Ruke's All-Purpose Thread of DA/Askar/Haste/etc and Win!

[fantasticdan]

Haste is a linear increase, you gain no additional benefit from stacking it like haste. Haste is an exponential curve, which is much much different.

Thanks for the info.

[Liquidedust]

Haste=>Double Attack is a linear increase, you gain no additional benefit from stacking it like haste. Haste is an exponential curve, which is much much different.

Don't post when tired RK not used to see you make mistakes!

[Tempestwolf]

Don't post when tired RK not used to see you make mistakes!

Agreed, I kept reading it over thinking I was making a mistake and just missing his logic <.<;.

~Edit~
To the OP,
How about adding an ability to gain enmity through taking damage to Defender as well as adding in dmg mitigation % in accordance to current defense. This will solve the hate issues of blood tanking while at the same time making blood tanking more viable all around. From what little I know about WoW they have something similar where you have to take damage to build up enmity.

That all being said, I don't see PLD/WAR becoming viable in terms of endgame tanking with those changes, or WAR for that matter, simply due to lack of CE enmity building tools. However, in terms of exping WAR could really step up as a tank.

[Amele]

Haste is a linear increase, you gain no additional benefit from stacking it like haste. Haste is an exponential curve, which is much much different.

Adding 2% DA when you have no DA at all, and adding 2% DA when you have 50% total DA before it, gives practically the same exact increase in damage.** Because of this, as you get into higher values of haste, each point of haste is worth significantly more than each point of DA. And as a result, full Ares would almost never be a better option than a typical build for TP. However it does appear to be rather amazing WS gear.

**It is actually slightly less of a boost, due to diminishing returns and the fact that if you land a killing blow on the first swing of a DA, you never see the DA happen.


Also, for Ares to be better than Askar (an addition of 7 STR and 17 attack), it would only have to add about 4-5 points to your melee average at a base of 180 damage/hit. Which, it can do pretty easily.

when viewed from %dps standpoint (which I think makes the most sense*) all the 'positive' stats (MAB, double attack, triple attack, crit, etc) actually have decreasing returns. (linear damage bonus for each +1 but less increase compared to your previous damage amount.)

adding that first point of DA gives a 1% dps boost (give or take a small delta for ws application since it makes you ws more and hit harder on the ws itself) adding each subsequent point gives you slightly less than 1%: to wit, 1.02/1.01 = 100.99% (.99% dps boost). at 50% DA; adding 1% is a dps boost of about 0.67% - and this is assuming you are attacking forever without interruption.

this happens because you can't DA an attack you already DA'd (likewise you can't crit an attack you already crit) - whereas until you reach the cap, you *can* continue to attack faster and faster (haste).

for very low amounts of a particular stat (say, like drg wearing a homam body for triple attack) the bonus is almost full value. (+2.5% triple really does give you just about +5% dps) but as soon as you start talking in the area of 10%+ (whenever you have a trait native) the subtractive bonuses like dual wield delay, fast cast, and haste are going to be better in equal quantities.



*it makes it easier to do apples to apples comparisons of inherently % based stats like haste and accuracy.

[mikazuki]

why don't we have a thread for producing a definitive spreadsheet that can be stickied and replace the 'what's better' topic with a 'how i use cells?' topic.

[Max™]

HOW I MINE FOR CELLS?!?!?

[Amele]

why don't we have a thread for producing a definitive spreadsheet that can be stickied and replace the 'what's better' topic with a 'how i use cells?' topic.

because we don't actually know enough to build the spreadsheet error free. (look at the number of assumptions that we have to make whenever we answer a question)

hell, every couple months Kirschy has to revise her haste model (kirschy's model is the most accurate) and the general community has considered haste 'solved' for like two years now.

as for the major questions that are still outstanding: to my knowledge we don't have a consensus yet on what pdif and cRatio functions look like in the post-boost-nerf-nerf world (particularly for 2handers). and we have never really had good confidence for how to get a useful pdif_average in the area where either pdif_max or pdif_min are constrained to 1.0. (above and below this area we've always assumed that pdif_average is the midpoint between min and max which may also be untrue).

zanshin rates are another thing I'd be interested in but not as critical as the pdif stuff of course.

[mikazuki]

i don't know the shape of an energy well, but that hasn't stopped me using the best estimate ^_^

i realise the model won't be perfect but we answer the questions based on the available model anyway, it doesn't matter much if it is revised after being collated.

i'm not aware of any revision to haste model since kirschy set it to 1024ths (or whatever), i think that measurement now has a sufficient degree of accuracy.

i agree pdif (and subsequently cratio) needs to be covered again, but it's easier than it looks; at least the part of hitting lv1 bunnies.
when calculating i was always lazy about taking average values, obviously if the average is in the lower or upper part of the boundary instead of directly in the middle that is problematic in terms of how valuable atk is; that solution can use some of the same data as finding the boundaries though.

zanshin is actually a much easier test, i think blinding potion (and of course no native dbl/trpl atk) will give you data very quickly. you can probably get the base 5% rate (and confirm it on non-samurai if you wanted) on a mob of relatively low lv to the character, then swish~swoosh!

[RKenshin]

I meant that DA is completely linear*.

Yeah, typo, not enough sleep lately. D:

Isla had a good idea in splitting this part of the discussion off of the PLD talk, probably best that way!

[Starr]

Rather than the Haste vs Double Attack graph already going around, someone should graph them as a function of increase in DPS.

[RKenshin]

It would look exactly the same. <.<

The haste vs DA graph plots total swings at different amounts of haste and DA.... Whichever is swinging more will do more damage. Not like a DA does more damage than a hasted attack or anything.

why don't we have a thread for producing a definitive spreadsheet that can be stickied and replace the 'what's better' topic with a 'how i use cells?' topic.

Some people have already done something similar to a smaller scale. I'm like 80% done with a large scale one and have been for a while, but not enough time to finish... It gets way too complex. I might come here at a later time for help with finishing it.

Posted/discussed it just recently in another thread, here:
http://www.bluegartr.com/forum/showthread.php?t=59886

[Starr]

but you're saying that DA has diminishing returns so wouldn't the DA line not be linear and have a decreasing gradient in an increase of DA vs increase in DPS graph where Haste Vs DPS would be increasing like the other graph

[isladar]

Split from the blood tanking thread, as this is an excellent discussion.

[kayn]

thx so much

[RKenshin]

but you're saying that DA has diminishing returns so wouldn't the DA line not be linear and have a decreasing gradient in an increase of DA vs increase in DPS graph where Haste Vs DPS would be increasing like the other graph

Well, it would be hard to accurately display the diminishing returns I was talking about. It would require guessing how often you would kill-shot a mob, and then applying the % of DA to that. But that is correct and that an adjustment can be made to it for that factor (my mistake then!).

However, even if it was a straight line it could still have diminishing returns, as another explained earlier where 0 to 1% DA does mathematically provide more of a swing increase to you than 50% to 51% DA.

[kayn]

I'd like to see your numbers showing the ares body out damaging the askar

[Aragon of Odin]

Here's how I understand it would work with Ares vs Askar.

Assuming you have no other DA just to make this simpler, and also assuming the 17ATT and 7STR will add 5 points of extra dmg per hit.

Lets say each hit averages 180/hit.

Using 100 swings as an example with Askar you would actually get 102x180/dmg = 18360 dmg.

With Ares body assuming it would raise you dmg/hit by 5 it'd be 100x185/dmg = 18500 dmg.

This was how I understood it anyways.

[kayn]

I think 5 is incredibly steap

[Amele]

I think 5 is incredibly steap

5 is completely reasonable.

we'll use a generic statless 90D great axe (since it makes the math easy). the fstr max is 18 so we won't hit it; not really important just putting it out there.

at 110~str for TP set we'll have an fstr of 11 vs. g.colibri. +7 str will add +2 fstr (remember this for later).


17 attack (it will sometimes be 18.) will be a boost of ~0.054 to cRatio - cheating and saying cRatio and pdif_avg vary proportionally, that'll be a boost to pdif_avg of 0.054.

to simplify things we will ignore weaponskills.

0.054 pdif * 101(fstr+D) = 5.5 damage. if we use a perdu voulge instead (107) then the amount of increase is about 5.7* edit: if you actually mocked up a proper TP build (I suspect fstr is higher if you can main an Ares's Cuirass) I wouldn't be surprised to see a final result of 6 or 7.

at higher ranges of cRatio the value could potentially be even larger above 2.0 cRatio (~2.5 Ratio) the increase to pdif_avg is probably larger than the increase to cRatio if pdif_avg ~= pdif_mid in the >1.2cRatio range. (a likely assumption)



*decimals in this case represent the approximate average damage over multiple hits.

[fantasticdan]

You have to take the extra tp gain/WS into account from the DA.