I'm pretty disappointed in the current Arms rotation, according to Celestalon it should be keep CS on CD, keep Ms on CD fill with whirlwind, which in an of itself is boring, and even with the later addition of Rend its still pretty boring, that however isn't my main concern; my main concerns are Rage management the raw number of empty GCDs and the Headlong Rush. MS costs 30 rage, WW costs 30 rage, CS costs 10 and Rend cost 10 as well. Our only source of Rage generation is white swings, which for arms generate 30 rage per swing and 60 Rage if they crit. As white swing crits are largely out of your control this means that to not cap out on rage and thus waste it, we need to either need to be particularly attentive of when our next white swing is going to happen or For the most part this means we will want to keep rage between 30 and 40 at all times, that's a very tight operating band, and pretty much make Glyph of Unending rage mandatory. I was under the impression that Glyphs where meant to other options, not illusions of choice.
Baseline 2 handed weapons have a swing speed of 3.6 seconds. Each swing gives 30 rage so to calculate Rage Per Second (RPS) generated it's;
RPS = 30 / 3.6Working Crit into the RPS gen relatively simple since each crit doubles the rage gen, we can model this as multiplying the whole thing by 1 plus crit rate as a decimal (so 10% crit would be 0.1);
RPS = (30 / 3.6) x (1 + crit)Haste also plays a part as it lowers the swing timer and it does this in the traditional way of base time divided by 1 plus Haste as a decimal;
RPS = (30 / ( 3.6 / (1 + Haste))) x (1 +Crit)
If I've done my maths right that means that baseline an arms warriors hould be producing 8.33 Rage per Second. and we can see that Rage will scale Linearly with Crit and Haste. if you have 100% crit you'll generate 100% more rage than you would if you had 0% crit,Haste is much the same but the two stats multiply the effectiveness of each other so if you had 100% crit and 100% haste you would generate 400% of the rage you would at 0% of both.
Now calculating Rage expenditure is a little bit more complicated to workout. I'm going make some assumptions because; generally Arms follows a 4 GCD cycle of MS>WW>WW>WW, so the average RPS expenditure is going to be 30 rage per GCD, or 20 rage per second, which means with no crit/haste we can use an ability ever 3.6s. if we factor in that Arms will use CS 3 times a minute, this would be a saving of 1 RPS, and if we assume that we Rend every 15 seconds its a further saving of 1.3 rps. Storm bolt on CD is another 1 rps saved, IV on cd 0.666. so the minimum RPS in we'd need so far to be GCD locked is 16,that's almost twice what we generate, so nearly half of our GCDs (47%) will be empty.
The New tier 3 Talents all serve to close the gap between how much we need and how much we spend. TfB adds 1RPS for every target with have Rend on, so on AoE particularly strong, but single target is marginal (though with sufficient Haste Crit could be enough). Sudden Death is unreliable, though should average out to a saving of 30 rage every 10 white swings, as such it should be 0.83 RPS, though it will scale upwards with haste. Slam has the most potential to help fill GCDs but frankly it's not enough. if we assume our rotation becomes MS>Slam>X>slam, where X is CS, Rend, Storm Bolt, IV, or WW or Ravager we're still 0.5rps from producing enough Rage. Of course this is a completely fixed rotation(so managing rage is a non-issue) and makes our tier 2, 3, 4, and 7 talents mandatory Choices. And this is rather a DPS loss, you wouldn't delay a CS by up to 6 seconds, and you wouldn't want to rend that often, you probably want to use Storm bolt inside a CS if possible and it usually isn't a good idea to use IV on CD. And this rotation really begs the question; what is the point of letting Slam stack to 2,. it worked much better when it stacked only once.
Moving on Headlong Rush sounds great, I was really happy when they announced it. Granted the name is a bit bleh, I'd have preferred something like"Quick to Anger" but whatever. As we increase haste it lowers the GCD and this has the effect that it actually increase the amount of Rage that you need to perform the rotation, the the point that until you hit the 1 second GCD cap Haste is actually a net loss in RPS; whilst you generate a Higher RPS through more through white swings, you will now have even more empty GCDs. Unless you maintain certain ratio of Crit to haste, all haste does is increase your white swing damage and insert more empty GCDs between your abilities. Much like Prot pallies, where haste is a direct DPS increase, it doesn't actually lead to less empty GCDs, it just means the rotation happens quicker, which is fine if you only get an empty GCD every 15s or so and have a proc to help fill it (Grand Crusader). When every other GCD is empty, it's not quite so great and this is where prot pallies have a second advantage, almost all of their rotational abilities also have their CDs reduced by haste as well, unfortunately the poor arms warrior whose relying on things like storm bolt, or IV, or CS, or rend whose CDs aren't reduced by haste to fill those GCDs means that he gets proportionally less low cost GCDs per time period as he increases haste.
For example at 0% haste you have 40 GCDs a minute, 10 of which are probably spent on MS, 3(and a bit) on rend, 3 on CS, 2 on IV and 2 stormbolt, leaving 20 GCDs to fill with Whirlwind. At 50% haste we now have 60 GCDs a minute, 15 of which go on MS, 3(and a bit) on rend, 3 on CS, 2 on IV and 2 stormbolt, leaving 35 GCDs to fill with Whirlwind.
This is graph of roughly how much Crit we need to sustain to be GCD locked as we increase in haste. Notice how it's only once we past the 1 second GCD cap that we start to actually not rely on Crit near as much;
The reason this works well for Protection Warriors and Paladins (both Ret and Prot) is that Those specs use their GCDs to build up resources, Arms uses it's GCDs to spend resources, and that's where the issue comes from. Prot, Prot, and Ret's fillers either don't have a cost or the cost is negligible (CS isn't limited by the mana cost). For those Specs it's the order of the GCDs that matter. With the Rework of Sudden Death, Arms is now about dumping as much rage in a CS as possible, that doesn't really play well with the fact that we have to keep ourselves to such a tight rage band; More than 40 risks capping out, less than 30 risks stalling out, and with so many Empty GCDs and such high costs on our Fillers we really are punished for using the mat the wrong time.
It's also worthy of note that unlike Live prot pallies, where haste falls in value once you hit the 50% 1 second GCD threshold, Arms's value for haste sky rockets as you now are actually able to start filling more GCDs.