Questioning the design of TP

[Shaiandra]

Yes the fact that skill speed has exponential returns with an animation hardcap in FF14 is funny/annoying/lulz.

I haven't studied the mechanics on this; so it has exponential returns by subtracting from the GCD a static amount? i.e. every X skill speed reduces your GCD by 0.01 sec?

[Kevee]

That wouldn't be Paeon staying relevant though... that would be it increasing in value.

No, it still gives the same TP and skills still cost the same TP.

Again, it would provide the same damage as it does with no skill/spell speed. Skill speed doesn't increase damage, it increases DPS.

[Viridiana]

Skill speed doesn't increase damage, it increases DPS.

It also increases TP consumption, meaning there's a higher chance I'd have to use Paeon; meaning that, yes, Paeon is now more valuable, as it goes from "never used" to "used on occasion."

[Kevee]

It also increases TP consumption, meaning there's a higher chance I'd have to use Paeon; meaning that, yes, Paeon is now more valuable, as it goes from "never used" to "used on occasion."

Yes, but if the mob was going to die anyways, it's going to die faster for the same TP amount. If x mob takes 3000 TP per kill, then Skill Speed will only let you spend that 3000 faster. You're still spending 3000 TP to get the kill. All Paeon and natural regen do is help you to sustain DPS to reach that TP threshold.


It's not "never used."

Most people just don't run a comp where it's worth using. Most groups only run 1 melee, and a DRG at that.

[Shaiandra]

Yes, but if the mob was going to die anyways, it's going to die faster for the same TP amount. If x mob takes 3000 TP per kill, then Skill Speed will only let you spend that 3000 faster.

That might be accurate if you were the only one DPSing the mob. If you consume TP 10% faster with upgrades, that doesn't mean the enemy will die 10% faster (assuming the rest of the group's DPS hasn't increased by 10% as well).

If your TP consumption exceeds your total available TP, you run out of TP and your DPS suffers. I don't know the current slopes of TP consumption vs available TP, but if more skill speed increases the former such that TP consumed and available TP now intersect earlier and before the battle is over, well, Paeon becomes a larger DPS increase.

[Viridiana]

Yes, but if the mob was going to die anyways, it's going to die faster for the same TP amount. If x mob takes 3000 TP per kill, then Skill Speed will only let you spend that 3000 faster. You're still spending 3000 TP to get the kill. All Paeon and natural regen do is help you to sustain DPS to reach that TP threshold.


It's not "never used."

Most people just don't run a comp where it's worth using. Most groups only run 1 melee, and a DRG at that.

It's going to die faster at the same TP amount, yes, but only if I can actually get that much TP in that much time. Skill speed will not let me spend that 3,000 faster if I don't have it yet, which potentially means using Paeon where I wouldn't otherwise.

[pandabearcat]

Does anyone know if skill speed affects autoattack speed?

I agree with what you say, Kevee, but the problem lies in that, if you say had 100% haste, you wouldn't actually be able to kill it twice as fast.

If you spend on average 70 tp every 2.5 seconds, it'll take you only ~125 seconds or so to run out of TP (without invigorate). This equates to 3500 TP of damage.

If you instead say, had 100% haste, then you'd run out of TP w/o invigorate in only ~28 seconds, equating to about 1556 TP of damage before running out of TP, and doing ~36% of your normal tps for the remaining time. That's what I mean by punishing. You'd expect to do 3500 TP worth of damage in 62.5 seconds, but instead you only do 1556 TP in 28 seconds, and only 690 TP more until 62.5 seconds, ending up with 2246 TP spent in 62.5 seconds.

Sure, you did more TPS then you would've without any skill speed, but here's the kicker...because we're limited by TP regen, as time taken in the fight approaches infinity no matter how much skill speed you have your TPS in an arbitrarily long timeframe will be the same.

No matter how much skill speed you stack, as the fight gets longer (say it takes 15 minutes to kill a boss), your TP spending smooths out to 20 TPS (aka the 60 TP you regen every 3 seconds).

With Invig it turns into roughly 27 TPS, 28.3 TPS for dragoons.

Barring differences in TP efficiency, every class is capped at that TPS given the fight is long enough (>12 minutes).

EDIT: Here's the other huge nail in the coffin for skill speed: if you notice in the example above, if we extended the fight out to 125 seconds that the "control" case runs out of TP, the 100% haste trial does exactly the same amount of TP spent. Which means thus: if you ever run out of TP in any fight, you will receive absolutely ZERO benefit from any skill speed, because your average TPS over the fight is the same.

EDIT2: Changed wording of "damage" to "TP spent", as TP does not correlate directly into damage. Efficiency of skills matters.

[EasymodeX]

EDIT: Here's the other huge nail in the coffin for skill speed: if you notice in the example above, if we extended the fight out to 125 seconds that the "control" case runs out of TP, the 100% haste trial does exactly the same amount of TP spent. Which means thus: if you ever run out of TP in any fight, you will receive absolutely ZERO benefit from any skill speed, because your average TPS over the fight is the same.

1. This argument against SS requires an "infinite duration 100% contact" fight with no extraneous TP recovery (Bardsong). Most of those conditions don't exist in practical situations. Edit: also only really begins to apply to higher amounts of skill speed where the increased TP per second becomes "real".

2. The argument is not absolute even with those conditions since skill speed allows you to compress more attacks into buff windows. E.g. even if you have the same total # of GCDs (due to TP limits), more of those GCDs are spent while Blood for Blood is active. Hence SS improves DPS in that fashion.

3. As a practical matter, due to the way FF14 mechanics work with the "SLOW, LONG AS HELL" GCD, DOTs and buffs tend to have 'big' gaps or clipping windows. Skill speed can serve to close those gaps or inject GCDs to remove the clipping. This generally improves TP efficiency by itself.

4. The inherent weakness of SS (aside from the fact that TP limits its effectiveness) is that SS's main benefit is that it adds filler GCDs, not high-potency GCDs. In other words, potent attacks (like Windbite or Chaos Thrust) are typically on hard or soft cooldowns, so SS doesn't affect them. Filler attacks like Heavy Shot or the DRG thrust combo are generally weaker than the average ability.

I haven't studied the mechanics on this; so it has exponential returns by subtracting from the GCD a static amount? i.e. every X skill speed reduces your GCD by 0.01 sec?

Yes.

For example, assume animations didn't exist and you start with a normal 2.5s GCD. That's 0.4 attacks per second. Add 100 skill speed. That's a 2.4s GCD -- 0.417 attacks per second, an increase of 4%.

Now pretend you have a metric ****ton of skill speed and your GCD was so fast it was 0.2 seconds. That is 5 attacks per second. Then, add the same 100 skill speed. Now you have a 0.1s GCD. That is 10 attacks per second -- double.

Not sure if 'exponential' is the exactly correct mathematical term, but yeah it escalates in value.

Of course, with animation hardcaps, you run into the caps way before the exponential effect really gets going in a big way, so the point is moot. (sidenote: and TP drain >_<)

It's just conceptually funny.

Generally speaking MMOs should use the "delay = base delay / (1 + haste%)" approach, rather than the "delay = base delay - %haste". The delay of the attacks (e.g. a 2.5s GCD) isn't directly what anyone cares about or measures. People care about the # of attacks per second (e.g. 0.4 attacks per second), but it's more awkward to label and communicate, so devs tend to use "2.5 seconds per attack".

That formula is much more scaleable and it makes items easier to itemize. The "delay-%haste" approach only works when the haste amount is very low (as it is in FF14) or you have random hard caps (as you have in FF14).

[pandabearcat]

Mmhmm, I agree with what Easy said.

It is obviously in a purely theoretical scenario, but the base point is there. It might seem obvious, but if you ever run out of TP in a fight, skill speed should be the last thing on your mind.

If they actually increased resource generation with haste (again, % based instead of stupid based like it is now), it would be a moot point and skill speed would be purely beneficial, esp for the more attacks per cd like Easy said.

Furthermore like Easy said, your filler attacks tend to be your worst, and also somewhat inefficient, so thats already a blow against skill speed. With haste increasing resource gen, at least that wouldn't be a strike against you.

Also @Easy "attacks per second" or rather "damage per second" is a function of 1/GCD, so with linear scaling down with GCD this is an inverse (or rational) function, which is much faster with its asymptote at 0 (exponential decay doesn't have an asymptote).

EDIT: The converse of the "if you run out of TP, get less skill haste" statement is also valid, as you will do the most DPS if you run out exactly at the end of a fight. Practically, this is a horrible thing to do what with procs, movement, crits, etc, but in a general sense if you find yourself finishing fights with a ton of TP, skill speed is at least worth considering.

Though considering that it doesn't affect any DoTs, and you can't even stack enough of it to make a noticeable difference, its value is slightly higher than "zero" but significantly lower than basically anything else.

[Shaiandra]

Yeah, I understand why subtracting from the delay will scale like that, I just didn't know if FFXIV used such a formula. If I'm not mistaken it does so with defense and damage reduction too, where every X defense prevents 1% of base damage and thus its value grows as you get more of it too. Thanks for clearing it up.

Is Greased Lightning III also -0.375 GCD? Or GCD (after skill speed) * 0.85? It sure isn't GCD/1.15.