DPS calculator?Follow

Not really, no. Besides, those things are way easier to just estimate the DoT improvements of rather than go through the long way and calculating out damage in full.

In b4 BrimstoneFox or whoever maintains that one calc now.



Edited, Mar 4th 2010 2:23pm by bsphil

Wow this is a great tool. Thx alot for the link, it has helped alot.

Well, you're also missing the human element. Hell, I've gone up against people geared equally as well or better as I was and I've crushed them damage-wise just by being more mentally prepared. Riding JA recast timers, autotargeting the next mob, WSing immediately at 100 TP instead of sitting on it, etc.

5hit on a 437 delay gk?

Well yeah that's one of the main reasons I advised against using the calculator, lol. Besides, calculating DPS with entire sets is needlessly complicated and far too much work when it's way easier to just compare two items back to back and how they improve the gearset as a whole.

Edited, Mar 8th 2010 10:16pm by bsphil

It wasn't too hard to work out what changed once I got sick of waiting. The tl;dr is that even things people think they know can change overnight.

Ok I'm going to point out a couple things here.

First off the DoT formulas are not right in v4.0 onwards. (this is easily proven by adding 1 sTP to any gear set and noticing it will minorly change your damage output - which makes no sense unless it changes your xHit build). Upto including v3.2 I had a factor called 'Effective Delay' which just modified your natural delay/haste/DA/TA/acc. into one factor with 100% accuracy. This worked pretty well, but for some multihit weapons it certainly wasn't ideal. Masa added a nice table to calculate probability for any given round in v4, but made the mistake of scaling TP to account of DA or accuracy instead of time. We'll looking at trying to do a full probability table, but the math is very intensive. Its been a good 10-12 years since I've taken statistics or probability and I don't have much free time to work on it, so its been slow going.

2nd: the tool is used for comparing setups in ideal circumstances, which really don't exist. The DoT portion can be tested agains Fortifications without Lag. (ie checking DA rates, average TP return, average TP started with on a ws etc..) Other parses you can compare more generic information such as your max/min for regular melee or crits. Obviously when there's pauses in battle, you get TP from getting hit, you get feather tickled the mob runs away because a mage pulled hate, etc. etc. the DoT piece won't match up at all. But that's irrelevant for comparing gear (most of the time at least, some situations the continuedness of the fighting is likely to make a logical decision as to why an answer might not make sense). Also comparing the data to a random parse isn't entirely valid just because gear swaps might not be consistent, debuffs and mob levels may not be consistent (although kparse can filter mob levels).

3rd: it clearly does not calculate ever niche circumstance correctly, I never figured out a good way to have the 15% AGI bonus of the Rogue Armlets+1 work (but you can easily do the math your self and add the bonus - then the sheet does the rest of the calculations). It attempts to do somethings though such as estimate Souleater damage based on your HP - although I'm pretty sure it assumes you are not cured.(I don't remember and I'm not sure if Masa changed it, I set it up well over a year ago). Buffs can be tweaked very easily if you have the world's luckiest COR you can just go to that page and change the attack modifier to the % needed for 11s.

As a melee paying attention and not being an idiot with adequate gear will often make up for getting uber gear+1. (engaging mobs quickly, not WSing @1%, riding recast timers).

I don't see a Butachi +1 in the default gear list, so I think you're talking about a different calculator.



But why stop there, maybe people shouldn't use a parser either, just go out and whack a mob once and compare the 2 numbers.

Finally, its obviously the goal to make this as accurate as possible, but I would tend to think some changes (such as the ws delay) will make little difference for comparison sake because it will be equal between both, but when its easy enough to have a tool do it all, you might as well. The other obvious goal is to have the tool report average differences. (since that's going to be what most people care about). I'm happy to look at anything someone thinks is wrong. Whatever is incorrect with the formulas should at least be equal and without bias, which is good from a comparison point of view.

Ditto, of course it was.

Why be for one type of automation but against another? It doesn't make sense, its like being half-Amish.

If you think the damage formulas hold any validity carry it through to the end, the whole is merely equal to the sum of the parts. Its not rocket science nor is it hidden voodoo magic, its just a long series of equations where the output of one is the input to the next, running other calculations is neither 'easier' or 'more accurate'. (after all the hard work has been done here, doing it the long way is both 'harder' and 'more error prone'). Its no different than trying to scribble damage numbers onto a piece of paper when a parser can do it for you.

Not all of the equations are linear for starters (in fact I'd say most probably aren't).

2nd of all, comparing things which alter a number of factors starts to be come more complex.

Quick and dirty numbers obviously have some use and validity, but there's no sense in swearing by them.

(of course the converse is with the spread sheet is if you hit an fSTR tier or something it can make the difference appear to be greater than it generally would be - say on a different mob).

I know it's a boring topic, but it is actually a very major change in the math and is one of the most significant in a long time.

Take for example,
(a)WS->100+TP in 10 seconds
vs
(b)WS->100+TP in 30 seconds

Not including a 2 second WS pause in (a) causes us to overstate a melee's TP cycle (~DPS potential) by a whopping 20%. In the case of (b), we'd be overstating it by ~7%. Hell, back in the day when ~94% haste was achievable by RC and Apoc, one could literally be doing a full 6 melee hits and getting 100tp before the WS animation was even finished. And essentially never be out of WS animation. As you can see, the implications get exponentially larger as the time required for a TP cycle decreases. For practical application in particular, I can see this being a very big issue for Sam's trying to judge the worth of a 5 hit over a 6 hit build.

I won't do a real comparison, but this is just an example to show again what I mean. Let's take Hagun and put it at 20% equip haste for both a 5&6 hit build, and assume haste / double march, and let's just say 50% hasso/seigan and generalize it to mean 5% for these purposes. So haste total is 60% and delay is 180. 5-hit and 6-hit are meant to include the WS, so the time spent meleeing would only require 4 and 5 hits respectively. For ease let's not consider double attack, etc.

5-hit total time,
(450*0.4)*4/60 = 12 seconds
6-hit total time,
(450*0.4)*5/60 = 15 seconds

This estimates a total WS volume increase (again not including things like med, etc) of 25% (15/12). But in reality with a 2 sec WS pause, the real time it would take to get 100tp would be 14 and 17 seconds. And thus the more accurate increased volume would be 17/14 or ~21.4%. So the estimate was overshot by 3.6%. To anyone that's not very familiar with gear/build comparisons, that might not seem like a whole lot. But for those that are very used to this sort of thing, 3.6% is enough to make/break a "what's best" conclusion. It's huge, actually. Of course, this is just one example.

Right, all of which can be very quickly compared the same way gear can.

You don't have to get an extra swing, you just have to beat someone else's swing.

I suppose more important than DPS is how short a fight is. The shorter, the better, in terms of the purpose of DPS. When you think about it that way, it makes more sense. And if all the DD's had that 1% extra haste, the fight would be shorter every time. Assuming you have a 1/3 chance of being the last one to hit the mob, the fight will be however much shorter 1% haste makes it. Or so that's one way of thinking about it.

The WS Pause thing is a derailment of this thread, but I just came across another situation where it makes a huge difference in calculations (Soboro/Yoichi Sam PT w/ Dnc .VS. w/out).