Dots and you: The Affliction Warlock Thread - Page 15

Kalle · 01/13/09, 8:23 AM

I noticed another error (or change by Blizz?) in "The math part": In the computation of drain soul's damage you do (1+0.15(Shadow Mastery)+0.60(Soul Siphon)). My testing on 3.0.8 PTR shows that SM and soul siphon are multiplicative.

phulshof · 01/13/09, 9:11 AM

Originally Posted by Leibniz

Cast the SB. If your UA has a cast time of 1.3 and your SB has a cast time of 2.3 your talent spec is wrong. If you have the Bane talent there's no possible way for your SB and UA to have the same cast time reduction. If you don't have Bane you're in a bad affliction spec. Regardless, still cast the SB.

Well, the numbers were just an example.

The question was just about wether there may be a point where the cost of waiting outweighs that of letting a DoT run out. Your answer seems clear: there is never a situation in which you'd prefer to do nothing, even for 0.2s, over casting the filler spell.

Trickykid · 01/13/09, 2:27 PM

Originally Posted by phulshof

I often come to a point where I have to decide between waiting and casting another SB. The question is: what is the cutoff point between those two? For example: I have 1.5s to go on my UA, with a casting time of 1.3s. Do I wait 0.2s or do I fire off another 2.3s taking SB, causing UA to not run for 2.1s? Any thoughts?

If you'll be able to use a life tap down the road, they make good fillers in situations like these.

Villeraz · 01/13/09, 9:28 PM

I posted about this in the Simple Questions sticky in response to a question, but I thought I'd share it in here as well, since it is most pertinent to affliction warlocks.

Haunt's healing threat is attributed to the player or mob that it was debuffing.

This was tested with two locks, engaged in a duel, without Fel Armor. Lock B body-pulls (0 threat, but present on threat table) a mob. Lock A pulls aggro on it by casting a curse (CoE works well for this). Lock A casts haunt on Lock B, and should also not be at full health. Once the haunt heal activates, the mob will return to Lock B due to the heal aggro.

Leibniz · 01/14/09, 2:16 AM

Originally Posted by phulshof

Well, the numbers were just an example.

The question was just about wether there may be a point where the cost of waiting outweighs that of letting a DoT run out. Your answer seems clear: there is never a situation in which you'd prefer to do nothing, even for 0.2s, over casting the filler spell.

I would not say "no point" but the reaction time required is probably superhuman. And the other person was correct, if you could use the lifetap use it vs. the filler. If you're already casting the filler, interrupting is a wasted gcd.

phulshof · 01/14/09, 4:38 AM

Originally Posted by Leibniz

I would not say "no point" but the reaction time required is probably superhuman. And the other person was correct, if you could use the lifetap use it vs. the filler. If you're already casting the filler, interrupting is a wasted gcd.

Not sure if I'd want to call it superhuman; usually you're in the middle of casting a SB when you notice that it will be complete before it's time to refresh your DoT. At that time you have to decide if you'll wait a brief moment after this SB is complete, cast life-tap (if beneficial) or cast another SB. The warning about never interrupting a cast spell was clear.

Draezaal · 01/14/09, 5:06 AM

A few things that I have said we want to look at are:

2) Make Affliction feel less dependent on dot timers and other external mods (nothing wrong with the mods -- we just think it's getting harder and harder to play the spec with the default UI).

Do most of you who play Affliction use DoTimer or some other mod? I've always found that the default UI allows me to monitor my dots just fine while avoiding screen clutter and maintaining more awareness of my surroundings.

EDIT: This is from a blue post regarding the upcoming changes in 3.1

Leibniz · 01/14/09, 5:32 AM

Originally Posted by Draezaal

Do most of you who play Affliction use DoTimer or some other mod? I've always found that the default UI allows me to monitor my dots just fine while avoiding screen clutter and maintaining more awareness of my surroundings.

EDIT: This is from a blue post regarding the upcoming changes in 3.1

I've always used DoTimer myself. I keep my timers in the upper left so I can easily keep an eye on myself, the mob, and not be too distracted from the main actions.

krilz · 01/14/09, 5:45 AM

ForteXorcist is what I use. It has a very nice looking timer, doesn't use up too much memory, supports a lot of classes (the only one missing is Rogue I think) and has a lot of extra functions (such as monitoring other warlocks shards and SS and the healthstones in the raid) which you can shut off if you don't use them.

Zaleiria · 01/14/09, 6:06 AM

Originally Posted by Villeraz

I posted about this in the Simple Questions sticky in response to a question, but I thought I'd share it in here as well, since it is most pertinent to affliction warlocks.

Haunt's healing threat is attributed to the player or mob that it was debuffing.

This was tested with two locks, engaged in a duel, without Fel Armor. Lock B body-pulls (0 threat, but present on threat table) a mob. Lock A pulls aggro on it by casting a curse (CoE works well for this). Lock A casts haunt on Lock B, and should also not be at full health. Once the haunt heal activates, the mob will return to Lock B due to the heal aggro.

Hm, that's interesting. Also, I know the combat log reads "Target's Haunt heals Warlock for X."

Kalle · 01/14/09, 8:25 AM

I tkink your calculation of the amount of haste eradication provides is off. I'm not absolutely sure on what your error is, but I believe it is the fact that every summand in your "average haste benefit ="-formula represents a different amount of time. Apart from the fact, that I derived a formula which produces different results, a simple simulation I wrote to check the results makes me quite sure that there is indeed something wrong with your calculation.

I will now show the derivation of my formula. I will need the limit of an infinite geometric series, which is
$\sum_{k=0}^\infty r^k=\frac{1}{1-r}\quad\mathrm{and}\quad\sum_{k=0}^\infty k\cdot r^k=\frac{r}{(1-r)^2}$
|r|<1 is required for the convergence of the series.

The average proc time is, as you wrote in the section about trinkets,
$\sum_{n=1}^\infty t_n \cdot p\,(1-p)^{n-1}$
where tn is the time the n-th spell (that can cause the effect in question) is casted or hits, depending on what causes the proc (with proc chance p). You could insert a real casting sequence here and use the first few terms of the sum as an approximation. But if we assume a constant amount of time between subsequent casts, we can actually compute the sum. That means, tn = t0 + (n-1)*dt with some "casting time" dt and the first spell at a time t0. Now we have:
$\sum_{n=1}^\infty (t_0+(n-1)\cdot dt)\,p(1-p)^{n-1}=\sum_{n=1}^\infty t_0\cdot p(1-p)^{n-1}+\sum_{n=1}^\infty (n-1)\,dt\cdot p(1-p)^{n-1}=\sum_{n=0}^\infty t_0\cdot p(1-p)^n+\sum_{n=0}^\infty n\,dt\cdot p(1-p)^n$
and with the formulas for the geometric series:
$\frac{t_0\,p}{1-(1-p)}+dt\cdot p\,\frac{1-p}{(1-(1-p))^2}=t_0+dt\,\left(\frac{1}{p}-1\right)$

With uptime=12s/(average_proc_time+30s) and average_benefit=uptime*20% this leads to the following average increases in casting speed:
1/3 2.35%
2/3 3.44%
3/3 4.21%

I used t0=0 here because if the 30s cd begins when corruption ticks, corruption (almost) immediately ticks again after it expired.

In the new section about trinkets, you wrote:

The "average proc time when no ICD is up" is

(time of first proc chance) * (proc rate) +
(time of second proc chance) * (proc rate) * (1-proc rate) +
(time of third proc chance) * (proc rate) * (1-proc rate) * (1-proc rate) +
(time of fourth proc chance) * (proc rate) * (1-proc rate) * (1-proc rate) * (1 - proc rate) and so on and so on.

The "time of n-th proc rate" is just (n-0.5)*average casting time. You can be lucky and hit a spell the moment the ICD is gone and get a proc, or on the next spell, or the one after that, and so on. The "0.5" helps to get an "average" moment.

For example, a 15% proc chance gives an average time of 10.88s between procs (if there is no internal cooldown present).

I would have to set t0=0.5*average_casting_time. From what you wrote I would expect that we get identical results here. But we don't. Did you use the average casting time also for the difference betwenn n-th and (n+1)-th proc chance? If you did, I think I need more details about your calculation to see the difference.

And a word of warning: the way you compute the uptime for trinkets is not applicable if the debuff lasts longer than the internal cd (eg if there is no internal cd). My formula neither.

Auze · 01/14/09, 11:32 AM

Originally Posted by Draezaal

Do most of you who play Affliction use DoTimer or some other mod? I've always found that the default UI allows me to monitor my dots just fine while avoiding screen clutter and maintaining more awareness of my surroundings.

EDIT: This is from a blue post regarding the upcoming changes in 3.1

I tried without DoTimer, and it was just impossible. I keep them down near the action bar so I can monitor the GCD on the spells. I put drain soul execute watcher down there so I can monitor the mobs life. With DoTs lasting so short, you pretty much need some sort of add-on. In games like EQ where DoTs would last literally forever, you had lots of time to wait to react. But WoW is such a faast paced game.

I think the only way they could make DoTs not require an add-on is to allow cast stacking. Ie. You click Haunt 5 times and it auto recasts it when it runs out up to 5 times. If you're in the middle of casting SB, it'll wait. If you end up with more than one waiting, it'll take the first one that is waiting.

necrodyssee · 01/14/09, 1:13 PM

Thank you for this great post,

When I went to affliction few days ago, I first read this forum to get helpfull informations and this post with Fallenman's are great, thank you for your great job.

However I have some questions :
1) In my first reading (first post of this topic), I saw immolate should be maintain ALWAYS, both below 35% and 25% of boss life. Then I saw yesterday it was edited and that it's better to drop it below 25%, but what about betwenn 35% downto 25% boss life?
2) I tried to answer to my first question and I asked myself (and so yourselves !) whether it wouldn't be preferable to compute DPCT for 35%-100% for all spells, and the DPCT for all spells between 25% and 35% (immolate included) then the DPCT below 25% to discuss if it's suitable to drop or not immolate instead of computing a "mean" value of death's embrace effect during the all fight by (1+0.12*0.35) coeff as it is done at present?
3) I also took a look to corruption (and other spells) DPCT computations and numerical application doesn't seem to be the good one.

(1+0.15(Shadow Mastery+0.10(Improved Corruption)+0.05(Contagion))*(1+0.2*hit chance)(Haunt)*1.1(Shadow Embrace)*1.13(CoE)*1.01(Spellstone)*1.03(Malediction)*1.03(Sanctified Retribution)*(1+0.12*0.35) (Death's Embrace)*(1740(Base Damage) + ((1.2(Base Coeff)+0.36(Emp Corr)+0.30(Everl. Affl.))*(2000(spell power))))*0.99(Hit Chance)*(1+0.15(Crit Chance)) =
(1+0.15+0.1+0.05)*(1+0.2*0.99)*1.1(Shadow embrace missing?)*1.13*1.01*1.03*1.03*(1+0.12*0.35)*( 1740 + ((1.2+0.36+0.30)*(2000)))*0.99*(1+0.15) = 9835.90

Without this missing (red part), I computed : damage done = 12214.28 instead of 9835.90
With the missing SE, it is : damage done = 13435.71

Could you tell me what's the correct value? and why?

4) I saw 0,15 crit for both Unstable Affliction and Corruption, is this to take into account Pandemic effect?

5) Last question, why don't you assume that the hit cap his 100% (even if we don't have a Draenei in the group, the 1% hit can be obtained by food for example)?

PS : sorry for my bad english, it's not my mothertongue (and english teaching in France has not a good reputation

), I hope my post will nevertheless be understandable.

Namnalia · 01/14/09, 1:51 PM

Hi Necrodyssee,

thx for the feedback, and now my answers:

1&2): Interesting question (I actually didn't really think of it), but I guess the answer is easy: As SB does roughly 2k dps and Immolate does 4k DPCT, it MUST be a lot better to cast it between 25% and 35% than to fill in half a shadow bolt instead, no matter who gets some +%. The same is true for all the dots, of course - so I can get around the computation for the 35-25% zone :-)

3) Yep, looks strange to me too, I'll have a deeper look into it. Thx for the hint.

4) Right. It works exactly like a 100%-Crit, it just doesn't look like one - at least that's what I assumed.

5) I DID assume the cap being 100%. The test person just does not reach it because his equip is too low. When I add the computation for a decent Ulduar starting equip, the cap will be reached for sure. (Btw, you are about the 100th person telling me that the cap is not 99%, so you're in good company^^)

Greets,
Nam

twotwentytwo · 01/14/09, 3:41 PM

Just wanted to let you know that the numbers for black magic are very wrong, unless im missing 800 damage per tic for something im doing wrong?

just trying to help ppl save money on mats

1400g X.X lol

fallenman · 01/14/09, 4:45 PM

Enchants:

Weapon:

Black Magic 115 dps
Applies a dot ticking for 1045 to 1155 every 3s for 15s. ICD 45s, proc rate 50% -> up every 45s+2.64s
Scales with some shadow talents, no tests yet, dps without these would be:
(1045+1155)/2 * 5 * /(45s+2.64s) = 115.44 dps w/o scaling!
Cannot proc molten core but some other effects like 2p T7.

Hey Namnalia, black magic does 209-231 damage per tick. I have no idea where you got 1045-1155 damage per tick from?

So, the correct math would be:

(209+231)/2 * 5 * /(45s+2.64s) = 23.08dps.

Namnalia · 01/14/09, 8:02 PM

Yep, that was an error, thx - probably added the ticks or something. Corrected it. I hope I did not ruin too many people :-)

Meepity · 01/15/09, 1:13 AM

First off, thanks for the info shared so far. I do have a question though. Would anyone here think that 6y more range on spells is worth a trade off to, say, 1% spell damage? From personal experience, it can be quite frustrating to watch your dots fall off just because one cannot get into range without taking serious damage, eg. Heigan during dance. I'm sure the topic of range has talked about so far, but I'm not exactly sure how much it's worth.

Leibniz · 01/15/09, 2:14 AM

Originally Posted by Meepity

First off, thanks for the info shared so far. I do have a question though. Would anyone here think that 6y more range on spells is worth a trade off to, say, 1% spell damage? From personal experience, it can be quite frustrating to watch your dots fall off just because one cannot get into range without taking serious damage, eg. Heigan during dance. I'm sure the topic of range has talked about so far, but I'm not exactly sure how much it's worth.

If your tank is doing their job correctly, you should never be out of range on Heigan. There are very few fights the +range is really that useful since only 3 dots are instant cast. It will help some on movement/positioning fights, but there are only a handful I can think of. That said if this is a consistent issue with you and your tank on a lot of fights a 1% dps loss is better than no dps because you're out of range.

Meepity · 01/15/09, 2:48 AM

It was actually during the dancing part, where I have to get hit by waves to be in range to reapply my dots on Heigan before moving again, but then again, it's one of the few cases where I ever get out of range.

Leibniz · 01/15/09, 2:57 AM

You're running too far away from the platform then.

blecap · 01/15/09, 3:04 AM

Hello. I've a serious problem over here, while some of the warlocks in here are able to hit 5k+ on Patchwerk, I'm struggling. My stats are 2200 SD w/ Fel Armor, 311 Hit, 15% crit and 590 Haste. Raid composition consists of oomkin, elemental shaman, UH DK and i'm only doin 3k-4k dps. My rotation is SB, Haunt, UA-Immo, corr, CoA, SL and spam SB. I do not clip my dots as i refresh at the correct time once the last tick had tick off. When boss is at <25% i refresh whatever dots needed except for immolate and drainsoul instead of SB. My glyphs are SL, Immo and Corr. Please help.

Nnayr · 01/15/09, 3:52 AM

@blecap

you talking 25 or 10man?

need a WWS to help you properly.

chang · 01/15/09, 6:32 AM

1) First of all: Do not use Black Magic. In the calculation the damage of the whole curse was used for the damage of one tick. I spend approximately 2000g to find out that it is wrong

But I am not angry, since this thread helped me a lot improving my playing style (which is more worth than any g).

2) I have seen in many forums / posts wrong calculations of the expected Eradication uptime. Some post before mine, there is a correct calculation. So I think it is worth investing more than 1 point, but perhaps only 2 and not the full amount of 3...

The mistake many people did is the following. They calculated the percentages and probabilities of uptime for one phase and then averaged over the percentages. Let us see, why this is not possible. Consider the case, where you have a percentage of uptime of 33.3% = 1/3 and 25% = 1/4 with equal probabilities of 1/2. Averaging over the percentages gives us (1/3 + 1/4) / 2 = 7/24 = 29.167%. On the other hand these 33.3% could result of a phase lasting 36 seconds with 12 seconds uptime and the 25% are caused by a pahse lasting 48 seconds with 12 seconds uptime (as it could be the case for eradication). Then we have an uptime of 24 seconds in 84 seconds, i.e. 24 / 84 = 28.571%. This counter example shows, that in general you cannot average over the percentages.

So the correct way of calculating the expected uptime is to calculate the epxected length of a "phase" (i.e. the time waiting for the proc plus the cooldown). Then divide the duration of the spell through the expected uptime. There are still some details to consider, e.g. if Eradication can proc with a tick of Corruption immediately hitting the end of the cooldown or the average time for the first Corruption tick after the cooldown, but I think they do not lead to big differences in the results.

The next thing is, that a lot of people use somehow complicated calculations, but in this particular case it is very easy. We are interested in the expected number of ticks of Corruption, until Eradications procs. And this is in fact a geometric probability distribution. The expected number of ticks is thus 1 / p, leading to an expected phase length of 1 / p * 3 + 30 = 3 / p + 30. So we have 12 / (3 / p + 30) for the expected uptime. The values are 11.429%, 16,471%, 20.000% for 1, 2, 3 points invested in Eradication. If we assume that the average time until the first Corruption tick after the cooldown is 1.5 seconds or 0 seconds (insted of 3 used in the calculation above) results change slightly, but not as much as for the wrong calculations. (You would get the same result with the formular some posts before mine. In fact there it is calculated "by hand", what I calculate with knowledge about the geometric distribution).

Since I am a new member I cannot open a new thread, but I think it would be worth to open a new thread with the correct calculations, such that all warlocks (not only affliction) will recognize it...

3) Now my last idea: I tried a lot of different affliction specs with the aim to optimize for damage output. Yesterday I looked at the talents again and my attention came to Shadowburn. I thought about investing 1 point in Shadowburn for some reasons.

At first, Shadowburn has half the cast time of Shadowbolt, but more than half of its damage. The disadvantage is, that it consumes shards every time and has a cooldown of 15 seconds. I do not the exact amounts of damage produced by those two spells, but let us assume Shadowburn does 70% of Shadowbolts damage. Having Shadowburn on cooldown you lose on average 1/2 Shadowbolt every 15 seconds, but getting 70% (more than 1/2) of Shadowbolts damage instead. Thus, in 15 seconds you get additional 20% of one Shadowbolts damage. Let that be 800 in average. Then you increase your dps by 800 / 15 = 53.3 which is not bad for one talent point and I think for well eq. locks this can be even more. Again, the disadvantage is that Shadowburn requires a shard each time it is casted. For a five minute fight you would need 20 shards, which is a lot

But there is another reason, why Shadowburn could be used. Consider (without any spell haste), you have 2.9 seconds until you have to renew Immolation and Unstable Affliction some dots (A similar scenario was describes some post above mine). A Shadowbolt would consume 3 seconds, and the dots would be applied 1.6 seconds after they ended. On the other hand using a Shadowburn would result in applying the dots 0.1 seconds after theiy dropped.

Another advantage is that you can use Shadowburn while moving with your char.

In my opinion investing 1 talent point for Shadowburn could really increase your dps. It makes the rotation harder and would require perfect skill. And you need tons of shards. But with 2 or 3 pockets that can contain 32 shards (and since you refill shards during a raid with your <25% filler) this would be possible.

Although I do not know if that would lead to an increase of real world dps, I think I will try that out in some of the next raids

blecap · 01/15/09, 7:38 AM

Originally Posted by Nnayr

@blecap

you talking 25 or 10man?

need a WWS to help you properly.

25 man. I don't currently have a WWS atm, will get them next time.