• Latito

Oh, also.. a while ago during one of my Dire Maul sessions I decided to test some cycles a bit.

I did 10-minute runs, using BF 5x and AR twice each test. I watched my dps on Recount and after 6-7 min it never jumped around more than 10-15 dps. Obviously this is with NO buffs, and no other debuffs on the boss. When doing 3s/5r I was hovering around 1200 dps, when doing 5s/5a (using Ald's abbreviation now) I got 1280 dps. Certainly a difference large enough to rule out some streaks, considering these were over 10 minute periods as well, not just 2-3. This was against the Gorduk Spirits in DM North, can't remember what their armor was at, but its likely around 3-4k base. That was without Imp EA.


So while that is more "anecdotal" evidence - it does match reasonably close to Ald's previous post where non-Imp EA is a personal dps increase over Rupture when no warrior is present. By the math in my first post, Imp EA is a raid-dps increase over Sunder. That said, you CANNOT EA on a boss when a prot warrior is tanking, it just kills threat too much. For druid/pally tanks however, it is worthwhile.

 

• songster

Any DPS increase is better than zero. So you won't have enough energy for a rupture on every cycle - no reason not to throw in a rupture when you do have the spare energy. So your effective cycle could end up something like 5s/5e/5s/5e/3r. Or even longer if that's still not quite workable. I'm honestly beginning to think that the mindset of "must find a cycle at all costs" is going to end up with bad rogues making stupid decisions. No cycle will ever perfectly match your energy income. You can either use the slack energy (and thus not use a perfect cycle, shock horror!) or waste it. It shouldn't take pages of theorycraft to work out which of those two is a better option!

Edited to add: The fact that Expose armor is a better finisher than Rupture on an unsundered mob should not be news to anyone reading this thread. In a raid situation Improved Expose is better than a full stack of sunders for raid dps. However, in practice you can't use it if you have a prot warrior in the raid. Even if the prot warrior isn't tanking, their personal DPS depends on the Sunder stack, which reduces or wipes out the net raid gain.

 

• Latito

Fair enough. Agreed, any dps increase is better than zero. That said, a cycle such as that still requires you to do "5s/5a/3r/5s" at one point in the rotation. Essentially, you must get 13 combo points within 30 seconds. Saving energy and dropping the first 5s while you have ~70 energy *should* likely allow this to happen.. but you could get unlucky of the next 30 seconds with ruthlessness and combat potency, thus dropping SnD (and likely EA 8-10 sec later also). The MARGINAL dps increase of an extra 3-point rupture every minute just doesn't seem worth the risk. We could do the math I guess and see what % chance you have of being able to pull off that sequence, figure out the dps loss if you lose SnD for X amount of time and see if that potential dps loss * chance to happen is less or greater than the extra dmg from rupture.

Agreed, cycles are never perfect - I sometimes find myself needed to cut out a rupture set from my cycles to keep SnD up and adapting to circumstances is what keeps a rogues dps more consistant, as opposed to flopping around and being kinda streaky. It is a balancing act,


On the topic of EA usage.. clearly it was news to Nanomann, and even Ald and myself wanted to double check. It makes sense, but its worth checking. I'm sure there is some level of high AP, low hit, low crit, no haste or arm Pen which makes Rupture better than EA on an unsundered mob. Likely its so damn high that the set of gear which achieves that does not exist, but still.

In a raid situation, I'm not sure how the prot warriors dps is affected. I realize they would no longer be able to Devastate, is this one of their primary moves while dps'ing, or do they pretend to be Fury and just HS/WW? Perhaps some insight from the prot warrior thread would be good. Good thought - it was something I hadn't considered.

 

• Aldriana

I don't see that it's quite that clear cut. First off, the fact that your energy bar is bigger than the amount of energy needed for any particular move gives you some ability to store up energy. The cycles produced by this sheet are average cycles; during some SnDs you'll generate more energy than average, and in some you'll generate less; hence, in order to keep SnD from dropping during low-energy cycles, you need to save up energy (to the extent you can) during surplus energy cycles. Now, this isn't 100% possible, of course - your energy bar is only so big, so there's only so much error you can absorb; this means two things in practice. First, sometimes you will generate more energy than you can store, and thus need to "waste" some SnD uptime by shortcutting the cycle; second, to make sure that you (almost) never have a drought so large than SnD drops, you need to use a cycle slightly longer than the long-term average case to give you a cushion against SnD gaps - which, in turn, leads to more wasted SnD uptime.

So, fundamentally, yes, one tends to have - depending on how you look at it - either surplus energy or surplus SnD uptime relative to the ideal cycle. And, from a purely abstract perspective, it would be beneficial to use this to generate a little extra DPS.

The problem is, it's not always possible to do so. First, there's the simple problem that you need to make decisions at intermediate locations in the cycle - that is, if you try to squeeze an extra finisher in before the SnD, you don't know if Ruthlessness will proc; you don't know how many Combat Potency procs you're going to get in the last 10 seconds of your cycle; if it's not a 5-point finisher, you don't know if Relentless Strikes will proc; and so on. Hence, you are still making a decision with some variance to it, and to avoid having SnD drop (or having to do a very small SnD followed by some ad hoc cycling to bootstrap back up to a normal-sized SnD, which is almost always bad) you still need to allow yourself a cushion - which brings us back to the aforementioned issues. So while you may be able to reduce the amount of wasted SnD uptime through such tactics, you will never eliminate in entirely.

Additionally, there's the fact that in many circumstances, just doing an extra SS or burning the SnD uptime is actually optimal. Determining which ad hoc cycle components actually prove worthwhile is a problem that I don't think most people have a good appreciation for. For instance, it has been suggested that dropping a 3-pt Eviscerate during high energy cycles of 5s5r might be better than doing the standard 3s5r. Now, my question is: is the damage gained by the higher energy efficiency of an occasional 3 point Eviscerate enough to offset the damage lost to reduced Rupture uptime? I would argue that the answer is almost always no; the overall energy efficiency of the cycle element is not enough larger to warrant this approach.

And that, I think, is going to be the key to analyzing these; the notion of energy efficiency of cycle elements. The concept of energy efficiency of individual moves is a fairly standard one by now; the trick is, it doesn't really tell the whole story. What you really want to know is the energy efficiency of a combination moves. Let me give a concrete example of what I mean.

Lets compute the energy efficiency of performing a single iteration of 3s5r. This requires 6.8 Sinister Strikes and does the damage of one rupture. Hence, the total energy expended is (roughly) 40 * 6.8 + 10 (for the chance Relentless Strikes does not proc), and the total damage is 6.8 times the damage of a SS plus the damage of one rupture. Working this out for the information currently in my spreadsheet gives an energy efficiency of about 37.76 damage per energy. Note that we're ignoring the damage contribution of the SnD; for purposes of computing components of ad hoc cycles, we're assuming that due to abnormally high energy regen or whatever, we're able to do this component with 100% SnD uptime; hence, that damage will just subtract out evenly accross all builds.

Now, if we compute the energy efficiency of 5s5r, we find that it's only 37.23, while the energy efficiency of a 5s5r3e component is 37.84.

So: what we learn from this is that a 5s5r3e element is actually more energy efficient than 3s5r (though barely) - however, it is only more efficient by .08 damage per energy, while 5s5r is .53 DPE less efficient; hence, roughly 87% of energy would need to be spent of 5s5r3e cycle elements and only 13% on 5s5r cycle elements in order for the ad hoc cycle to come out ahead - which is clearly not sustainable. So we find that 5s5r with adhoc 5s5r3e is inferior to 3s5r.

Now, I can easily imagine circumstances - gear combinations and the like - where this will not be true; I'm sure there are times that 5s5r with adhoc 5s5r3e is better. The problem is, this level of analysis is not something you can do on the fly; hence, making cycles up as you go is, in most cases, likely inferior. And while I'm sure it would be possible to understand your gear and energy tradeoffs well enough to figure out for yourself when you should do what, it's most defenitely an advanced technique that only the most knowledgeable and experienced of rogues would want to attempt; it's certainly nothing I'd recommend to the average spreadsheet user.

Fundamentally, I guess my point is this: the relative value of Rogue finishers are such that I don't see that a compelling case can be made to do ad hoc cycles; in the vast majority of cases, XsYr type cycles will be optimal for appropriate choice of X and Y and good energy management. If at some point Eviscerate/Envenom/whatever becomes competitive, ad hoc cycles might be sufficiently good to revisit; but I don't think that's the case right now.

If the warrior is tanking, you certainly can't expose; however, the damage that a prot warrior gains from sundering isn't actually that large; even assuming they're spamming Devastate for DPS every cooldown, they're only going to gain 175 damage every 1.5 seconds (further modified by crit chance, hit chance, and armor reduction) - or something on the order of 120 DPS. And I can almost guarantee you that the damage gained over the raid by going from 2600 reduction (Sunders) to 3075 (Imp Expose) is enough to offset both that and the ~100 DPS the Exposing rogue would be doing with Rupture. So if one is in a guild that regularly has feral druids tanking raid bosses, it might be worth speccing Imp Expose, as it is almost certainly worth using whenever a warrior is not tanking, even if there is a prot warrior or two DPSing.

Edit: I'm assuming, of course, that the warrior can still Devastate, which I believe is true; the Sunder just doesn't go off. If this isn't true, the answer may be different... but I suspect the 475 extra armor pen is still enough to cover it, at least in a 25 man.

 

• songster

It is in the case I was talking about - keeping up Expose Armor. A 5s/5exp cycle gives you 100% uptime on both and some spare energy. You have three options for the extra energy.

1) Waste it completely by cutting the S'n'D or the EA short
2) Use it for extra Sinister Strikes, and waste the combo points
3) Use it for extra Sinister Strikes and an occasional rupture.

As long as you don't drop below 100% uptime for S'n'D and EA, solution (3) is clearly the best option, since Rupture has better energy efficiency than Sinister Strike. In practice it's difficult to achieve (3), since an energy buffer of 100 doesn't let you save up enough spare energy to be sure of dropping a sensible-sized rupture.

The 5s/5expose cycle costs 352 energy (8.8 sinister strikes, free finishers), and your expected energy income for a 30 second cycle is ~390, i.e. you have one spare sinister strike per cycle. So if you're thinking in "cycle" terms, that's a 5s/5expose cycle with an extra SS inserted somewhere at random in each cycle.

However, I think the better strategy is to save up as much of the energy as you can, right up to the point where you're about to tick over. Then you spend the extra energy all at once - i.e. rather than doing one extra SS per cycle, you do 2 or 3 extra SS every second or third cycle. If you happen to get lucky with CP procs at that point, it will allow you to drop in a small rupture (better energy efficiency than SS), while if you don't get lucky, it's not wasting anything more than you otherwise would.

Question: What's the smallest rupture that has better DPE than a SS?

 

◊ Vulajin

According to my spreadsheet, a 2-CP Rupture or above will be superior DPE to an SS, but only by ~11. Since this figure doesn't account for Relentless Strikes energy regen, I'd say it's still not worthwhile to perform a 2-CP Rupture, especially since a 2-CP Rupture only gains 2% of your AP per tick, whereas 3 and above all gain 3% per tick. So the smallest worthwhile Rupture is essentially 3-CP.

 

• Aldriana

Agreed, for the 5s5a case, weaving the occasional rupture is probably worthwhile if you can do it. I was referring more to the line

Fundamentally: yes, there are times (such as 5s5a cycles) where it usually makes sense to weave in something if you can. But there's also a *lot* of circumstances where it doesn't, so on the whole I think cycles are a good thing - I believe it is the exception and not the rule when weaved cycles are able to beat static cycles.

 

Ekto

I have a question about a specific cell in the spreadsheet: cell G74 on the Xs5r worksheet. I assumed that each column on this sheet calculates the dps when you add one more of the relevant combat stat. So, for example, column G calculates your dps when you add 1 more hit rating to your stats. If that is the case, shouldn't the last term in G74 have a minus sign, NOT a plus sign. The plus sign increases your miss chance and, thus, calculates your dps if you were to LOSE one hit rating. There is a similar concern with cell I75. I did not check the other worksheets for this possible error, so you may need to fix those, too, if I understand this worksheet correctly.

 

• Aldriana

Adding and removing one of each stats should have more or less the same effect; what you're really attempting to do is approximate the derivative of the function at that point, for which either endpoint approximation is equally valid. For maximum accuracy, we should really be doing a midpoint approximation, but as that would take 2 columns per stat as opposed to one, I figured endpoint approximations were probably sufficient. And as either endpoint is equally good, I can use whichever is more convenient for the particular stat. In the specific case of the cells you mention (which will be the same in all calc sheets) I elected to compare by removing one hit rather than adding one (and doing it in G74 and I75 instead of G7 and I11) to work around the hit cap. Subtracting 13/205 in G74/I75 would risk giving a negative hit score for people at the hit cap, which may potentially cause odd results. Hence, I add in instead so the hit stays in the valid range. Since the actual comparisons are performed against the absolute value of the difference anyway, it all works out in the end.

Note that most of the procs/set bonuses do this sort of thing as well: if you have the buff, they remove it, and if you don't have the buff, they add it. Hence, at any given time, some are being scored by "subtracting" and some by "adding". But as both are equally valid, this is fine.

 

Ekto

ah ok, makes sense then.

In my own spreadsheet, I too have had to deal with the less than aesthetic ways of dealing with evaluating a function at different points. Is there no way in Excel to define some functions or functions in terms of letters and then just evaluate the functions for given sets of data points?

For example, something like "LET(f = 1-h/1577-.0025*e/3.924-.01*p)" where f is supposed to be your chance to miss, with h your hit rating, e your expertise rating, and p your rank of precision (and LET is some Excel function I made up that ostensibly defines f algebraically). Then in some cell specify a range of cells to represent the desired values of h, e, and p. and then in another cell evaluate f for the given data. I suppose that this would require Excel to have some sort of CAS built in, and so if it's not possible, I would understand.

Thanks for the usual speedy response.

[Edit] For my own gear, the dps difference between +1AP and base stats on the Xs5r is ~0.308. The (absolute) dps difference between -1hit and my base stats on the Xs5r worksheet is ~0.795, giving a relative value of hit to AP of 2.584. but the dps difference between +1hit and my base stats on the same worksheet is ~0.117, giving a relative value to AP of 0.381.

Now I know that the sheet is not meant to give a static AEP table, but an AEP table for each data point corresponding to each possible combination of stats. But if the plus/minus signs in cell G74 and I75 are supposed to be equally valid, they should give (at least approximately) the same AEP for a given set of data. Going from 2.584 to 0.381 is quite a difference.

 

• Aldriana

I think there are ways to do such things - in fact, I'm pretty sure the Rogue DPS sheet does them. The problem is that they all involve macros, which starts to cause issues with OO compatability and the like (not to mention being well beyond my knowledge of Excel) which is why I chose not to use them. It's one of the limitations of the spreadsheet format - which brings us back to the oft-mentioned "write this in a real programming language" point (which I'm working on, I swear). Basically, doing things in a spreadsheet makes the UI work a lot simpler but it does put some fairly heavy limitations on your ability to evaluate certain things.

 

Bluefish

I'm having trouble convincing the Balance Druid community here on EJ that Imp Faerie Fire is worth spec'ing into. The Druid themself loses a significant amount of either DPS, threat reduction, or sustainability to spec into it, but according to the Gear Sheet, the benefits are pretty crazy for those that aren't hit-capped. I show 1491 DPS with WF / LotP / Imp BS / Might / Kings / Drums / Enh Shaman, no CoR. When I compare each of these buffs to Imp FF, it comes out ahead of Might, Windfury, even LotP! Can you confirm that the Gear Sheet is properly modeling Imp FF? If so, could you possibly stop by the Balance thread and let them know?

 

• Aldriana

Well, the thing to be careful about is that while Imp FF does come in ahead of those things, the marginal difference between FF and Imp FF is somewhat smaller.

Using, for instance, the stat weightings Vulajin posted in the Roguecraft 101 thread:

Imp FF gives 3% hit ~= 47 hit rating ~= 2.3 * 47 = 108 EAP, plus 610 armor reduction = 610 * .29 = 177 EAP, for a total of 285 AP - which is more than Blessing of Might, almost as much as Battle Shout, and more than the 5 * 22.1 * 1.75 = 193 EAP one gets from Leader of the Pack.

The thing to be careful about it: regular Faerie Fire already gives the armor pen, so speccing into Imp FF gives roughly 108 AP *additional* benefit beyond what we'd otherwise get from the druid. Hence, each talent point gives something on the order of 36 AP benefit per rogue, which, in practice, means about 11 DPS per rogue per point. So if you have 3 rogues, it gives a benefit of roughly 33 DPS per point, or about 100 DPS for all 3 points... plus whatever it does for the hunters, dps warriors, enhancement shaman, etc. Also note that if you're tank isn't hit capped (and a lot of tanks aren't) it also significantly increases the tank's aggro.

So, long story short: the incremental benefit from Imp FF is not quite as large as some of the others you list; however, it's still a very helpful talent for all your melee DPS and probably some of your tanks as well. I can easily see 3/3 Imp FF increasing the tank's aggro by ~4% and increasing raid DPS by ~200. How that compares to the other talents you could be taking, I couldn't say... but it's certainly a very respectable talent from the perspective of the rest of the raid.

 

royaljester

And let's be honest, balacne druids aren't there for their amazing dps and sustainablity. They are there for the 5% spell crit (amazing for pallies/mages omg) and possibly the imp FF. I think the biggest benefit comes from being able to reitemize your gear with that druid using some off peices with less hit and stack more ap/crit. I know hunters are able to drop talent points and get some other talents.

If the druid is worried about losing pts on his own dps/survivability/sustainability, rather than contributing the raid buffs....yeah, probably shouldn't be raiding.

 

Khrashz

Minor issue (oh hi I rerolled rogue from enhancement shaman):

Fool's bane is missing under maces. I know it's not exactly a prime choice for raid dps, but .... it dropped and Spiteblade hasn't.

 

• Aldriana

Yeah, there's a couple weapons missing - specifically, the socketed ones. Weapon socketing isn't working, and given that there's a grand total of 4 socketed weapons that anyone would think about using raids, and all of them prove to be fairly mediocre for when you can get them, it hasn't been a priority to fix so far.

 

◊ Shaker

Fool's bane should be slightly behind Spiteblade, and a little ahead of Big Bad Wolf's paw. It's a fair chunk ahead of Vindicator's Brand, iirc. Honestly, just copy the stats from another mace weapon and hardcode in the stats (with or without socket bonus), and see where it lines up.

 

roq

I was wondering if you guys could help me figure something out, because I haven't been able to figure out an answer:
Wow Web Stats

This is our Morogrim Attempt last night and I did 1588 DPS. Now that may not sound weird to most of you guys, but it is for me. I even had 2 Graves. If we look at the spreadsheet, I should go no higher than about 1370 with the buffs I usually have, but I beat that by about 200 DPS. We don't Use CoR and from the WWS, it doesn't even look like we had FF up. We don't Shaman swap or anything special to up our DPS.

The only difference in what I did this week than what I usually do is, I used AR twice, because I used it earlier than i usually do in the fight, but will that make up that much of a difference?

My Group: Enhancement shaman, Arms Warrior, 3x Rogue.
Buffs, that may not be standard: 5/5 Shouts with Solarian's Sapphire, Imp Might, Imp Hunter's Mark, Expose Weakness (~1050 AGI).
I did use one haste pot, but no food, no Elixir/Flask, no Drums.
Gear: Armory, i may have my riding Trinket on, but i use WSC/DST

One note as a whole Morogrim died alot faster than I have ever seen him die before in our guild at least. As you can see in the WWS, most DPS classes are about 1K DPS and that never happens.

 

frozenkex

I dont know if this was brought up before, but id like someone to tell their opinion on how envenom would perform in DPE compared to rupture using t6 equivalent gear, having 4/5 vile poisons (20/41 spec) AND considering each envenom is occassional and performed ONLY when Stormstrike (20% more nature damage)debuff is up, which is tricky , but possible.

 

Left

Do the calc. It's going to depend on gear and stats. A lot.

For the sake of argument, let's pick 2500 AP, 35% crit. (IE, high end stats.)

Both rupture and envenom are not mitigated by armor. We'll evaluate for with or without mangle up, as well as with or without Stormstrike

Rupture
Damage = 1000 + 0.24*2500 = 1600
Damage w/ Mangle = 1600*1.3 = 2080
DPE = 1600/25 = 64
DPE w/ Mangle = 2080/25 = 83.2

Envenom
Damage = (900 + 2500*0.15)*1.16*1.35 ~= 1997
Damage w/ Stormstrike = 1997*1.2 = 2396
DPE = 1997/35 = 57
DPE w/ Stormstrike = 2396/35 = 68
But, you also lose your Deadly stack for a few seconds, which is a big chunk of lost DPS.

In short, the DPE is marginally better for Envenom if Stormstrike is up but Mangle is down, but you lose your Deadly stack. Losing your deadly stack is a huge chunk of lost DPS, but it might be ok in situations where you are supposed to "stop DoTs" or if the mob is going to die. (Although, if the mob is about to die you shouldn't Rupture anyway.) The comparison is actually more favorable for Envenom with T5 2pc, since you tack on an extra 200 base damage to Envenom.

If Envenom crits, it is worth it, as the crit more than makes up for the lost DP stack. So Mutilate rogues, for example, can use Envenom to great effect coupled with Cold Blood. (Anecdotally, on a Netherspite kill I used Envenom to wipe my DP stack after holding the blue beam for nearly a whole phase. It crit for 13k.)

 

• songster

Extremely poorly. Envenom is worse than Eviscerate, which is in turn worse than Rupture. And you should bear in mind that taking the Stormstrike proc (which is pretty much impossible in practice) doesn't do as much as you think for raid DPS, since you're reducing the DPS of whoever would have got the Stormstrike if you didn't.

Use Envenom as the last finisher on a dying mob, if you don't need to refresh your S'n'D. Ignore it at all other times. Unless you're Mutilate, when the higher crit and faster poison stacking may make it useful.

 

• Latito