DoTs EXPLAINED

[epicgrim]

DOTS
It has become clear to me that many people don’t fully understand how DoTs work. I’d love to try to clear up any confusion on the matter. DoT’s since Cataclysm act a bit differently, below are the gory details. *note the details below are based on my knowledge of how moonkins work, if that's not how [insert some terrible class here] works then whatever.

TERMINOLOGY
DoT :: Damage over Time
Δt :: Time interval between each tick.
DPET :: Damage per Execution Time
DPS :: Damage Per Second

Haste
DoTs are affected by haste. As you gain more haste your DoTs Δt decreases.
Haste increases DoTs DPS

Example
The top image is at 6845 haste, the bottom one is at 8190.


Breakpoints
By now most of you have heard of breakpoints. These are the values of haste at which there can be another tick of the DoT within Δt of the base duration of the spell.

Another Tick! Yay more DPS. Right?
Yes, but why is not at first clear. Even though the DoT is now ticking once more per application, it’s still ticking at the same rate. That means that the DoT’s DPS is actually unaffected by haste breakpoints. What is effected though is even more important, because the DPS a single spell does isn’t really important in the large picture. What matters is the DPET. This accounts for the fact that you can be casting other spells while your dots are ticking.
What a haste breakpoint does is increases the DPET of the DoT.

Example

Observe the first column, on top is before a breakpoint at 8190 haste, then on bottom at 8210 haste. The small difference in DPS is because of the 20 haste I added, not because I went over a breakpoint.

Now the interesting part

Ignore the moonfire bar it’s not relevant or accurate. As you can see the DPET of the graph on the right is a good deal higher. 1680 higher to be exact. The diffeance in haste is only 20.

Below is the DPET of Insect Swarm at the bottom of a breakpoint compared to at the top.

As you can see the DPET is very close considering the haste difference here is 1345. However you can see the DPET increasing with haste. That is because of haste’s effect on the GCD, all instant cast spells are effected by this.

RECAP:
Haste -> DoT DPS
Breakpoint -> DPET increase

If you are clipping your DoTs you can ignore breakpoints because they do not effect you at all.

- Hope this helps some of you

PS. Sorry about any misspellings, I can't spell. :/

[Tierbook]

not all dots are affected by haste if i remember correctly the hunter serphent sting isnt

[ImpTaimer]

Warrior Rend / deep wound doesn't scale with haste, and this is on purpose by blizzard because they don't want warriors to make use of haste and murder the charts more than they already try to.

Color me a bigot, but the facts don't lie. I really hope they demote Ghostcrawler one of these days and give someone else a crack at class balance.

It would help to change the thread title since this is in the druid forums. People on the home-page can't see the difference.

[epicgrim]

ah, I'll edit the post. I never pretend to know anything about other classes :P
Moonkin or GTFO

[Angelic]

I love how random people talk about balancing classes and are willing to fire game developers with years of experience who surely know nothing about game designing, right? What you got to show?

Twelve years experience of holding the mouse and whining?

[shadowboy]

Only caster DoTs are affected by haste. Physical dps class DoTs are not (this includes any/all bleeds, serpent sting, piercing shots, deadly poison, blood plague, frost fever, unholy blight, etc. I can't comment for sure on censure, but I believe it is also not affected by haste.

But the single biggest most gigantic flaw in your post is using simulationcraft as "proof."

Regardless of the truth of your post, if you want to provide evidence of the basis for game mechanics, you need to use game mechanics--not a simulation of game mechanics.

That's like me using electronic structure calculation software to prove that a hydrogen atom's 1s orbital shape is a sphere--it doesn't prove anything, correct or not.

I'll leave it up to you to figure out why it's incorrect methodology.

[angelx7x]

you can't clip dots so long as you reapply before the last tick, but after the tick that precedes it. so you still can't -ignore- breakpoints unless you're actually clipping (which in MoP they may even be making even harder by allowing 2 ticks to be "clipped" without a dps loss)

should also be pointed out that the dot with more haste has nearly a whole second taken off from the tick intervals.

[epicgrim]

Well I guess I figured that the people who made Simcraft knew how DoT work therefor would be good enough to illustrate my point. If you want I'll make a quick mod that sum this all up in game. I can't reach those high haste numbers which is why I wanted to Simulate it, to prove beyond a doubt what is going on. Unless you are arguing with the theory though your point is true yet also somewhat irrelevant.

EDIT:
After I though about this for a minute I realized how wrong you really are. Simcraft is a tool for simulating situations over very large data sets. like I said before, unless they did something wrong in their code, witch they didn't because the math involved in the test that I did for this post could be done by a 5th grader, then using a tool like Simcraft makes it more clear. It further proves that there isn't RNG. any variation can be explained by mechanics and mechanics alone. So unless your saying that Simcraft can't be trusted to do math like (dot tick DMG) * (Ticks) / GCD then I don't see your point.

[MasterNewbie]

Only caster DoTs are affected by haste. Physical dps class DoTs are not (this includes any/all bleeds, serpent sting, piercing shots, deadly poison, blood plague, frost fever, unholy blight, etc. I can't comment for sure on censure, but I believe it is also not affected by haste.

But the single biggest most gigantic flaw in your post is using simulationcraft as "proof."

Regardless of the truth of your post, if you want to provide evidence of the basis for game mechanics, you need to use game mechanics--not a simulation of game mechanics.

That's like me using electronic structure calculation software to prove that a hydrogen atom's 1s orbital shape is a sphere--it doesn't prove anything, correct or not.

I'll leave it up to you to figure out why it's incorrect methodology.

Computation that is based on a generally accurate model of mechanics (WoW combat mechanics, quantum mechanics as per your example, Newtonian mechanics, etc.) is typically sufficient to show something. I don't really see how it is a flaw when it can do a thousand times over what you can do in the same time period. I don't know if the moonkin model in SimC is accurate at all, that is another story. To me, this would be acceptable, assuming the moonkin script is correct. Unless you have a counter or show that the moonkin script is incorrect, you don't really have much room to talk about his being incorrect for using a tool to demonstrate his point.


OP, aside from a needed clean up (I originally believed you were incorrect), looks fine to me. Though it wouldn't hurt to verify in-game before posting results if possible.

[Deleted]

Dunno if I'm missing something here, pretty tired but

on the second picture you say that the difference is 20 haste and you don't go past a breakpoint, yet it says that with

8190 haste you get 399 ticks

and with

8210 haste you get 400 ticks

that is going past a haste breakpoint, no?

edit:
yea obv. what u say is correct, just wasn't thinking straight

[angelx7x]

Dunno if I'm missing something here, pretty tired but

on the second picture you say that the difference is 20 haste and you don't go past a breakpoint, yet it says that with

8190 haste you get 399 ticks

and with

8210 haste you get 400 ticks

that is going past a haste breakpoint, no?

ha, just noticed that for IS. that would indeed be going past a breakpoint.

[Braindwen]

Dunno if I'm missing something here, pretty tired but

on the second picture you say that the difference is 20 haste and you don't go past a breakpoint, yet it says that with

8190 haste you get 399 ticks

and with

8210 haste you get 400 ticks

that is going past a haste breakpoint, no?

It IS going past a haste breakpoint, but you are missing something. The reason it got that 400th tick in is because of the 20 haste: it ticks just a little faster and lets it get 1 extra tick in during whatever amount of time the sim ran for (10 minutes I think). Notice that the first picture, despite crossing no breakpoints, gains 24 ticks from haste (375->399).

What it gains from the breakpoint is shown in the "count" column. In the first (8190 haste) row, it takes 30.5 casts to get 399 ticks, in the second row it takes 29.4 casts to get 400 ticks. You spend fewer casts on DoTs when you cross a breakpoint for the same number of ticks, which frees you up to cast other spells. Your DPS from the DoTs themselves doesn't go up significantly if you're keeping them up all the time, but your DPS from other spells will go up because you have more time to cast them.

[Windry]

I might not be an expert, but this is my understanding. Correct me if I'm wrong of course. With more haste, dots do 2 things.

1) They tick faster -----> delta t DECREASES, not increases.
- You can find this proof by looking at any World of Logs and looking at the time between each tick of corruption of an affliction lock that gears a lot of haste vs a demo lock that barely gears any haste (make sure the demo lock is not under demon soul:felguard).

2) The dots last longer. I don't see you mentioning this in your post. With more haste, the dot last longer and you have to refresh it less.
- You can find this proof by turning on enhanced tool tips and looking at the tool tip of one of your dot spells when you have a lot of haste and when you have very little haste.

- Increasing your haste by 1 point past a threshold does not increase the DPS of that dot more so than increasing your haste by 1 point anywhere else.
- However, increasing your haste by 1 point past a threshold DOES increase your OVERALL DPS more so than increasing your haste by 1 point anywhere else.
- Bottom line then... haste thresholds are valuable targets for gearing, and the value of haste in your gear is very high at a threshold, but significantly lower once you past that threshold.

[Braindwen]

Pretty much spot-on, except:

2) The dots last longer. I don't see you mentioning this in your post. With more haste, the dot last longer and you have to refresh it less.
- You can find this proof by turning on enhanced tool tips and looking at the tool tip of one of your dot spells when you have a lot of haste and when you have very little haste.

Is only true at breakpoints. At all other times, Haste decreases the length of your DoTs to keep the # of ticks constant. I think you already knew that, just clarifying .

DoT (and HoT) lengths are roller-coasters in regards to haste, if they scale with it. At some value of haste they may last a second or two less than their original duration, then you cross a breakpoint and they shoot up above it. Adding more haste will lower the duration again until the next breakpoint, where they jump back up again, and so on.

[mmff4]

ha, just noticed that for IS. that would indeed be going past a breakpoint.

Of course it has more ticks because haste causes dots to tick faster (Faster ticks = more ticks!). There are breakpoints, yes. But breakpoints have close to no effect on the DoTs DPS assuming you keep them up with 100% uptime because when you reach a certain breakpoint (extra tick) the DoT duration resets back to the original duration.


Note: Durations aren't based on any real spell.

Q: If breakpoints have to effect on DoTs DPS why do healers want to have an extra hot tick?
A: More overall healing but less HPS as you don't keep your HoTs up with 100% uptime under normal circumstances.

[Psilar]

I might not be an expert, but this is my understanding. Correct me if I'm wrong of course. With more haste, dots do 2 things.

1) They tick faster -----> delta t DECREASES, not increases.
- You can find this proof by looking at any World of Logs and looking at the time between each tick of corruption of an affliction lock that gears a lot of haste vs a demo lock that barely gears any haste (make sure the demo lock is not under demon soul:felguard).

2) The dots last longer. I don't see you mentioning this in your post. With more haste, the dot last longer and you have to refresh it less.
- You can find this proof by turning on enhanced tool tips and looking at the tool tip of one of your dot spells when you have a lot of haste and when you have very little haste.

- Increasing your haste by 1 point past a threshold does not increase the DPS of that dot more so than increasing your haste by 1 point anywhere else.
- However, increasing your haste by 1 point past a threshold DOES increase your OVERALL DPS more so than increasing your haste by 1 point anywhere else.
- Bottom line then... haste thresholds are valuable targets for gearing, and the value of haste in your gear is very high at a threshold, but significantly lower once you past that threshold.

The DoTs don't necessarily last longer. Sometimes as Haste increases and you hit a breakpoint the overall duration of the spell itself will change. Usually it stays relatively the same. Sometimes at one specific breakpoint, usually the first one, the duration will actually increase resulting in a longer duration per cast and one extra tick - but also at a smaller delta T(ticks occur faster). Most future breakpoints, if not all of them, will decrease the duration.

Break points are important because it adds another tick. When you post these gigantic sims with 1/2 casts and shit you begin to get lost in the numbers and forget about reality. Half casts do not fucking matter because you cannot cast 1/2 of a spell. A lot of the time the haste will do nothing for you up to a breakpoint because of this. It would require you to change your rotation, at a DPS LOSS, simply to keep the DoT up or from clipping.

Furthermore and here's the important part... when your DoT HoT or whatever expires...that is the crucial point in deciding what to cast next. How long has the DoT been off the target? Did it expire in the middle of a cast? Do I delay a cast a split second so as to not clip a DoT? Is that even worth it? Do I alter my rotation?

Also, I believe censure is effected by haste.

[Windry]

@Braindwen and mmff4 - Thanks, I actually didn't know that it fluctuated, learned something new.

@Psilar - My understanding is that (with 4 exceptions) you should not let your DoTs expire. Clipping the last tick still gives you that last tick.
Example:
- My corruption has 7 ticks and ticks roughly once every 2.5 seconds.
- If I refresh my corruption at 2.5 seconds left (after the 6th tick, and before the 7th tick), I will still get that 7th tick of the original corruption followed by 7 more ticks of the new corruption.
- For this reason, you should never run into the situation of, "when your DoT expires, do I do x or y?" (Unless you were doing a 3 second hard casts, only destro locks do this with soul fire, I think)
- If you were doing a 3 second hard cast, you should have noticed in those 3 seconds that your DoT was about to fall off and get ready to queue up that DoT.

4 exceptions:
1) Living bomb
2) Bane of Agony
3) Bane of Doom when current BoD was under very high procs and cds
4) Right before an eclipse state (not positive on this one)

[Deleted]

sorry stopped reading at
DoTs are affected by haste. As you gain more haste your DoTs Δt also increases.

that's 'decreases' right?

@mmff4
But breakpoints have close to no effect on the DoTs DPS assuming you keep them up with 100% uptime because when you reach a certain breakpoint (extra tick) the DoT duration resets back to the original duration.

The reason the breakpoints give more dps is because you have to use less GCDs casting the DoT over the period of a fight allowing you to cast something else in it's place. The DPS gain of reaching a breakpoint can be calculated by seeing how many GCDs you gain over a time period and seeing the damage you can cause during those gained GCDs from other spells.

[darkwarrior42]

Computation that is based on a generally accurate model of mechanics (WoW combat mechanics, quantum mechanics as per your example, Newtonian mechanics, etc.) is typically sufficient to show something. I don't really see how it is a flaw when it can do a thousand times over what you can do in the same time period. I don't know if the moonkin model in SimC is accurate at all, that is another story. To me, this would be acceptable, assuming the moonkin script is correct. Unless you have a counter or show that the moonkin script is incorrect, you don't really have much room to talk about his being incorrect for using a tool to demonstrate his point.

Except computation doesn't equate to proof in the way it's being done here, or in your example.

Here's what the sequence of events in the OP, and why it doesn't constitute proof:

#1) The people who work on simc study the mechanics of the game.
#2) The people who work on simc break down how haste works, and how it affects dots (in this case, insect swarm/moonfire/sunfire).
#3) They test their theory, and verify that their model is correct.
#4) Having proven their theory correct, they create a program that simulates the spells being used, and the effect of various stats on those spells.
#5) Epicgrim uses their tool to generate numbers
#6) Epicgrim cites the numbers that their tool genreates as proof that haste functions as described in this thread.

The flaw, if it isn't obvious, is that the tool doesn't provide proof of anything. It's the other way around; if the methodology weren't proven to be accurate using methods other than simc, simc would not have been created in the first place. It isn't the proof, it's the result that can only happen after the theories have been proven to be sound.

In general, simply iterating on the same problem over and over again does not constitute proof. Proof is when you demonstrate using evidence and calculations that the model described is correct, not when you use someone else's model that is itself a derivative or simulation, as any error in the simulation means that your "proof" is rendered useless. (And to head off the obvious counter-question, what I would consider proof here is a simple combat parser, combined with in-game testing at various levels of haste, demonstrating the various points at which you gain additional ticks; a combat log that shows timestamps on the various ticks would suffice for proof that haste increases dps by decreasing the interval between ticks. As for the proof that hitting a new breakpoint increases DPET, I don't think that needs proven at all; it's pretty much by definition. When you get 10 ticks instead of 9, you get more damage from one cast of the spell. That's literally a definition of higher DPET.)


Do not misunderstand me, I appreciate the effort that Epicgrim put into this post, and it's a topic worth expounding upon. And I don't see any claim he's made that is wrong, but using simc numbers doesn't prove anything (particularly when different simulations can provide different results, as a result of taking different factors into account in different ways). I don't even think using the simc numbers is wrong, if he were simply providing them as an easy example... the only issue I see at all is that he's using simc numbers to prove the point, which they don't.

[angelx7x]

@Braindwen and mmff4 - Thanks, I actually didn't know that it fluctuated, learned something new.

@Psilar - My understanding is that (with 4 exceptions) you should not let your DoTs expire. Clipping the last tick still gives you that last tick.
Example:
- My corruption has 7 ticks and ticks roughly once every 2.5 seconds.
- If I refresh my corruption at 2.5 seconds left (after the 6th tick, and before the 7th tick), I will still get that 7th tick of the original corruption followed by 7 more ticks of the new corruption.
- For this reason, you should never run into the situation of, "when your DoT expires, do I do x or y?" (Unless you were doing a 3 second hard casts, only destro locks do this with soul fire, I think)
- If you were doing a 3 second hard cast, you should have noticed in those 3 seconds that your DoT was about to fall off and get ready to queue up that DoT.

4 exceptions:
1) Living bomb
2) Bane of Agony
3) Bane of Doom when current BoD was under very high procs and cds
4) Right before an eclipse state (not positive on this one)

as it stands right now (4.2) living bomb is not an exception. you can clip it as you would with other dots, as the explosion damage right now is negligible (by clip i mean refresh before that last tick goes) BoD is also similar, BoD ticks every 15 seconds, you can refresh it when its duration hits 14 seconds if you so choose