Tuesday, February 3, 2009

Eclipse: Part 2 - What's it Worth?

This is Part two of my four part series looking at Eclipse. In this post I will look at the value of Eclipse and try and provide you with an idea of how it will help your DPS. Part one looked at the two Eclipse buffs and showed which is the better buff. Part 3 will try and answer the question, Should you reapply DoTs during Eclipse? Finally, Part four will summarize my findings and I will share my general thoughts about the talent.

For those of you that don't like reading math, please feel free to skip to the bottom for my TL:DR version of the post.


Ok, How will Eclipse affect my DPS?

Please Note: Most of the assumptions I made in part one will apply to this post as well. The one exception is to my 5th assumption. For this section I am assuming that you have the [Idol of the Shooting Star] equipped when casting both Starfire and Wrath to maximize the Starfire Eclipse buff.

It is difficult to assign Eclipse a concrete value. Given it's random nature it is impossible to nail down an exact DPS rotation to measure it, and when it does proc the buff might be wasted due to fight mechanics. So, the number I am presenting in this section are kind of a best case scenario.

There are 5 steps to providing a value to eclipse.

1. We need to determine how long our Precasting period will be and how much damage we will do during it. We can determine the average number of Wrath casts it is going to take to proc the Starfire Eclipse using the following formula:
# of Casts to proc = 1 / (proc rate * crit rate) or
# of Casts to proc = 1 / (0.60* 0.35) = 4.7619 Casts

Since it is highly unlikely that you will see the buff proc before your casting your next spell I will assume a 1 spell cast buffer. This will increase the number of Wrath's cast by 1 and reduce the Buff period by Wrath's casting speed.

So our Casting Time and Damage for the Precasting period looks like this:

Avg # of Wrath's = 4.7619 +1 = 5.7619
Avg Wrath Cast time = (((1.5/1.17) * (1 - 0.35)) + 0.35) = 1.1833 seconds
Avg Wrath Damage = ((((588 + (2000 * 0.6714)) * 1.13)*1.0712)*(1-0.35))+((((588 + (2000 * 0.6714)) * 1.13)*1.0712)*2.09 * 0.35) = 3228.77 damage

Precasting period cast time = 5.7619 * 1.1833 = 6.8181 seconds
Precasting period damage = 5.7619 * 3228.77 = 18603.85

2. Now we need to determine how long our Buffed period will be and how much damage we will do during it. The Duration of the eclipse buff is 15 seconds, but I am assuming that we will cast one Wrath before we recognize that we have the buff. So the buff's duration is reduced by Wrath's cast time of 1.1833 seconds. Also remember that Damage is calculated at the end of spells cast. Therefore I will ignore any partial casts and round down to the nearest number of whole Starfire casts.

So the numbers for the Buffed period look like this:

Avg SF Cast time = (3 - (0.5 * 0.68)) / 1.17 = 2.2735 seconds
Avg # of Buffed SF casts = (15 - 1.1833) / 2.2735 = 6.077, rounded down to 6 casts.
Avg SF Damage = ((((1285 + (2000 * 1.2)) * 1.1)*1.0712)*(1-0.68))+((((1285 + (2000 * 1.2)) * 1.1)*1.0712)*2.09 * 0.68) = 7560.48 damage

Buffed period cast time = 6 * 2.2735 = 13.641 seconds
Buffed period damage = 6 * 7560.48= 45362.88

3. The combination of the Precasting period and the Buffed period are what I call the Eclipse Rotation. So lets combine the values from the previous two steps.

Eclipse Rotation cast time = 6.8181 + 13.641 = 20.4591 seconds
Eclipse Rotation damage = 18603.85 + 45362.88 = 63966.73 damage
Eclipse Rotation DPS = 63966.73 / 20.4591 = 3126.57 DPS

4. Now that we have the DPS of the Eclipse rotation we need to compare it to what we would be casting otherwise. Once again I am ignoring DoTs and assuming that they don't need to be refreshed. Therefore, the only other DPS options are Wrath Spam or Starfire Spam.

The average DPS of Wrath Spam can determined using information from step 1 above and the average normal Starfire DPS is also listed above.

Avg Normal Wrath DPS = 18603.85 / 6.8181 = 2728.60 DPS
Avg Normal SF DPS = 2556.78 DPS

Therefore, the Eclipse rotation increases DPS by:

Value of Eclipse Rotation (Comp w/ Wrath) = (3126.57 - 2728.60 ) / 2728.60 = 14.59% DPS increase
Value of Eclipse Rotation (Comp w/ SF) = (3126.57 - 2556.78 ) / 2556.78 = 22.29% DPS increase

5. The above numbers ignore the cooldown, so we need to determine what portion of our total casting rotation is taken up by the Eclipse rotation. We already know that the Eclipse rotation will last 20.4591 seconds on average. The total casting rotation is the sum of the precasting period minus the buffer cast, plus the 30 second cooldown.

Total Casting Rotation Duration = 30 + (4.7619 * 1.1833) = 35.6348 seconds
Eclipse Rotation "Up Time" = 20.4591 / 35.6348 = 57.41%

So,

Value of Eclipse (Comp w/ Wrath) = 14.59% * 0.5741 = 8.38% DPS increase
Value of Eclipse (Comp w/ SF) = 22.29% * 0.5741 = 12.79% DPS increase

As I said before these numbers are a best case scenario. In actuality the will be significantly smaller. This is because you will have to refresh dots and you will miss some of the buffs due to movement and such. However, even if the benefit is cut in half you're still talking about a 4% to 5% increase in DPS from 3 points in Eclipse.

Edit: Down Ranking Eclipse A couple of people have asked in the comments that I value Eclipse point for point. I won't detail the math because it is really just repeat of the math above with a change in a variable. However, please know that I have used the same approach to come up with these numbers.

In the math above I showed that an Eclipse rotation is a 12.75% DPS increase over a normal Starfire rotation and a 8.45% increase over a normal Wrath rotation if you put a full three points into Eclipse. Please remember that these numbers were calculated under ideal circumstances.

If you put just one point in Eclipse it will increase your DPS by 11.27% over Starfire and 6.42% over Wrath. With two points in Eclipse it will increase your DPS by 12.31% over Starfire and 7.83% over Wrath.

This means that mathimatically you can receive 75% to 90% of the benefit of Eclipse from just 1 point. This is because of the 30 second cooldown. It has a greater impact on determining when Eclipse procs then your Crit chance.

Personally, I will proably still put 3 points into Eclipse, eventhough those last couple of points will have a much smaller impact then the first point. I will do this for two reasons. First, there isn't a great talent to put the points into. I would love to pick up Gale Winds but It is hardly necessary. Second, I want to cut down on the randomness. Eclipse is already very dependent on the RNG and I want to reduce some of that impact.

TL:DR Version

The Eclipse buff is very powerful buff. In an ideal situation it can improve your DPS by up to 12%. However the ideal situation is near impossible to achieve, but even if you miss half the buff your still increasing your DPS by close to 6% for a 3 point investment.

This indicates that Eclipse is a must have talent for raiders. However, you don't have to invest a full three points into the talent. You can receive a majority of Eclipse's value by spending just one point on the talent. However, Eclipse is already a very random talent, and for that reason I recommend investing 3 points to reduce that randomness.

10 comments:

Genetic said...

I agree that 3/3 Eclipse does prevent randomness but can you please show the math for the difference in DPS between 2/3 and 3/3 Eclipse under a patchwerk style fight? If random things are random then if the difference in dps under perfect conditions is only say 1-2% then wouldn't that point be better spent in other more worthwhile or utility talents?

wammes said...

I have enjoyed reading your blog for quite some time now. My question is related to the DPS numbercrunching you're doing. When i see figures hovering around 3kdps, i feel comfortable because i can relate to those numbers.
What's totally unclear to me however, is why are moonkins able to push well over 4k, even reaching 5k when theory only accounts for about 3k-ish.
Or to rephrase it: i feel much like the moonkin you're describing, but am no where near the 4k barrier

Wammes/ Darkmoon Faire

Anonymous said...

As like wammes, my moonkin is at about 3k DPS on patchwerk and I am nearly all epics. I believe I play well and pretty much maximize my DPS using proper spell timing and procs. My SF crits even hit for around 12-14k! However, my DPS still isnt venturing much from that 3k line. Is it the gear or the playstlyle that is keeping us "normal" raiding moonkin below the 3.5-5k DPS line.

Beefiest, Moonkin
Haumarush

Elthy-san said...

Wammes and Beefiest:

While I lack the authority of Graylo, I can offer two pretty standard answers to why you might be seeing the discrepancies you describe.

1. Group composition

If your 10 man raid group looks like mine, then you don't have heroism, replenish or a shadow priest for misery. Hopefully it is clear why your dps would be significantly lower than other people.

2. Group composition (I know, I know...)

Imagine a crazy good group. It takes them 1 minute to kill patchwerk. As a consequence, you spend almost the entire fight under the effects of heroism. I bet your dps looks really great. now if the same fight takes 5 minutes because your group is pretty new, suddenly the influence of heroism is muted, and your personal numbers look worse. (even though your technical performance doesn't get worse in the last 4 minutes of the new fight)

This happens more subtly with other cooldowns. A fight that lasts 2 minutes will give me worse dps than a 2:15 fight, because I get one fewer use of my trinket. Once you start looking at trees, starfall and Sundial, you realize that timing is everything. ;)

The last, and least popular option to consider, is that you just might not be as good as the other guy. Obviously I have no way of knowing whether you're competent or not, but I've gotten a lot of millage out of assuming I'm terrible, and if I listen to what other people say I might get better. :D

Graylo said...

@Genetic

I'll add the numbers, but off the top of my head the value of the 2nd and 3rd point are very small. For instance I said eclipse was worth a 12% increase against SF spam. I think 10% of that was from the first point.

Personally, I think the talent is to random as it is so I Put a full 3 points in it, and I don't have another place to put those points anyway. I don't need the mana, Brambles would help a little, Genisis sucks, and Gale Winds would be fun but not a big impact.

@Wammes and Beefiest

Elthy-san has made some good points. Group comp plays a lot into DPS, and cooldowns can mess with your DPS number, but here are a couple of other points.

1. WWS and recount are great tools to evaluate yourself, but start to lose value when you start comparing your self to other people not in your raid.

Remember that your situation is not the same as the theirs.

2. No one brags about their bad WWS report. While you may see a bunch of 5k WWS reports doesn't mean those guys do it consistently. I saw a WWS a while back where some one had a 6k+ Patchwerk. What he didn't mention was that he had +80% crit rate on both SF and Wrath.

3. Actual numbers will rarely equal Theory number. Theory crafters generally work in ideal situations. We don't facter in resists, lag, movement, and a combination of other things that bring actual numbers down. If a spreads sheet says you sould be doing 4k but your only doing 3.5k don't sweat it. Your actually doing pretty well.

4. Finally, I know what it's like to stress about your DPS. We all want to be the best, but lets put it into perspective. If your numbers are fairly close to that of your guilds then you don't have a big problem.

We all have room to improve, but there is no reason to stress about it if no one else is. Review your guilds WWS and look for ways to improve. other then that there isn't much you can do.

Genetic said...

The other question which I have always wondered is why 1/3 Moonglow or 2/3 Moonglow instead of 1/5 or 2/5 Genesis. (I don't spec Nature's Reach because my alt is a afflic warlock so I don't want to get used to extended range casting.) Moonglow, as has been proven, sucks from a mana conservation standpoint being the weakest talent, whereas Genesis buffs about 15%-20% of a Moonkin's damage. Yes the points are expensive, but increased damage is still increased damage.

Graylo said...

@Genetic

Genisis is such a bad talent that people generally just don't care and prefer to have the safety of mana regen/mana savings.

If I get to a point were I feel compfortable getting rid of all mana regen then I will put one point into Geneisis, but it is so minor that it isn't worth woring about. Each point is worth about a0.2% DPS increase at max. Its just not that important.

Regarding Natures Reach, I have an affliction lock also. I would still pick it up, You will be surprised at how important those 6 extra yards are at times, and the threat reduction is very nice also.

Anonymous said...

was just looking at gear...
http://www.wowarmory.com/item-info.xml?i=40439
Mantle of the Eternal Sentinel
these shoulders need to be praised more. These look like best in game for Moonkin atm. I would say get these over t7 and get the other 4 pieces for the set bonus.

(these shoulders drop from sarth no drakes. have seen them 4 times and have 2 pairs lol)

-Beefiest, Moonkin
Hoamarush

Genetic said...

From a pure dps standpoint the Eternal Sentinels are probably the best dps caster LEATHER shoulders in game. But there are probably several cloth pieces that overshadow it. The problem with those shoulders is that Boomkins will want 4pcT7 for the 5% crit bonus. Looking at Head/Shoulders/Chest/Legs/Gloves you can see that the Tier 7 Shoulders are very well itemized, no wasted budget on spirit and it has hit which is very rare on Leather Caster items. The weakest slot compared to alternative non tier items is usually considered the legs. Combine this with the fact that you can buy shoulders with 60 badges of Valor and this means that most boomkins will always use their Tier7 shoulders.

Graylo said...

@Beefiest & Genetic

The Mantle of the Eternal Sentinel is good, but it doesn't hold a candle to the T7 shoulders.

The T7 shoulders have more of every stat except for crit rating and the 35 hit easily makes up for that. The only way the Mantle is better is if you don't need the hit or the set bonus.

As Genetic said, the generally accepted slot to get away from the set bonus is the lets, because there are a lot of great lvl 226 legs out there.