I didnt say it's for my current gear , I have head and shoulders to change and im doing dome tests so i was wondering what my rotation should be.anyone ?

It's worth noting that the rupture/evis decision not only depends on gear, but also on raid buffs. In particular, if your target doesn't have mangle/trauma up (either you don't have it in the raid, or the only person who can apply it is doing something else), you may well find that it's worth swapping to the evis cycle with only 2 ToC items.

Side note about engineering since I'm already posting -- don't forget that it can hit more than one target, so in fights with more than one mob, bombs often have a correspondingly larger benefit.

Yup but my question is : once you get to the point that Evis only will make more damage than Low/High rupture cycles , is it worth eviscerating with 5 cps or also with less?

Like the majority of finishers, 5 pts is what you want to do if you can maintain snd uptime.

Eviscerate damage scales linearly with combo points last I checked, atleast for the level of AP I had when I looked at this. Although the larger eviscerate you do the less energy is costs through the talent Relentless Strikes (on average). So its not a straight up damage nerf, but you are spending on average more energy per eviscerate if you are always spamming 3 point finishers.

The eviscerate playstyle is like it has been said before, all about getting the largest finishers off you can, but trying not to waste energy/duration around your SnD refresh, while keeping it up all the time.

The only added twist to this is that for combat, everyone uses a glyph of sinister strike, which makes it so that each sinister strike gives you close to 1.5 combo points per application. So on average going from 4 combo points to 5 combo points, you are wasting 0.5 combo point, which should be taken in consideration. For mutilate, the waste is even higher.

Last I checked 4+ rotation is not statistically different in terms of dps from 5-combo point finisher rotation for combat. For mutilate, I checked in the past that 4+ rotation strictly beats 5-rotation. 4+ means to use a finisher as long as you have 4 or more combo points.

Eviscerating at 4 combo points is less efficient. I took the gear I was already wearing and adjusted the spec to use eviscerate talents and glyphs, and it's showing a 1542 efficiency for 5 points, and 1339 efficiency for 4 points (this is adjusted damage per combo point. adjusted damage accounts for the energy cost.)

Let's assume 50% crit rate. It's a nice number.

If I eviscerate every time at 4 or more combo points, I have about a 20% chance of skipping from 3 to 5 and getting a 5-pt anyway. I used Markov chains to calculate this and I will spare you the details. So your average eviscerate efficiency for this strategy would be about 1380, and no combo points would be wasted.

If I eviscerate only at 5 combo points, my average efficiency is obviously just 1542. I have a 20% chance of "wasting" a combo point every time I get to 5. So .2/5.2 combo points are worthless. Only 5/5.2 get used, so multiply this ratio by our efficiency and we end up with 1483 which is still way ahead of the 4 point eviscerates.


Basically, you shouldn't ever use a 4-pt eviscerate unless you have to dump your combo points fast because the target is dying in a matter of seconds or you have to switch targets immediately.

EDIT: Ok I ran the simulation of combat dps with what I consider BiS in 3.2.2, which is 15/51/5, with double wound, eviscerate-only rotation at 1000 iterations, and 5 combos did come out ahead of 4+ by 40 dps which was statistically significant with p-value of less than 0.1%. Standard error with this many iterations is around 6.6 dps.

So, I can't totally swear where he's coming from here, but the language does seem familiar - some of it appears in my spreadsheet - so I'll take a few minutes to define how *I'm* using the terms. If this isn't what Dinadin is talking about, I apologize, but it seems pertinent to the topic at hand regardless.

Basically, the philosophy is simple: defining the optimal usage of energy and combo point usage is straightforward - ideally, SnD has 100% uptime with no overlap, and is refreshed as efficiently as possible; if rupture is worth using, it, too, has 100% uptime and no overlap, and is refreshed as efficiently as possible; any additional CP are spent on the most efficient size of Eviscerate; and all energy not spent on these finishers goes to Sinister Strikes.

The right way to think about this thus as follows: energy is converted to damage by sinister strikes at some baseline efficiency; combo points then give bonus damage above and beyond this level. Hence, the place to start it to consider the damage per energy efficiency of Sinister Strike in the absence of finishers or combo points. Well, this is pretty straightforward - it's just the damage of a sinister strike, divided by the energy cost of a sinister strike. So, for instance, with the default gear in 1.3 beta 4, if you open up the calcs sheet, row 390 (among other places) has "Baseline DPE", which has value 124.2.

So, now we can calculate how much additional damage we get from an eviscerate - for instance, a 5pt eviscerate costs 10 energy (after ruthlessness), so we need to subtract 10*124.2 damage from the tooltip amount - hence, the actual average damage done by a 5pt eviscerate comes out to 10010.71 (row 398), reduced by the energy cost, to get the adjusted damage of 8768.7 (row 412). However, what we want to know is damage per combo point, so we need to know how many combo points that Eviscerate "costs". Well, we spend 5 on it, but we also get .4 back (on average) from 2/3 Ruthlessness. So the cost is 4.6, right? Well, no. Because the 10 energy cost, had it been spent on sinister strike, would also have given combo points. Hence, we need to account for this as well, to get a true CP cost of 4.91 (row 419). We can then use this to compute the damage per CP we get from a 5pt Eviscerate, which we find to be 1785.79 (row 426). Since we're using the "most efficient" size of eviscerate as a CP dump, we compute this value for all sizes of eviscerate and find that 5pt eviscerate is indeed the most efficient; hence, the marginal value of a combo point, fully realized, is 1785.79.

We then adjust the damage per energy for the fractional combo point loss, so as to avoid needing that last step hereafter; a SS gives 1.X combo points for 40 energy, so we increase the DPE by 1.X * 1785.79/40 to get the true DPE conversion of 179.61 (row 430).

We now know the value of a point of energy, and the value of a combo point; using this, we can find the "true" damage of rupture. Rupture does however much damage, costs some energy, and some combo points; we thus adjust the damage by these costs to find the increase in damage we get by using a rupture instead of an eviscerate. And what we find is that the most efficient rupture is a 5pt rupture, which does -2843.02 damage (row 459), and thus averages -177.69 true damage per second (row 466). Thus, it's a damage loss to use rupture - which is why an evis-only cycle is recommended.

Similarly, SnD costs some amount of energy and CP per time to keep up, so we can compute the maintenance cost. And we find the cheapest way to do so is with 5pt SnD, at 264.99 damage per second (row 493) spent maintaining it.

So, with these numbers, we can compute the damage lost to performing any given move. For instance, a 4pt SnD has an efficiency of 276.49 per second, and a duration of 26.5 seconds (allowing a half second of overlap for human reaction time/latency), and is inefficient to the tune of (276.49-264.99) = 11.5 damage per second, for a total damage loss of 11.5 * 26.5 = 304.7 damage. Meanwhile, Sinister Striking at 4 CP costs .24 CP (row 370), for a damage loss of 430.53. Thus, it is marginally more efficient - by 125.83 damage - to use a 4pt SnD rather than sinister striking to make 5... assuming both waste the amount of SnD uptime. However, in practice, building the extra CP often allows you to wait a bit longer to refresh and thus waste less SnD uptime - and since SnD uptime costs 264.99 damage per second, provided we gain at least 125.83 / 264.99 = .47 seconds by so doing, it's worth it to do a 5pt SnD.

With eviscerate, on the other hand, the situation is a bit more clear-cut - a 4pt Eviscerate has an efficiency of 1538.99 and costs 4.07 CP, hence wastes (1785.78-1548.99)*4.07 = 1003.34 damage, so it's clearly better to perform the extra SS than it is to do a 4pt Eviscerate. The numbers would work out similarly for Rupture were it actually worth using. Hence, my earlier statement: 5pt damage finishers are usually superior (or rather, they are with every gear/spec combination I've checked), but the situation is a bit more ambiguous for SnD.

If SS were worth more CP -- say, in the case of Loatheb, where you get the spore buff and have essentially 100%+ crit chance, ultimately increasing how often you got that extra CP -- this would change, correct? It's not really relevant anymore, but on that fight in particular I've often played around with 4+ finisher cycles just to see if I could tell one way or another if it was more effective. There's always been too much variance in raid composition to know, but I do wonder if the math supports it.

Kind of

Specificaly what Aldriana's talking about would be a combat rotation, not mut, for mutilate the formula would change a little bit, for one thing you'll have 3/3 Ruthlessness instead of 2/3, but you would also generate extra CP from Seal Fate (which is I think what you're asking about)
So no for Combat the numbers shouldn't change, for Mutilate yes they'd change based on how much you're criting on your CP generating abilities, remember if you're trying to apply the formula to Assassination, you'd also need to account for Mutilate generating 2 CPs

I understand the logic of the calculator approach, but it only makes me appreciate the simulation approach more. Simply because I do not have to assume the seamless flow of combo point generators and finishers into a rotation with 100% snd uptime and 100% rupture uptime, if it's worth using. If you get extra energy from using a different finisher or using a higher combo point finisher, the DPE of that energy is worth a different value each time, depending on how many combo points you have relative to snd and rupture duration. For instance, if the extra energy allows you to perform an extra sinister strike and apply a higher-combo point snd when it's due to expire, the value of that is in turn determined by how much time you gain and how long you can delay using snd again. If you can perform an extra sinister strike and apply a higher combo point rupture, the cost of that extra bit of energy in the beginning is going to be different. In some situations it's just a combo-point adjusted damage of a sinister strike. All the math above is a model under several simplifying assumptions, which I admit works well enough because the results of the simulator and calculator are usually close enough, so it gets the job done. I've seen though when the value of a stat, such as expertise for instance, was different in mutilate simulation because in reality it could lead to a cycle disruption, but in the spreadsheet, at least at some point, the cycle was assumed to go without a disruption (i.e. expertise cap was simply assumed). Whether average-rotation approach works well under all circumstances, I do not know. There can be some rotations that can be written in a calculator that will end up being purely theoretical and not sustainable in real-life combat. High-rupture rotation appeared to be an example of that because it still assumed snd was never dropping.

But that's the point: this is an ideal cycle, and not what we're actually basing damage calculations on. It's clearly totally unobtainable - and it's not like we can even really try. But that's not why it's useful. It's useful because it allows us to assess our real-world options in terms of how much they deviate from this ideal cycle, and thus figure out which of the available (not optimal) options are least harmful. We can compute how much damage we lose to rupture downtime, we can compute how much damage we lose from undersized finishers or early SnD refreshes - and thus, we know which option to use.

I *really* don't want to turn this into a simulator versus calculator debate, because I'm totally sick and tired of the argument and I think it's a waste of time. But just to illustrate the point:

If, in a simulator, you get to the situation where you have 5 CP and have pooled as high as you deem safe, and there's 7 seconds left on your SnD, do you eviscerate or SnD? What about 5 seconds? 3? Presumably you define some rubric, and play with it till you give something that seems to give optimal results... but what happens when your stats change? Do you reoptimize it?

With these sorts of calculations, it takes out all the guesswork out of cycle optimization. We specifically compute what our move priority is and when to make these transitions, and thus can model the actual move counts. Admittedly, depending on cycle complexity it may work more or less well - for instance, I'm very confident in my Evis-only numbers, but less so in my high-rupture numbers. But I think in either case it probably gets closer than simply guessing at what the optimal move priorities should be.

There's nothing unsustainable about "high rupture" in practice, the only question is about what you give up for it. Even without any raid buffs, it's pretty easy to keep snd up 100% of the time and have a new rupture ready to go the second the old one drops, as long as you aren't trying to weave in any eviscerates. If rupture did 20,000 DPS, this is pretty much exactly what rogues would do in practice because the minor extra damage from getting an eviscerate in would be greatly outweighed by the off-chance that you let rupture drop for a couple of seconds.

Rupture obviously isn't 20k dps, so the approach needs to be more nuanced. Perfect play instead involves weaving some eviscerates in, and the question is how aggressive you should be about trying for them. The difference between "low rupture" and "high rupture" (as far as I know) is precisely that. With "Low Rupture", what the spreadsheet is telling you is that the benefit of an extra eviscerate outweighs the cost of allowing rupture to drop for a few seconds here or there, whereas "high rupture" suggests that the cost of letting rupture drop is rather high, so you only want to try to toss in an eviscerate if you're sure you'll be ready to put rupture back up when it falls off. Yes, there's some RNG involved, but you have a few seconds to make a decision. If you got to pool before putting rupture up and with 16s left on rupture you have 5cp, then you're in great shape for tossing in that evis. If you barely got the last rupture off in time, however, and as 12s left you only have a few cp, refresh snd and make sure you have the cp for rupture later.

Partly. I think the key difference is more along the lines of the following: High Rupture prioritizes rupture uptime over all other options (save SnD uptime, of course). Low-rupture instead prioritizes efficiency - it attempts to minimize SnD waste, combo point waste, and so forth. What this means in practice is that you have a lot more surplus combo points to spend on eviscerate, but you can't pull all the pooling and truncation tricks necessary to maximize rupture uptime.

That's what has to be done, to be as accurate as possible, and instead of ad hoc rule, you will get something that will work best most of the time.

As for sustainability, I definitely meant without going into options that make you lose dps, so I was thinking of situations when high-rupture cycles (90+) allowed you to gain dps.

When using 4pc T8 rupture is a very big part of your dps. That is why the high rupture cycle is recommended. I would guess, adding even more rupture damage to the picture by picking up serrated blades just adds that much more.

With 4pc T8 and 15-51-5 spec... MH dps is 3985, OH dps is 1530, Bleed dps is 647, and Magic dps is 1291. If I switch to 07-51-13... MH dps is 3791, OH dps is 1568, Bleed dps is 844, and Magic dps is 1284.

MH dps - 3985 vs 3791 = -194
OH dps - 1530 vs 1568 = +38
Bleed dps - 647 vs 844 = +197
Magic dps - 1291 vs 1284 = -7

Total difference = 235 - 201 = 34 dps gain by using 7-51-13

However, I believe in the old spreadsheet (2.0 or whatever)... running the 7-51-13 spec was worst, beacuse it was harder to keep a high rupture rotation going. The cycle of 5s/5r/5e would be broken.

When I was theory crafting to shift my Rogue to T9, I got pretty much the same set of gears (except the trinkets and hardmode gears), but I was wondering about the Mace Spec. I know if the talent offered a static 15% ArPen it would be unrivaled when comparing to other weapon specs however it is listed as "up to 15% ArPen reduced" so wouldn't this pose to be an issue to be analyzed? Because it is not a static 15%, it could mean that you'd get 0% from one hit and a max of 15% the next hit, so what would be the average of this talent? If someone could tell us how good it is, it may become worth while to use.

Not something you need to worry about; the phrasing on the tooltip for armor penetration is also "Enemy armor reduced by up to X%". This is because of the way armor reduction is calculated and not due to a random value calculated each swing. As I understand it ArP would allow you to ignore X% of the remaining armor (post debuffs), which, depending on the debuffs, may be considerably less than X% of the initial armor value value.

Mace Spec is very valuable right now; in fact, it always has been, but Ulduar lacked the maces to make it interesting. That said, with the ArP nerf in 3.2.2, Hack n' Slash starts to pull out ahead as highest DPS; partly due to the OH interactions and partly due to the itemization.

Not quite. On low-armor targets (say, clothies in PvP), you do actually reduce the full percentage - that is, 100% ArPen will reduce the target to 0 armor. Against higher-armor targets - such as most raid bosses - there is a "cap" somewhat lower than the total armor of the mob that limits how much you can reduce the armor of the opponent; thus, a given amount of ArPen reduces the armor of the target by a percentage of that cap. Hence, 100% ArPen won't actually reduce 100% of the opponent's armor - it will still have some small amount less - hence the "up to" in the ArPen wording.

The salient point here, however, is that Mace spec works exactly like all other ArPen - it reduces armor by 15% of the maximum amount that can be reduces.

However, it's also worth noting that that gear list is relatively out of date - following the 3.2.2 ArPen nerf, axe setups will beat mace-dagger setups both for softcap/Mjolnir setups and straight hardcap setups. And a handful of the other slots change for related reasons. Hence, I wouldn't focus on that particular list too much - it's still not a *bad* setup, but there are decidedly better ones.

Ive been doing alot of reading here and elsewhere and Im trying to clarify what i need to get maximum dps as mutiliate. I need to understand the acceptable losses vrs possible gains.

first assuming no additonal raids buffs
Hit Cap = 315 or %10 , acceptable loss , your regular melee attacks will miss approx %10 of the time on bosses since reaching the 700 plus cap is not feasable I understand that and made adjustments to get closer to 315.

Expertise Cap = 26 ? Im currently at 42 and still getting parried about %3 of the time on regular melee Im sure i wont be dodge or parried at %11 . Is the 26 exp cap suggesting i take even more dodged and parried melee attacks ?

So if I dropped that expertise to 26 i see about a %6 parry on a boss.
Is this an acceptable loss and would i then stack attack power or agility instead to improve overall dps.

Help me understand if this Loss\Gains info is correct

Thanks

Correction: I tested again and 26 exp shouldhave no dodges or parries on reg melee but i will miss %10 of the time . Is that about right ?

The Expertise 'cap' to which rogues refer to is the number at which dodge is eliminated from the bosses attack table. Parry can be eliminated by simply standing behind the mob. If you are being parried by a boss as a rogue, you are, 99% of the time, Doing Something Wrong.

Similarly, while the poison hit cap is 315 totally unbuffed, it is generally pretty reasonable for most rogues to expect Misery or Improved Faerie Fire in their raid and thus, tend to gear around the assumption of those buffs - a reduction of 3% in the cap to 237. The reason for this is because the DPS loss of being geared/gemmed for hit above 237 when the debuff is present is larger in most cases than the DPS loss of being below the cap when it isn't. Your mileage on this one may vary though, and of course, you should always gear around the raid circumstance that you are most likely to find 'typical'.