Best in Slot for SMN after 2.1?

[Flips]

I was wondering if most of this data was parsed with food active? I see a lot of SMN threads indicating eating Buttons in a Blanket or the HQ version. I was just wondering as a lot of people are reaching i90 gear if Black Truffle Risotto will be a worth while food to eat? I was just curious because if that is the case people might want to provide parsing BiS data while eating the BiS food.

[EasymodeX]

I still see a solid 360+ with the BLM around, although I don't really like throwing out numbers when I don't have an actual parse to relate it to.

Yeah our BLM and I both average 320-360 depending on RNG and fluctuations with T4.

If anything though this all serves to corroborate what I previously mentioned about actual encounters being a poor metric for figuring out best gear sets.

Most fights are ok. Fights with significant AOE components are variable. T4 is exceptionally variable because of all the multi-target options. SMNs don't have a simple and direct AOE methodology, so different AOE group compositions will significantly impact how a SMN accrues AOE damage, or not. Hell, you could even cycle in Ifrit egi IIRC, since Enkindle is some type of AOE right?

Fights with only ST or nearly-only ST are pretty reliable. For example I parse +/- 5 DPS (1.5%) on Atomos basically every single time (unless the next groups sucks at standing on the pad). I'm usually always consistent on Cad as well, although this week I somehow parsed 20-25 DPS lower than normal, which really weirded me out.

[EasymodeX]

I was wondering if most of this data was parsed with food active? I see a lot of SMN threads indicating eating Buttons in a Blanket or the HQ version. I was just wondering as a lot of people are reaching i90 gear if Black Truffle Risotto will be a worth while food to eat? I was just curious because if that is the case people might want to provide parsing BiS data while eating the BiS food.

Buttons HQ > Risotto/HQ.

I assume everyone uses food for Coil. Food is pretty cheap and doesn't expire on death anymore .

In the end though, food only nets like 5 DPS or so, maybe a tad more.

[Seobit]

Those numbers are with HQ buttons. I checked again recently to see if risotto might be worth using now and while I don't remember the exact numbers, I just remember that we're still a long way off from getting full value from it.

I really don't like atomos parses lol. It's pretty much just a dumb training dummy inside of a dungeon that people try to use as a reference for doing optimal dps in an "actual boss fight". And your caduceus example is exactly what I mean! It's a short fight so small details like being able to fit in an extra blood for blood at the end can probably make a noticeable difference, and its short nature also prevents rng elements from being normalized more so factors such as crit, good or bad, can actual sway the outcome more than you'd like, for data gathering anyway.

Both garuda and ifrit have aoe enkindles. Garuda's is a centered aoe, much like flare is (about the same range too I think), while ifrit's is a cone similar to rock breaker. Cycling pets on t4 is definitely a bad idea regardless, it's a a harsh fight on SMN's MP.

EDIT: Just to be a bit more specific. Assuming the difference in gear sets is actually quite small (torin mentioned summoner's "worst in slot" is only about 2% worse than the BiS, and I'd assume a similar picture can probably be painted for most classes, bards and crit being a possible exception), even the smallest of variations in fights are enough to skew data more than what should be considered acceptable. If this is correct, in most cases the variations you will see in a fight from performance whether it be personal or group related, should be greater than any variations caused by the gear itself, making the exercise mostly futile. Any difference in results that is large enough to be cause for concern is most definitely unrelated to the choice of gear set.

[T0rin]

Just so I can get some help on improving it and try to make my 'findings' more open than they are currently, here is a link to the simulator I'm writing:

http://logicfix.net/sim/?wd=71&int=4...0&food=buttons

You can obviously plug whatever stats you want for those stat parameters.. valid food parameters are 'buttons' and 'eggs'.

I just added rouse/spur/enkindle/raging strikes, but the second cast of raging strikes is poorly logic'd out. The first one just kicks off at the beginning before DoTs are cast, I need to come up with some good logic for when to tell it to cast the second one. Enkindle for instance waits on both rouse and spur to be up, but since it is on a 5 min CD, it will only ever get cast once during the sim.

Speaking of which, it is a hard coded 5 min sim right now. Soon (once I test the impact of stress of allowing significantly longer durations), will open it up from like a 1-60 min sim, for purposes of averaging out inconsistencies in the rotation between sets because of GCD differences. (and find a good logic for casting RS)

Damage is averaged out to account for RNG variation, and crit damage is averaged in as well. There should be no RNG impact. SS buff occurrence rate is based on crit chance.. higher crit chance = more frequent crits = more frequent SS buffs.

Things I still need to add:

* Mega-int pot usage
* Optional party buff (INT, speaking of which, does anyone know exactly how much this is? Is it a flat amount or something like 3%?)
* Mechanics-based cast interruptions (random or specific intervals, i.e. ex titan fight)
* AoE sim
* ACN mode that uses B4B and IR to allow comparisons between ACN and SMN dps
* Gear/race picker to streamline stat inputs
* Different pets (Ifrit) once I figure out how to account for his auto-attacks accurately and how damage is derived on his attacks (compared to Garuda)
* Add more food
* Add mana counter, to determine if/when you might run out
* Figure out some way to intelligently involve Foe's Requiem

The first line of the sim is going to look something like this:

WD: 71, INT: 499, DTR: 253(13), CRT: 468(9), SS: 460, GCD: 2.38, PGCD: 2.86, CC: 14.1826, PC: 2.83652, CPP: 35, Food: buttons

WD = Weapon Damage
INT = Intelligence
DTR = Determination (number in parenthesis is the amount that food provides in addition to the base amount)
CRT = Critical Hit Chance (number in parenthesis is the amount that food provides in addition to the base amount)
SS = Spell Speed
GCD = Calculated GCD based on SS
PGCD = Calculated Pet GCD based on SS
CC = Critical Chance %
PC = Proc Chance for SS buff per pet cast
CPP = Casts Per Proc, average of how many casts it takes to proc an SS buff
Food = What food you told it to use, always defaults to HQ food

Disclaimer: It isn't perfect. It is a work in progress. If I add some code to it and it breaks it, you'll probably see a blank page. If you do something idiotic like pass int=yourmom, it's probably going to break. I haven't added much variable handling simply because there is no security risk currently as it utilizes no SQL connections, so there was really no need. I usually won't leave it broken for long if I do break it in the process of development though. If you find any significant flaws in the rotation logic (doing something it shouldn't, or whatever), please let me know so I can fix. Also, any ideas to improve it are always welcome. And don't nitpick stupid crap like Festers going off 0.01 seconds after an instant cast (Bio1, Ruin2, etc.), I did that for ease of scripting, even if in reality you would have to wait ~1 second to cast. It is only ever trying to fit a single OCGD spell into an instant's GCD, so it isn't doing anything that is impossible.

Edit: Also, I have yet to do BiS based on party buffs/food, so things may change based on that. Once I figure out how much party buffs are, I'll add it in and run some numbers through it. But of course, anyone else is free to do that as well.

[NexusArchmagus]

http://i.imgur.com/k3b7NIJ.jpg

Not any of the BiS sets and not accuracy capped (forgot to switch gloves back for coil).

200 dps on t4 is simply too low. I'm not saying you did bad personally, just saying there's more to the results you posted than gear choice (could be parser was messed up maybe). And besides there's too much variation for dps in actual encounters for it to be a reliable measure in figuring out the best set of gear.

Okay, not sure where I got 200 from, but I made ACT combine my DPS and my egi's and it seems to fall more in line with what you have there!

http://imageshack.com/a/img823/20/nql4.jpg

But you said you weren't using either set, so... who knows? lol

Were you using Zenith or Allagan book?

[Seobit]

Gear used was http://ffxiv.ariyala.com/LR3E

[T0rin]

Just ran some numbers..

dtr bis, buttons = 101258 (top)
dtr bis, eggs = 101073 (-0.18%)
crt bis, buttons = 100828 (-0.42%)
crt bis, eggs = 100660 (-0.59%)
hybrid bis, buttons = 101020 (-0.23%)
hybrid bis, eggs = 100788 (-0.46%)

party dtr, buttons = 103729 (top)
party dtr, eggs = 103556 (-0.17%)
party crt, buttons = 103518 (-0.20%)
party crt, eggs = 103250 (-0.46%)
party hybrid, buttons = 103685 (-0.04%)
party hybrid, eggs = 103348 (-0.37%)

Between 3 different sets (albeit hybrid is only 1 item off of crt), there was a maximum different of 0.59% between the 'best' of the best, and the 'worst' of the best. And nobody should use deviled eggs.

[Flips]

Just ran some numbers..

dtr bis, buttons = 101258 (top)
dtr bis, eggs = 101073 (-0.18%)
crt bis, buttons = 100828 (-0.42%)
crt bis, eggs = 100660 (-0.59%)
hybrid bis, buttons = 101020 (-0.23%)
hybrid bis, eggs = 100788 (-0.46%)

party dtr, buttons = 103729 (top)
party dtr, eggs = 103556 (-0.17%)
party crt, buttons = 103518 (-0.20%)
party crt, eggs = 103250 (-0.46%)
party hybrid, buttons = 103685 (-0.04%)
party hybrid, eggs = 103348 (-0.37%)

Between 3 different sets (albeit hybrid is only 1 item off of crt), there was a maximum different of 0.59% between the 'best' of the best, and the 'worst' of the best. And nobody should use deviled eggs.

Huh, interesting. Did you use NQ & / or HQ for d. eggs & buttons?

[T0rin]

Huh, interesting. Did you use NQ & / or HQ for d. eggs & buttons?

HQ

/10char