/WW/MS rotation will make more use of a higher gear level through Impale and the ability to keep Deep Wounds up without Rend. A lot of the recent posts with DPS comparisons seem to be using end-game (close to best in slot) gear stats supplied by Shha.

Here is the post made by Tankietka which gives a better look at what DPS low-level gearing provides in term of the two rotations: http://elitistjerks.com/f31/t27758-w...92/#post903118