Why You Shouldn't Use Fracture

[Leiron]

10 characters

I c wut u did thar

[Coramac]

Like I said, I disagree with the idea that the cost is irrelevant, but I agree with you on its tp/damage efficiency.
If Kitru made an error that is fine, but it doesn't really change the fact that he did agree Fracture is TP/DMG efficient, but more costly in terms of TP.
If you feel the tp pool doesn't matter that is fine, but from personal experience, there are cases where its best to stay as TP pool efficient as possible.

Take a look at the "bouncing" comparison between 1000 TP, 3750 potency attack and a standard BB combo. Also take a look at the 100 TP attack that does 33% more damage than the 75 TP attack. The TP pool is irrelevant. If run out of it, the rate of ability usage slows. If you are using more efficient abilities and run out of your "TP pool" sooner, it is irrelevant because the advantage you built with the pool is insurmountable to the less efficient abilities. You can continue the rotation longer without downtime, but it will eventually run dry just like a higher TP cost rotation will. The difference is that the higher TP cost rotation is ahead of the game when the lower TP cost, less TP efficient runs out.

Can't you guys make a post and then cut and paste edit it rather than just reserving with "10 characters" :-p

[Coramac]

Leiron,

Here's possibly a better way to conceptualize it.

We're looking at two different things when with Fracture here. The first is damage per usage. This is relevant for "burst" fights. Does it cause an increase in DPS when used? One of the ways to determine this is to compare to what we would be using instead. Normally, this would be a single ability, but here we use combos. We look at the average damage (potency) of the combo. Is it greater? Yes. So, on a shorter fight, we use it if X is met. X is the number of ticks that it are required in order to be superior. A burst fight is where a TP pool may come into play.

The second thing is a long term fight. All we need to know here is damage (potency) per TP and not on a net basis. Is it more efficient here? Yes. So, on a longer fight, we use it if Y is met. Y is the number of ticks that are required in order for it to become more efficient. This ignores TP regeneration because it is irrelevant. By looking at the most efficient usage of TP, it doesn't matter when if things slow down which they will eventually. Fracture does not run you out of TP significantly that much faster. It costs 2.67 TP a second, 6.67 per GCD if you spread the cost over the duration you want it to run.

Typically, in these circumstances, when we think in terms of efficiency, we are considering things that stretch out what we do over time longer. That is typically the case, but it is not always the case. When comparing Fracture to non-Fracture, there are essentially three phases if you want to conceptualize it.

1.) Fracture rotation and non-Fracture rotation both have "TP pools" undrained.
2.) Fracture rotation having a drained "TP pool" and non-Fracture having an undrained "TP pool."
3.) Both rotations having a drained "TP pool."

In phase one, Fracture rotation builds up a damage advantage over non-Fracture rotation. For phase two, the non-Fracture rotation has the advantage. It begins cutting into the damage advantage that the Fracture rotation built up. However, phase three occurs before the non-Fracture rotation has overcome its disadvantage from phase one. In phase three, the Fracture rotation has the advantage.

We know that the non-Fracture rotation will always be superior because it is both more efficient and more damaging. Let's pretend that the Fracture rotation were simply more damaging but less efficient. It would build up an advantage over non-Fracture in phase one. At some point after phase one, the non-Fracture rotation would overtake it permanently.

It's not the complete picture, but for practical purposes, it is most of it. You can create some odd situations like the 1000 TP 3750 potency attack. It immediately starts in its phase three. However, because all of its damage is completely frontloaded, it distorts things. All things being equal, frontloading damage is always better. The damage is so heavily frontloaded that the standard BB combo which I used for comparison, does not overcome the initial burst until the 18th GCD. If both were putting out what they averaged per GCD, the big attack would be at a disadvantage until the BB combo ran out of steam (which it does at 75 GCDs I think). So for GCDs 18, 19, and 20, the BB combo is superior. If the target were to die in these GCDs, the BB combo would outperform the giant attack. Every cycle, it gains ground on the giant attack. However, it is inferior a majority of the time before for it runs out of its "TP pool" and forever after it runs out. Such is the power of frontloading.

This relates to Fracture because, in a sense, it is frontloading your damage and is more efficient.

[Sapphidia]

The other thing is that some people are arguing for Fracture not being used due to THREAT, not DAMAGE.

Yes, over long fights, Fracture use will starve you of TP quicker. As Coramac says, if you're only worried about dealing as much Damage as possible, then that TP starving is fine! Your TP will have run out faster but you'll have done more damage with it, so the fact you occasionally have a few unused GCDs is irrelevant. More damage has been done for the TP. Fracture will ALWAYS be a DPS gain to use in any rotation that cares about damage done.

But for threat? No. Fracture is 300 potency of threat for 80 TP. Combo'd Skull Sunder is 600, Combo'd Butcher's Block is 1400. If Threat is remotely an issue then fracture shouldnt be used. I'd likely not use it on Titan for instance, depending on the group composition, because heavy white mage medica spam can always potentially out aggro the tank if the fight lasts ages. But when it's only DPSers getting close to your threat, chances are you'll have plenty of chance to slip Fractures in for higher damage at no real worry of losing aggro.

[Kitru]

This relates to Fracture because, in a sense, it is frontloading your damage and is more efficient.

You've gotten most of my point with the exception of this.

Fracture is *not* more efficient. The entire point of this discussion is that Fracture is *not* efficient and ends up being inefficient because of its 80 TP cost. Using Fracture increases your damage by a whopping ~2.3% (and, yes, I'm including auto-attack in this because we're looking at total damage without looking at total damage irrespective of TP cost) and TP cost by ~8.3%. That's not an efficient tradeoff. An efficient tradeoff would be if the increased TP cost were equal to or less than the increased damage. Since Fracture increases your TP burn by more 3 times the amount that it increases your total damage, it's not an efficient tradeoff. The only "advantage" it provides is in the frontloading of damage, as you have stated.

The problem with this is that the frontloading advantage isn't really an advantage over the long term. Every time you use Fracture it ends up costing you 75% of your total damage per GCD 2.5 minutes of straight combat later *and pretty much all endgame fights end up with the tank doing this*. The only time it would benefit you to use Fracture is when you run out of your TP reserve and are only able to spend what you're generating because that's when straight TP cost is all that matters (since you're only able to spend what you're generating rather than what you've got saved so base TP cost is actually what you're measuring against).

Here's some math:

I'm going to ignore auto-attacking for the purposes of this, which means I'm using the rotational advantage rather than total comparative advantage numbers since it simplifies the math a bit (all adding auto-attacks would do is reduce the differences between the two because we'd be multiplying the numbers by .8 to account for the fact that auto-attacks make up ~25% of total damage). Non-Fracture provides you with 100% damage for 72 GCDs and 77.5% DPS for all GCDs after (net loss is 14.44, which means that you'll need to have 22.4% of your GCDs empty to keep using it; 50 / (50 + 14.44)). Fracture provides 103.3% damage for 60 GCDs and 78.7% DPS for all GCDs after (net loss is 15.64, which means that ~23.8% of your GCDs will have to be empty to use it; 50 / (50 + 15.64); 1.033 * (1-.238) = .787).

At 60 GCDs, Fracture provides 3.3% more damage than Non-Fracture. At 70 GCDs, Non-Fracture overtakes Fracture on total damage dealt (60 * 103.3 + 78.7x = 6000 + 100x; x = ~9.3). At 120 GCDs, Fracture once again overtakes Non-Fracture (60 * 103.3 + 78.7 * (12 + x) = 72 * 100 + 77.5x; x = 48.08) and provides a further 1.55% more damage (for each additional GCD, not total DPS; the total DPS contributions would be so heavily diluted by all damage done before before that it wouldn't register until you tracked to the ten-thousandths) from attack skills than Non-Fracture.

So Fracture maintains a slight advantage until shortly before 3 minutes, Non-Fracture maintains a slight advantage up until 6 minutes have passed, and then Fracture maintains a slight advantage from then on out. To put it into context, after the first 3 minutes, you'd have to track to the ten-thousandths place to notice *any* kind of difference in total damage dealt, and, all of those advantages are actually ~33% higher than they actually are for total DPS because ~25% of total DPS is contributed by the auto-attack, which those numbers ignore. The 1.55% advantage in damage per GCD for GCDs after the 120th is actually a ~1.16% increase.

All-in-all, we're talking a *lot* about something that makes an absolutely *tiny* difference, which is what most of my point was. *Yes*, if you ignore TP cost, Fracture is a net increase to your damage, but the increase is miniscule. Factor in TP cost, and Fracture ends up being a very small net loss over time up until you get very far down the line. If you look at damage dealt within discrete windows, Fracture is slightly stronger at the start of the fight, becomes slightly weaker towards the middle, and becomes ever so slightly stronger once again when the fight is finishing (it takes several minutes of being without TP for Fracture to actually close the gap).

Fracture does not add real value to your rotation. It's, for all intents and purposes, a break even that becomes *slightly* on incredibly short boss fights (fights have to be *obscenely* long for Fracture to actually overtake Non-Fracture again; you'd have to fight for twice as long as it took you to run out of TP for Fracture to overtake Non-Fracture and, most of the time, I run out of TP about 90% of the way through a fight with Non-Fracture). Its presence on your bar isn't going to make one whit of a difference upon your actual damage dealt. It would be like having an attack that dealt the average damage of the BB combo while also costing the average amount: it doesn't really *add* anything but doesn't really take away anything either (and, yes, I've revised my position on this after redoing my math to account for Coramac's point about bottomed out TP causing baseline TP cost to become the fundamental factor). It amounts to being just a button you press for the sake of pressing something different (though a strong case could be made for it reducing your average Wrath stack generation per GCD, both by consuming TP that could be devoted to Wrath stack generation when bottomed out as well as the GCD consumption itself, which ends up costing you *something*).

[Coramac]

Non-Fracture provides you with 100% damage for 72 GCDs and 77.5% DPS for all GCDs after (net loss is 14.44, which means that you'll need to have 22.4% of your GCDs empty to keep using it; 50 / (50 + 14.44)). Fracture provides 103.3% damage for 60 GCDs and 78.7% DPS for all GCDs after (net loss is 15.64, which means that ~23.8% of your GCDs will have to be empty to use it; 50 / (50 + 15.64); 1.033 * (1-.238) = .787).

Where do you get 60 GCDs and 72 GCDs?

1000 / 14.44 = 69.25 GCDs

1000 / 15.64 = 63.94 GCDs

You appear to have extended non-Fractures time to empty by 2.75 GCDs and shorted Fracture by 3.94 GCDs. Correct this and non-Fracture never falls behind.

Again, Fracture never loses out provided it runs for a full duration.

[Sapphidia]

Darn, Coramac beat me to it.

And even so, I still don't see why Net Loss is needed. You have a fixed amount of TP to spend. It starts at 1000, it goes up 50 every 2.5 seconds regardless of whether you attack or not. Literally the only thing that matters is Damage Done per TP. Every calculation I make has Fracture coming out at being around 3% dps more than not using it. It doesnt matter that you expend your TP faster. Sure, a non-fracture rotation might appear smoother with less gaps but the gaps don't matter. If you've spent 1000 TP and done 1030 damage with it using fracture, and the non fracture rotation spends that 1000 TP and does 1000 damage with it, it can literally NEVER catch up in damage done. Never. Every point of TP that comes in once you bottom out at 0 and have to wait for a regen gets spent at 3% more damage per TP on average with a fracture rotation.

The fracture rotation might miss out a GCD here and there but the damage of that GCD is present in damage-done-per-TP. GCDs are irrelevant to discuss. The only time it matters to discuss GCDs is when you dont have enough GCDs to spend all of your TP.

Fracture in your rotation increases your Damage-per-TP-spent by about 3%, assuming you use it once every 30 seconds at most. If Fracture wasnt a DoT, but was a 300 potency instant attack for 80 TP, the way to do max DPS as a warrior would be to spam Fracture (well, between keeping Maim up) and then just auto attack for a few blank GCDs.

[Kitru]

Where do you get 60 GCDs and 72 GCDs?

I go off of 950, when you start hitting empty GCDs waiting to regen.

[Sapphidia]

There's no way 60 and 72 can be correct before bottoming out. That's almost close to literally spamming fracture.

As we've said, a non-fracture rotation (BB BB SE) is 64.4 TP per GCD, a net loss of 14.4. The average potency of a GCD is 196.7. It does 3.05 damage per TP.
A fracture rotation puts in 1 fracture every 4 triple-hit-combos, and works out at 65.6 TP per GCD, a net loss of 15.6. Average potency per attack is 204.6. It does 3.12 damage per TP on average.

Just from the above calculations you can see that the damage per TP is higher with fracture inserted. The ratio of 14.4 to 15.6 is 0.92. The ratio of 60 to 72 is 0.83. It's 64 and 69 as coramac says.

Non-fracture rotation will never, ever catch up in damage done over any time period. Taking that 69 GCDs...

After 69 global cooldowns, the point at which non-fracture hits zero, it has done 69x196.7 potency. That's 13,572 potency of damage. After this point it generates 50x3.05 damage every GCD (152.5 per gcd)
After 64 global cooldowns, the fracture rotation has done 64x204.6 (13094) potency and then hits zero TP. It then has five globals where it generates 250 TP. 250x3.12 is 780 potency over 5 GCDs, which is 156 potency per GCD.

So after 69 global cooldowns, both rotations are then running on the .77/.78 damage calc. The non-fracture has done 13,572 potency, the fracture has done (13,094+780) = 13,874.4 potency. Which is about 3% more. After this point, non-fracture will never catch up as it can only muster 152.5 damage per global cooldown whereas Fracture rotation can manage 156.

TLDR - Including 1 fracture every 30 seconds is ALWAYS going to increase your DPS by around 3% regardless of when you run out of TP.

[Kitru]

There's no way 60 and 72 can be correct before bottoming out. That's almost close to literally spamming fracture.

Going off of 950 TP, spamming Fracture would get you 31.67 GCDs so I'm not sure what you're even talking about.

I go off of 950 TP because it doesn't matter if you've got 59 TP or 0 TP: you're not using any of your attacks. As soon as your net loss reaches 950 (since TP costs are all in intervals of 10), you're at the point where you enter cost triage mode. Going off of 1000 ignores that.