2013-03-26, 05:03 PM
Takebacker Wrote:What do the % increases look like for other attack speeds? Say i am at (5), what's the increase for going to (4) or (3)?
My reference is from LazyBui's thread: http://www.southperry.net/showthread.php...0Reference. It is a very old thread, and the data may be outdated/obsolete/inaccurate, but there aren't any other ways for me to obtain strong numbers otherwise. JoeTang might be able to provide you more accurate numbers...?
A quick survey reveals that most attack speeds have a decrease in delay at the final stage between -30 to -60ms when the delays are in the range of 600ms, with the exception of most larger-delay skills ((old) assulter, Dragon roar, soul driver... that sort of >800-900ms delay skills). But even then, other than Final Attack which is the odd one, the decreases in delays from Faster(3) to Fastest(2) are virtually always more than or equal to the decreases in delay from Fast(4) to Faster(3) [so e.g. for Bow, Strafe(4) -> Strafe(3) is -60ms, and Strafe(3) -> Strafe(2) is -60ms too; also, for XBow, the same skill gives -30ms and -60ms respectively]. By virtue of the fact that this is in fact not diminishing returns but a decrease in the variable of an inverse function, this gives rise to accelerating returns.
Therefore; the greatest benefit derived from a single-stage IAS is from Faster(3) to Fastest(2), and in general I average that to be approximately (t/600) / (t/660) = 10% increase in raw DPS. using nice numbers for convenience. It is usually a little less than that (see below if interested).
[everything below is extra stuff if you only want to know where that 10% came from]
The problem always comes from the fact that most/all changes in delays go in 30ms steps (possibly to limit computational turnover [taxing!] to 2000 a minute and provide a more practical form of numerical resolution). So, to be fair, it is more accurate to say that due to rounding errors, the maximum average increase in DPS from a single-stage IAS is about or less than 10%, and that almost always comes from Faster(3) to Fastest(2). Since we're looking for an off-hand rule-of-thumb for changes in DPS, I would say that logically speaking, the average change from (4) to (3) would be in the order of (1/660) / (1/720) ~~ 9.1%, or based on rough correction it would be closer to 8-8.5%. And likewise for (5) to (4) (Angelic Busters?) it would be 8.3% before estimating corrections.
You must understand that IAS looks so desirable because it is (probably) the absolute final modifier that affects DPS, and it is not exactly cumulative. All other stuff we have suffer from diminishing returns, especially PDR after Unlimited, but not (necessarily) IAS.
Just to state the obvious, summons aren't affected by IAS, but Final Attack-type skills are (which we have always approximated to proc based on the rate that we attack; the delays for FA-type skills are no longer accurate because FA mechanics have changed since BB, and so the approximation is indeed the correct way to do calculations). If your char uses a Hurricane-type skill (Phantom, BM, Corsairs etc.) or if you're already at speed cap then obviously you should obviously not be going for IAS!
If you're an XBM, a +1allpassiveskill IA ain't so bad I think, if I didn't do my numbers wrongly; might be better than IAS in fact. Ditto for BaMs because of ADA (I think!) but prior to CO. For everything else, %Boss is your man, followed by a tie with %crit and %BD. Poorer (read: unfunded/poorly funded, in the likes of <30%stat before MW) people might actually want to consider solid +WA/MA but it's decent everywhere. If you just want to grind, +1target... depends... [anybody with a Lumi out there tried this with the relevant hyper skill?] But at any rate, a single stage IAS from inner ability is pretty decent to begin with.
Hadriel