Connecting the DoTs: A Summoner Guide

[Sidra]

Ok - so I just went through my retainers and put on all the spell speed I could find, and with food ended up at 686 and repeated my test. I got the exact same 394 Burning Strikes over 20 minutes.

http://i.imgur.com/s01Djob.png

Here has been my testing methodology:
1.) Verify my stats then summon Ifrit
2.) Ifrit on Obey so OGCD usage doesn't impact GCD timings.
3.) Place Ifrit at target (didn't do this first time and lost 1 attack to movement). Cast Ruin 2 (so parse timer starts with Ifrit casting instead of use of his ability)
4.) type /echo end in chat log and stare at mini parse, hit "enter" on the first tick I see over 20 mins to end the parse.

Across 3 totally different Spell Speeds (354, 592, 686) I got the EXACT same 394 Burning Strikes. Also of note, then even my 354 Spell Speed test is faster than your high speed test (basically 3.05 vs 3.1 attacks per second). That could perhaps be attributed to our differing methodologies.

Additionally, if the spell speed buff we get (it's on the player not the pet) impacted the pet - the chances of it causing 3 different 20 min parses, with differing crit rates and base spell speeds, to have the exact same number of attacks would be super slim. It's much more likely it doesn't impact it at all, and out of everything we have here the only thing seemingly off is your 354 speed test. Everything else combined between the 2 of us basically shows speed having no impact.

I also find it interesting that Garuda's 10 min parse is exactly 200 attacks, while Ifrit seems to lose 3 attacks every 5 minutes if you assumed an actual 3 sec recast time. It would make no sense for them to mathematically slightly alter Ifrit's ability - I actually think that the combined animations from it's Burning Strike and Auto-attack cause a fractional delay in his ability usage which adds up over extended timeframes.

[Charybdis]

Snip

Just did my own, similar test.

http://i.imgur.com/WsQT6zo.png

http://i.imgur.com/tfCmrva.png - Buff Procc/Durations over 28:02 test for Dervy

549 CRT
727 S.SPD
1682 seconds

558 AAs
529 Burning Strikes

[Dervy]

24 enhanced procs
28*60 +2 = 1742s
1742/24 = 72.58333333333333 per proc

549 CRT = 11.2% Crit
= 4.43% Chance to Proc
/3 = 0.01478
1/ 0.01478 = 67.6504 seconds per theoretical proc

;D Glad that's working kinda accurately.

[Charybdis]

Dervy what data points am I collecting at this rate? Still seeing about damage and pet spell speed or does that seem like a bust? I'm getting all sorts of confused by the whip around information in the last few pages!

[Dervy]

As long as it has been verified that Spellspeed is not a factor in Ifrit/Garuda Damage Calculation, then just go and collect WD/INT/DET/CRT data whilst eliminating those pesky enhanced procs.

[Charybdis]

As long as it has been verified that Spellspeed is not a factor in Ifrit/Garuda Damage Calculation, then just go and collect WD/INT/DET/CRT data whilst eliminating those pesky enhanced procs.

Gonna give my Spell Speed one more solid boost to verify no damage changes then move on to ACTUAL stat collection with Ifrit and Garuda. I will eliminate pesky procs (ie they won't attack during them).

[ShinryuReishiki]

The impact of SS on the recast of Burning Strike seems to be very close, if not the same, as on the player. At 737 SS, I was getting 2.83s recast on Burning Strike; at the same SS, Shadow Flare is a 2.82s cast. That said, it seems like most of the impact is lost in practice.

There's an arbitrary delay between each Burning Strike, and it was consistently shorter with 354 SS than it was with 737 SS. When the it was on the shorter side, it nearly evened them out; the 737 SS set only pulled ahead when the 354 SS set had a relatively long delay.

https://www.youtube.com/watch?v=Zj0mcBs97eY

If both sets had an equal delay between Burning Strikes, the higher SS set would win, but as it is, it seems like it only pulls ahead occasionally. It's as if the pet's AI doesn't know that it can attack more frequently than 3s. As such, the only consistent benefit from SS seems to be the impact on the longer recast attacks.

[Dervy]

In other words, Ifrit-Egi is dumb as hell and can't tell the difference between an AA and Burning Strike. Got it. Could it be a weird interaction between Ifrits AAs and its Skills? Like how your AA's are delayed until the end of your cast, maybe when Ifrit is in an AA animation, he's unable to cast Burning Strike until afterwards?

At 354 SS, does Ifrits AAs and Burning Strike happen at the exact same time?

If so, that's definitely a slight bug that needs to fix. It essentially makes SS a worthless skill in the long run for Ifrit.

Seeing as Garuda has no AA, you could re-run the test with Wind Blade and another long Cooldown ability and see if you notice the same inconsistency. If you do, then Pet AI is dumb. If not, then it's definitely related Ifrits AI, or his AAs.

So if SS affects pet recast (and isn't visible on their tooltip which is absolutely dumb and needs to be fixed) and not damage, then that makes life a hell of a lot easier.

Praise Yoshida.

[Sidra]

snip

Can you do the same thing with Garuda? I just repeated my 10 min test, except at base 354 Spell Speed, and again got 200 attacks. So on Obey, using only Wind Blade, she gets off 200 attacks at both 354 and 564 spell speed per 10 minutes. This is baffling me.

[ShinryuReishiki]

Seeing as Garuda has no AA, you could re-run the test with Wind Blade and another long Cooldown ability and see if you notice the same inconsistency. If you do, then Pet AI is dumb. If not, then it's definitely related Ifrits AI, or his AAs.

Similar outcome with Garuda.

@354
Wind Blade recast: 180 frames (3.0s)
Delay: 6-7 frames (0.10-0.11s)

@737
Wind Blade recast: 170 frames (2.83s)
Delay: 19-20 frames (0.31-0.33s)

Also checked with Spell Speed buff up (737 to 884), doesn't appear to impact pet at all. It really seems like the only benefit from Spell Speed is being able to squeeze in more Aerial Slashes, Flaming Crushes, etc... Well, I imagine the DoT portions of Inferno and Earthen Fury would benefit.