Optimal rotation; theory vs reality - Page 2

Morwen · 11/10/08, 4:38 PM

Originally Posted by vassleer

Affliction Warlocks are likely the ones with the "hardest" rotation now.

Affliction is 'hard' in a sense, but it is not hurt much by movement/pauses.

As an affliction warlock it is possible to be within a few percentage points of optimal dps even if you consistently mess up the timing, because every ability is on a relatively long timer. When your concentration slips and you keep losing 2 seconds of uptime on a DoT that lasts 15 seconds and accounts for 10% of your overall dps, it is no big deal.

By comparison, if you are, say, a ret paladin, then your main abilities have short timers between 6-10 seconds, so you are going to be hurt more if you keep losing 2 seconds.

vassleer · 11/10/08, 5:55 PM

While I agree that the dps loss of a single instance of a Ret paladin clipping the last few seconds off of one of their effects will surpass that of a warlock clipping the last few seconds off of UA or SL (perhaps even CoA), I think it much more likely that the this would occur more often to the warlock juggling 6-7 spells/effects then the Ret. I also strongly disagree about the affliction lock not being hurt much by movement or pauses.

Much of Afflicition's dps is made up of the synergy between Haunt, Shadow Embrace and the remainder of their dots. Allowing either Haunt or SE to fall off their target results in a significant dps loss to the lock, and will take them some considerable time to replace - resulting in a dominoe effect of additional dots "falling" off while they attempt to fix it. The Ret pally may have all their effects fall off, but they can more easily (and quickly) reapply their effects and return to their optimal dps rotation. (Ignoring for a moment the differences in movement in various fights between melee and ranged dps).

The analogy I like to use when I describe to my friends the difficulty in playing a quality affliction lock vs. other dps classes is that of a juggler. Imagine juggling 3 balls and getting the timing down to be able to keep all 3 balls going and not let any hit the ground. Now throw in 3-4 additional balls and you have the affliction lock. Due to the differences in dot duration, cast time, and especially priority (as Haunt + shadowbolt to keep the Haunt and SEx2 debuff is of primary importance) you have to work just to keep all the dots running - forget about worrying over hurting your dps by clipping a few seconds off of UA, SL, Immolate (if you use it) or god forbid CoA. A few seconds of movement makes it even worse as the lock has to scramble to recover while the dots falling off are causing a dominoe effect that can require the lock to simply start their rotation from scratch in order to quickly get them all back up.

All of this seems to lead to that "tunnel vision" effect on locks that I mentioned previously. Their focus on keeping their dots and maintaining their optimal dps has made it much more difficult to pay attention to their environment - causing them to violate one of the primary tenets of raiding. That being don't stand in the pool of stuff that is killing you.

Montegomery · 11/10/08, 7:39 PM

Tunnel Vision is a problem, but one would expect the typical hardcore raider to quickly learn the flow of your rotation or priority queue. It's a lot like learning to drive, at first trying to drive the speed limit without staring at the speedometer seems impossible, but eventually judging your speed becomes so second nature that you only peek at the speedometer as an occasional reference point. Granted, this happens faster and to greater effect on a class with a simpler rotation.

It's my experience that the people with major tunnel vision problems will suffer from them regardless of what class or spec they're playing. There are plenty of people whose blinders are just as big when they play Destruction as when they play Affliction. The latter spec certainly doesn't help, but it isn't the root cause so much as a factor that exacerbates the situation.

Shadewalk · 11/10/08, 8:26 PM

Another point about rotations is how badly some, very notably pre 3.0 hunters suffered from lag. As little as going from 100 to 200 to 300ms could cause very significant decreases in dps as the old rotations simply fell apart. This is also important when hammering a rotation to match the GCD, quite sure the old warrior rotation would have to varry a little depending on the latency by dropping a devistate if you lost enough time every rotation, but in some ways this just caused a painful adaptation of the rotation to fix as possible.

As said though the game seems to have moved on a bit from simple rotations even if only by making the theroretical rotations longer than the stationary burst dps times on fights. The introduction of powerful procs can ensure that a rotation cannot be held even for a short while, as I'm sure many warriors have found out now with all the free shield slams even in a stationary tank and spank.

Another point has got to be used CDs like Inner focus. Theoretically I want to use this just after a clearcast proc to ensure that I can escape the 5 second rule for a moment and futher if lucky get another proc given the extra crit to help. In practice though I tend to be dropping renews and PoMs after my flash/greater heals and cannot really adjust much at all to make good use of what should be a few extra seconds of mana.

Nethris · 11/10/08, 8:35 PM

Originally Posted by Darian_TruBlade

Tunnel Vision is a problem, but one would expect the typical hardcore raider to quickly learn the flow of your rotation or priority queue. It's a lot like learning to drive, at first trying to drive the speed limit without staring at the speedometer seems impossible, but eventually judging your speed becomes so second nature that you only peek at the speedometer as an occasional reference point. Granted, this happens faster and to greater effect on a class with a simpler rotation.

It's my experience that the people with major tunnel vision problems will suffer from them regardless of what class or spec they're playing. There are plenty of people whose blinders are just as big when they play Destruction as when they play Affliction. The latter spec certainly doesn't help, but it isn't the root cause so much as a factor that exacerbates the situation.

If the affliction rotation is enough harder that there are a non-negligible number of people that do develop tunnel vision, or it takes much longer on average to get comfortable enough with affliction to escape tunnel vision, then since giving affliction higher dps to reward pulling said rotation off would violate the balancing they've been working on of having most classes and specs more or less equal, that's a strong argument that either affliction needs more X refreshes the duration of Y talents, or making some of the debuffs involved exclusionary and buffing individual damage if necessary to keep damage throughput about the same. Variety is fine, but it sounds as though affliction is becoming rather extreme in how much extra complexity it has, to the point of being a detriment to the spec - I believe Blue posts have indicated Blizzard is looking at that and seeing how it plays out though.

One interesting tangent of this discussion of course is how the disruption of PvP, from opponent actions, the need to use crowd control, etc, affect different rotations - if 2 classes/specs do similar raid damage with their optimal rotation, the one less disrupted by PvP is going to do more damage in PvP, especially if the definition of disruption is extended to "necessary" talent changes. Since Blizzard wants most/all of the dps specs to be viable in raids and most of them to be reasonably viable in PvP, and class/spec variety leads to some specs being more disrupted by PvP than others, this leads to some of the problems with PvP vs PvE balance that occurred in the past - to some extent, Blizzard may be shooting for all specs to be so totally disrupted by high level PvP that it's almost entirely a separate balance, though that seems unlikely to exactly occur.

Lucinde · 11/11/08, 7:44 AM

I think some players overlook how useful abilities and talents you can use while moving are. It's easy to look at a spreadsheet assuming you're fighting patchwerk and make all of your decisions from that.

For moonkin in particular, many people skip typhoon and starfall when these are significant damage boosts since you can use them while moving.

I just want to emphasize this a bit more, because as said before, in practice rotations or even spell priority systems get disrupted by a boss' abilities.

For example, on A'lar I used to skip refreshing SW:P and hitting SW:D close to it moving to the next platform, so I could use the time I spent moving casting instants. Similarly, on Illidan when he was being kited away from the blue flame circles (forgot the name), I used to "safe" SW:D so I could its GCD to move back in MF range.

Also, on bosses that have adds, it's usually better to refresh your DoT's instead of casting high prio spells right before the adds come up. Or there's Lurker where close to it submerging it was kinda pointless to refresh DoT's and more useful to get in another MF or MB instead.

There are countless other examples where you would choose to cast a spell of a lower priority than the highest one available because of what will happen in the next 20 seconds. In fact, you probably spend more time "managing your optimal rotation" than actually performing it. I think anticipating to that kind of situations is what seperates the good players from the bad.

Everyone can learn and execute a rotation within 95% of perfect on Patchwerk. Maintaining ~85% of the DPS of that optimal rotation during a real fight where targets change, non-rotations casts are neccesary (think dispelling, combat res, interacting with fight-related mechanics) or you need to move at semi-predictable intervals cannot be done by everyone. Those that can are the ones on top of the charts and those that get assigned to do the non-standard stuff (ie, handle core dumping on Vasjh). Those are good players.

Glodd · 11/12/08, 12:00 AM

Originally Posted by Balog

Moonkin were actually what prompted my post. In their thread I saw a lot of people skipping AoE or regen talents to try to work in Eclipse based on how much of a DPS boost it would be while up. I've also seen a lot of reports from people actually raiding that the actual benefits are not as great as the theoretical due to proc time being lost to other things. I was wandering if other classes/specs with proc watching rotations/priority systems had the same experience; and if so, would that devalue the proc talents, or at least make regen/survivability/utility talents more wortwhile in comparison?

As a 3.0.3 Marksmanship hunter, the only proc watch I have is Improved Steady Shot, which gives extra damage on the next Aimed Shot or Chimera Shot. I don't really have to think about it, because at least Chimera is always part of my rotation.

Also, with regard to DPS while moving, that is why I took Aimed Shot in 3.0.2, and it is why I am using it still in 3.0.3. That extra instant shot really helps a lot.

PSGarak · 11/13/08, 11:46 AM

Warlock rotations were harder to maintain in TBC, but they were also much more resiliant to movement than most rotations. Lifetap basically means a warlock has two filler spells instead of one. We have some ratio between the two we have to maintain, but there's an amount of flexibility that's basically unseen elsewhere in the game in brick-wall min-max DPS scenarios. Pretty early on, warlocks learned that what separated the good warlocks from the ok ones was timing lifetaps to coincide with movement. However, we probably suffered the worst from interruptions that stop casting altogether, like stuns, fears, silences, or phase transitions. As far as I can tell, in TBC we had the third-highest ramp-up time after mages stacking scorch/WC and shadow priests stacking shadow weaving (which we also had to wait for), but our debuffs took less time to drop off. In WLK we have the overall highest ramp-up time.

It's interesting to look at the graph of DPS vs difficulty. Affliction warlocks and shadow priests were widely cited as being the hardest classes to play in TBC, but I think that's slightly inaccurate. At the high-end of min-maxing, mages and rogues took just as much precision, control, and awareness to reach their top DPS. The difference is the dropoff; an affliction warlock that was merely ok could easily lose a third of their DPS or more, while a combat rogue that just hit SSx5+S&D could hit something like 93% of the theoretical maximum (maybe a little more with cooldown stacking), and the differentiation in mage skill was mostly in cooldown-stacking.

I think the fallout from this is that the classes that get labeled as "easy" are not the ones where you can hit your max DPS with a paperweight (destro locks) since that paradigm is basically dead. The beginner classes will be judged by how close you can come to your max DPS with minimal effort. Frost mages, combat rogues, and destro locks I think still fall into this category, although none of those are the highest-DPS spec of their respective class so it's not a huge issue, and the gap is generally a bit wider now than before. For mid-level guilds, there may be actual tangible benefits in going these unoptimal specs because it frees up some attention for the encounter, depending on the encounter and the actual DPS loss.

kemikalkadet · 11/15/08, 9:46 AM

Originally Posted by thorpac

The key is not knowing "what", but "why". If you understand why your rotation is set up a certain way, you can prioritize your next move when you are interrupted.

This is exactly right, and it's the difference between those that simply read basic theory to get an optimal rotation and those that really read into the subject and question why each move is done and why its done then etc. Someone that really knows and understands their moves will, when things get out of whack, know they should skip X move/spell and jump back into their rotation at Y.

I think the best way to go about it is to prioritise all your moves, so in a 6 move rotation there are 2 that you absolutely must do whenever they're up, 2 that are important but if you can only fire off one move in a particular period you'll skip for the top moves, and 2 that will always be the first to be left out. A very basic hypothetical scenario of course but that's a simplified view of how i look at move prioritisation.

Shaewyn · 11/15/08, 12:51 PM

I think that (perhaps obviously), getting closer to an optimal rotation also depends on a proper button/bar setup.

Using Bartender (or another addon) and setting up your Keyboard shortcuts so that you can use the ability you want to without having to hunt for it is very important... If I'm doing a high-mobility fight (eg: Keristrasza in the Nexus), I want to be able to go "fireblast" or "Living Bomb", not "uh... press 3". Hunting for the right button or having to do some strange finger contortions to find the attack you want can cost you a LOT of DPS.

Nostrum · 11/15/08, 7:13 PM

Apart from it been new I think DK's are the only class that needs real thought once your main priority list is nailed down. Anytime you use a non standard threat/dps ability the rune setup for your next few abilities can be changed severely depending on what you just did. I can't see it taking long to figure out the common detours but for now it's a lot of fun working it all out.

Pheus · 11/17/08, 7:51 PM

Originally Posted by PSGarak

As far as I can tell, in TBC we had the third-highest ramp-up time after mages stacking scorch/WC and shadow priests stacking shadow weaving (which we also had to wait for), but our debuffs took less time to drop off. In WLK we have the overall highest ramp-up time.

Certain specs of rogues and feral druids who need 5 point rip/rupture as well as ferocious roar/slice and dice and mangle have a pretty high ramp up time before their full dps cycle gets going.

PSGarak · 11/18/08, 9:06 PM

Originally Posted by Pheus

Certain specs of rogues and feral druids who need 5 point rip/rupture as well as ferocious roar/slice and dice and mangle have a pretty high ramp up time before their full dps cycle gets going.

I had forgotten about rogues and combo points, although I think it's not quite the same. You can do a low-point S&D to last you long enough to pull a 5-point S&D. This smaller S&D is basically the ramp-up, and then you start on the normal rotation of 5s/5r or whatever. The problem is more accurately a very long cycle with some backloaded damage (Rupture) than actually a high ramp-up time. Nonetheless, in practice this ends up having the same effect: your damage is much more affected by movement, since you may never advance to the high-damage move in your cycle. How it's changed is an issue of granularity, rather than ramp-up, which means that it's less affected by periodic movement (if you switch to a cycle that fits in the period).

songster · 11/19/08, 4:43 AM

Originally Posted by PSGarak

The problem is more accurately a very long cycle with some backloaded damage (Rupture) than actually a high ramp-up time.

I think this is a meaningless distinction. Moreover, there are significant aspects to rogue DPS that are genuine "ramp-up" issues even by your restricted definition. Stacking of Hunger for Blood, initial stacking of Deadly Poison, and the ratcheting from a single point S'n'D to a full length one are all ramp-up issues.

PSGarak · 11/19/08, 4:21 PM

The distinction isn't extremely production but it's not entirely meaningless. When your fight has period interruptions and the window of contact time is the length of a cycle, the rogue cycle doesn't have much trouble (assuming they run in with 5 CPs for an immediate S&D) while the warlock's ramp-up is more dependent on time off-target than time on-target. It's also easier for a rogue to deal with different (smaller) window sizes, as they choose the S&D that fits. On fights with random interrupts, the distinction is not productive.
To summarize: On fights with periodic interruptions of DPS, needing to ramp-up is a function of time off-target, while long cycles are a function of time on-target.