Spell interruption down cap and Aquaveil Tests

[ringthree]

Nah. That's not what you were arguing. Let me quote you :



You are clearly saying that at 99% the skill difference has an impact whereas it has none (lol sample size).

No, I think you are missing how he comes to that conclusion. If there is a cap at one point the next point down should have only 1% spell interrupts if it was simply a factor in the described spell interrupt reduction percentages. This is obviously not the case because below your cap point your observations see 25% - 33% interruption rate.

This simply means that something has to be going on besides straight described spell interruption rate to determine interrupts.

[rog]

RDM/NIN: The new supertank?

New?

[ringthree]

Merits: (8%) 8/256

Am I missing something here, or does this not match up?

[ringthree]

I'm guessing much like haste, spell interruption rates are rounded values, so the gear won't reflect the true value. Not of much important since you that in total, you need 102% or 77% Spell interruption. Now the question becomes, what can we change about FFXI in light of this new information?

Edit: Have you done tests on higher level mobs as well?

Honestly, it doesn't change much at all, but it could be helpful for casting sets for RDM. The real question though is if the juice is worth the squeeze.

[Kirschy]

Am I missing something here, or does this not match up?

I messed up there. I meant to write 16/256. (The way merits work would put 2/256 as 1%, so it would follow that 2% would be 4/256. (But it could just as easily be 5/256.) So 16/256 or 20/256 =/. My post was full of errors, I'll fix it tonight though -_-

[willriker]

What effect would you think an iron will Paladin will see with that fully merited? Fully merited Iron Will Paladin, and everyone in aoe range, would recieve 95% interuption down. paired with knightly, PLD would see 104%. Or with merits, 103%

I mean still kind of useless in application, unless you like tanking many enemies as pld/rdm

[ringthree]

What effect would you think an iron will Paladin will see with that fully merited? Fully merited Iron Will Paladin, and everyone in aoe range, would recieve 95% interuption down. paired with knightly, PLD would see 104%. Or with merits, 103%

I mean still kind of useless in application, unless you like tanking many enemies as pld/rdm

Izman and I tested it. It's 95% for everyone in range and it's very easy to get to 100% non-interruption... Don't ask how Iz ended up with Iron Will merits though... well, actually do ask him. The story is quite funny.

Interestingly enough, it's a real 95% interruption rate (observed from some rather weird but interesting tests on Sky gods) not the weird below the cap rates that Pchan observed.

[Nermie]

Here's my theory based on an idea using in computing (after all this is a computer game):

Assume that the value for a spell interruption down effect is stored in a single 8-bit memory location and that the memory is stored as follows:
bit 1: 1/1 space
bit 2: 1/2 space
bit 3: 1/4 space
bit 4: 1/8 space
bit 5: 1/16 space
bit 6: 1/32 space
bit 7: 1/64 space
bit 8: 1/128 space

What happens here is that any number that is not an exact multiple of 1/128 is truncated).

<Item>: <Defined %> => <8-bit binary representation> => <Resulting decimal due to truncation>
Club: 25% => (0.0100000) => 25%
Club: 25% => (0.0100000) => 25%
Torque: 5% => (0.0000110) => 4.6875%
Magnetic: 8% => (0.0001010) => 7.8125%
RDM AF1 Body: 10% => (0.0001100) => 9.375%
Belt: 6% => (0.0000111) => 5.46875%
Feet: 15% => (0.0010011) => 14.84375%
Merits: 8% => (0.0001010) => 7.8125%
Total: 102% => (1.0000000) => 100%

I didn't come up with this idea out of thin air. It's a simple computing method that is used to represent fractional numbers in a small amount of memory space. Of course it leads to truncation errors that lead to even more disgusting results.

What it all comes down to: PS2 8-bit limitation

[Spekkio]

Here's my theory based on an idea using in computing (after all this is a computer game):

Assume that the value for a spell interruption down effect is stored in a single 8-bit memory location and that the memory is stored as follows:
bit 1: 1/1 space
bit 2: 1/2 space
bit 3: 1/4 space
bit 4: 1/8 space
bit 5: 1/16 space
bit 6: 1/32 space
bit 7: 1/64 space
bit 8: 1/128 space

What happens here is that any number that is not an exact multiple of 1/128 is truncated).

<Item>: <Defined %> => <8-bit binary representation> => <Resulting decimal due to truncation>
Club: 25% => (0.0100000) => 25%
Club: 25% => (0.0100000) => 25%
Torque: 10% => (0.0000110) => 4.6875%
Magnetic: 8% => (0.0001010) => 7.8125%
RDM AF1 Body: 10% => (0.0001100) => 9.375%
Belt: 6% => (0.0000111) => 5.46875%
Feet: 15% => (0.0010011) => 14.84375%
Merits: 8% => (0.0001010) => 7.8125%
Total: 102% => (1.0000000) => 100%

I didn't come up with this idea out of thin air. It's a simple computing method that is used to represent fractional numbers in a small amount of memory space. Of course it leads to truncation errors that lead to even more disgusting results.

What it all comes down to: PS2 8-bit limitation

we're all well aware of the issue w/ converting a rational number in base 10 into base 2 and the rounding that happens. we see this with slow/haste, damage taken, and other effects. the issue is the exact values they use to round are rarely logical, and flooring can muddle the math some. it's much the same with needing "26%" haste to reach the real 25% gear haste cap. kirschy did a LOT of figuring on this and found it did not conform to fractions of 256 as we'd originally postulated and had to expand it to fractions of 1024 before (s)he found a fit. we know it's an implementation issue, but we actually have to grind it out to figure out the exact values which can be tricky w/ something that's random.

as for the ability to test if skill affects it, it's not impossible, but we can't use the same method of testing since we can't stack skill till we stop rolling interruptions. what we can do is look for a statistically significant variance. cast reraise 100 times while getting hit with 75 skill and 37 skill. see if you get more interruptions. try casting it against different lvl mobs that hit for roughly the same damage. try against similar lvled mobs that hit for different damage. you just need to show that there's some non-insignificant change and you can show if skill does or does not affect interruption rate.

[Byrthnoth]

Torque: 10% => (0.0000110) => 4.6875%

I like the above explanation, but figured I'd point out a typo anyway.



Apart from RDM/NIN, BLM and WHM are two other jobs that might get some use out of this. Obviously the downside for BLM is that it has to choose between massive -% physical damage taken and this, while the downside for WHM/NIN is that it has to switch its clubs.

If I'd had infinite inventory and spellcast, I might have used this on BLM/NIN when soloing Aerns in sea on the rare Ichi cast. Probably not though, -% damage taken is pretty sweet.

[Nermie]

I like the above explanation, but figured I'd point out a typo anyway.

Thanks - typo fixed (torque is 5%)

[Kirschy]

Yinnyth was nice enough to let me kill him /w 1k needles and test Earth Staff + Lv 70 JSE Earring for BLM. (These work in /256ths btw.) Earth Staff: 51/256, JSE Earring: 76/256. So fixing my mistakes and re-doing my numbers:

/256 system:
(Estimating values of course.)

Club: (25%) 64/256
Shield: (10%) 25/256
Torque: (5%) 12/256
Earring: (8%) 20/256
Belt: (6%) 15/256
Feet: (15%) 38/256
Merits: (8%) 20/256

Total: (77%) 194/256(~75.78%)

Aquaveil would need to be at least: 62/256 (~24.21%)

/1024 System:
(Again estimating values):

Club: (25%) 256/1024
Shield: (10%) 102/1024
Torque: (5%) 50/1024
Earring: (8%) 80/1024
Belt: (6%) 60/1024
Feet: (15%) 152/1024
Merits: (8%) 80/1024

Total: (77%) 780/1024 (~76.17%)

Aquaveil would need to be at least: 244/1024+ (~23.83%)

My bet is on the /256th system and Aquaveil being 25%~

[RyaWHM]

Regarding additive/multiplicative (maybe?):

WHM75/NIN37 with Lv.26 skill, casting Utsusemi: Ichi with -8% (no gear, just merits) and WITHOUT aquaveil:

attempted to cast 200 times ... succeeded 2 times >.>

----

WHM75/NIN37 with Lv.26 skill, casting Utsusemi: Ichi with -8% (no gear, just merits) and WITH aquaveil:

attempted to cast 200 times ... succeeded 3 times >.>

Doesn't this tell us anything about Aqua, now?

(I was testing on a Lv.63 Robber Crab, with ninjutsu skill=26)

[Kaylia]

with Lv.26 skill,

I feel bad asking this, but you should totally test on a lv ~40 monster, cap your ninjutsu, and try again.

Might need some eva- though, and the process would be a torture with low accuracy and sample size needed.

[pchan]

Regarding additive/multiplicative (maybe?):



Doesn't this tell us anything about Aqua, now?

(I was testing on a Lv.63 Robber Crab, with ninjutsu skill=26)

It does show that aquaveil is not very efficient in your test, which gives credit to the multiplicative system. In other words aquaveil will be efficient only if you got high skill (level 75 skill) or stack spell interruption%. So if you want to limit interruption on /nin you need to stack spell interruption, but if you are nin main, aquaveil will benefit you even if you don't stack. It would be interesting to test what the minimum of spell interruption is needed to get 50% interruption, it's probably less than 50%.

[RyaWHM]

The funny thing is, I bought a buncha whichever tool was cheapest, a while back, and was gonna skill to 37cap (114 I think?), but actually didn't specifically with interruption testing in mind... I dunno if that's good or bad.

I guess it's bad until I make use of the pitiful skill level, by testing some more angles on this. It's just, once I level it up, I can never lower it again...

you should totally test on a lv ~40 monster, cap your ninjutsu, and try again.

With the aforementioned irreversible..ness... in mind, I'd rather just hop on /DRK and test Drain or something. However, if I'm gonna do that, I think I'll start on Robbers again, because we have enough unknowns... I wanna cling to constants where I can (then this perfect-world, never-bored me, would move to a 40 like you suggested, and test /DRK there, as well as Ichi again)

[Argettio]

It is interesting that this is possible.

But based on what has been discussed, Pchan has proved by maxing out one side of the 'equation' you can reach a point where it is impossible to be interrupted.

There is still a lot of unknowns, like how skill/mob level/damage take affect the rate of interruption. I would be good to get a better understanding of these aspects so we can work out if there is such a thing as viable spell interrupt set for Jobs that want to conserve TP or aren't overly blessed with 'interrupt' gear.

But the info as it stands has some obvious uses, the main one being casting Utsusemi for mages.

RDM/NIN could become (even more of) a powerhouse with this sort of set:

Wands x2
Magnetic
Karasutengu Kogake
Solitaire Cape

= 81% (which should comfortably cap it will aquaveil)

And you have lost 2-3% haste. (I think that's the best trade off I can come up with for RDM).

[pchan]

RDM/NIN could become (even more of) a powerhouse with this sort of set:

Wands x2
Magnetic
Karasutengu Kogake
Solitaire Cape

= 81% (which should comfortably cap it will aquaveil)

And you have lost 2-3% haste. (I think that's the best trade off I can come up with for RDM).

Actually for rdm/nin it's better to use genbu's shield (you keep the benefit from shield and -10% pdt). The setup I use, /nin or not, is :

merit(8%)
club(25%)+genbus's shield
magnetic(8%)
body(12%)
belt(10%)
karasutengu(15%)
=78%


You can put pdt on cape/leg/hands/head/rings/shield. and swap to recast right before the spell goes off.

[Argettio]

Actually for rdm/nin it's better to use genbu's shield (you keep the benefit from shield and -10% pdt). The setup I use, /nin or not, is :

merit(8%)
club(25%)+genbus's shield
magnetic(8%)
body(12%)
belt(10%)
karasutengu(15%)
=78%


You can put pdt on cape/leg/hands/head/rings/shield. and swap to recast right before the spell goes off.

For Ichi I wouldn't want to give up my Relic body.

With spellcast you could get the haste gear in just before the Ichi finishes, although it would be a major FUUUU moment if you got hit in the moment your switch to haste gear.

For Ni there is a good chance (with lag etc) the spell will go off before you get your haste in. But then again, you should be able to time your Ni casts well enough not to really need an interrupt set.

Although I did forget about the merits, which would allow the removal of Solitaire Cape or Magnetic (or add willpower and remove the karasutengu which are 1mil on my server) in the set I proposed.

[Demosthenes11]

wait so this is literally saying at 102% spell interrupt down, you won't get interrupted ever?