DPS system? Formulae

[smoothielover]

Ive seen many parsers etc with DPS worked out and a realtime one also.

i look at Seq's combat and mmob screens alot as im a stats whore and notice its based on time from mobs start to finish or somehting similar which throws out the figures significantly if mezzing etc.

now i also know its not really possible to work out the formulae for calulating DPS off Haste damage etc because of varying mobs.

i have yet to get to the riversource and parce a long file but i also dont want to drag a chante etc to workout the best calculations from it.

in case any dont know the riversource is like a lvl1 mob with a billion hitpoints...doesnt regen too fast and cant hit/doge etc great for a static mob test.


well anyways i ramble a little..

Is there any way we can clean up DPS system a little on seq...

i get random mobs in my list from zoning
i get issues when i faction tag a mob it throws dps out by miles.
i get issues when im not actually attacking a mob for 100 percent of time...ie oor or not targetted etc.

i expect if we look at the target and attack = true to use as a counter that would help ALOT. it still doesnt help the OOR stuff where you are Fluxed or such away from a mob. can we also remove the time between the first out of range message and fir next swing landing.


also just and idea but on the average dps wouldnt it be wise to remove the top and bottom 10 percent? that would cater more for the od caster in the mix etc. makes it lso a truer average i beleive

anyone have any odea how the MOB defensives are worked on? is there a logical way to estimate a mobs defensive rating so we could build a curve?

[S_B_R]

I usually look at just the "Current DPS" on the "Mobs" tab, it seems to be mostly accurate. That "Average DPS" is usually totally out of whack.

But I would also like to see this portion of SEQ brought up to par with the rest of the App.

[Manaweaver]

Something to keep in mind is that a majority of pure casters don't both meleeing... so if attack = true than a pure casters DPS would be zero, constantly.. regardless of how much damage they did. Since auto-attack wouldn't be on. Its a real hazy line, the start and end of a battle. At least in terms of stuff that is going on.

[smoothielover]

wouldnt a casters dps be irrelevant?

DPS for casters is mathmatically based. resists are not consitant i find so its nor really possibel. wouldnt the casters prefer a Spell Aggregate only like in the combat window?

ie (Spellname (231)) number of cast number of hits max hit min hit average Number of partial resists and posibly split this by mob type? name lvl?

DPS is more a melee based issue i thought. Its more variable

so we would have a new panel names like melee DPS and Spells Totals. Now i am a melee so i dont know for sure what a caster wants but if we are looking at devising a mod to increse parsing accuracy of DPS and such which i am i think we really should Split it to two areas ie like hitmode.

also a Pet DPS would be kewl too if possible. necros mages Sks etc

this is a wish list for me as a Stats whore player. I dont code well enough to help with the project source but i do have skills in the areas of networking etc & and mathmatics & Programming and realise its not QUICK FIX material but if any of you Developers play melee alot and particularly like me as a warrior im sure you would have used parsers etc before.

It seems a little lacking in comparison to the other areas of seq i expect because of parsers being available.

[whiteviperx]

Has anybody looked into this issue in farther?

[S_B_R]

it's working again. The average DPS still is out of whack. Zaphod got it working again a couple patches ago.

[perlmonkey]

I spent a lot of time writing a log parser that tried to calculate dps numbers, and I've come to the conclusion that dps is a meaningless term.

Perhaps a more cogent term would be dph (damage per hour), but even then you are making certain kinds of assumptions.

Let me pose an example: you're a caster. You're nuking the mob, and you see this in the log:

Tim delivers a critical blast (2300)!
Tim blasts a gibbous sandworm for 2300 points of damage.
Bob blasts a gibbous sandworm for 1200 points of damage.
Joe delivers a critical blast (3600)!
Joe blasts a gibbous sandworm for 3600 points of damage.
a gibbous sandworm curls up and cries for its mommy before falling silent.

Ok, so how much damage did Joe do? In practice, probably next to none. But, it's hard to tell that in the log parser, but one thing is for sure: all three of them spent the mana for their spells, so over the course of the whole session/encounter/raid/adventure/whatever what they just did was effectively INCREASE that mob's HP total.

dps in this case would usually be calculated by dividing damage listed by time engaged. I find that a much more meaningful number is damage listed divided by the time time from "go" to "gate". I usually calculate this in terms of ticks for reasons of preference, but seconds are fine too.

When you do this, you start to see some of the things that make EQ hard to figure any other way: the benefits of mana regen to sustainable damage for casters; the benefits of styles of pulling for various mixes of caster and melee; the damage value of buffs (I'm always amazed by shamen who don't realize that they can do more damage with focus than they can directly); the pain caused by rezzing downtime, etc.

My next log parser will be one that analyzes "encounters" such as an enitre raid or an entire LDON adventure. When you're taking on monolithic mobs (no pun intended) like AoW, Emperor SS, or planar gods, you certainly end up thinking in terms of that single fight as a discrete encounter and nothing else around it is on par, but that's rare even for raids.

Perhaps I'm just sticking on strategic vs. tactical thinking, but it seems to me that the conventional dps-wisdom is flawed.

[S_B_R]

Any measure of Damage over a given amount of time is as relevant as any other. Caster DPS is somewhat of a different issue from DPS of a Melee class. Hybrids further muddy the waters. But to say DPS is meaningless only to turn around and say Damage Per Encounter is important is silly. Obviously the total duration of the encounter can be divided into seconds and thus DPS.

The problem in the current Combat window is the "Average DPS" field. Personally as a melee player, Damage Per Mob is fine this me. Where DPS becomes useful though is when you have mobs of varing level and therefore HPs. All things being Equal except mob HPs, the more HPs a mob has the more Damage you're going to inflict. So DPS at that point is a valid measure of efficiency.

What it comes down to is Melee Damage Vs. Caster Damage or simply Burst Damage Vs. Sustained Damage.

[S_B_R]

And another thought. The best measure of overall efficiency in EQ would be based on XP over time, not damage over time. The way I see it, Damage Per Time measures personal skill more that efficiency. If I'm a Melee and I keep falling out of Melee range on mobs, I'm going to see that as a drop in my DPS. If I'm a Caster and I mis-time my nukes I'll see that as a drop in my DPS. I suppose damage measurements for Melee types mean more on a second by second (or tick by tick) basis. Alternatively damage for Casters would mean more on longer durations...

[perlmonkey]

Originally posted by S_B_R
And another thought. The best measure of overall efficiency in EQ would be based on XP over time, not damage over time. The way I see it, Damage Per Time measures personal skill more that efficiency. If I'm a Melee and I keep falling out of Melee range on mobs, I'm going to see that as a drop in my DPS. If I'm a Caster and I mis-time my nukes I'll see that as a drop in my DPS. I suppose damage measurements for Melee types mean more on a second by second (or tick by tick) basis. Alternatively damage for Casters would mean more on longer durations...

I disagree, but I understand where you're coming from. You're looking at this from a fairly basic point of view where there is only one variable: do you put out enough damage over time to kill a mob? I understand that, but there are, unfortunately, far more factors.

You mentioned efficiency, and that's one very good example. There's also the idea of damage mitigation. If all you want to know is: is my pig-sticker better than joe blow's pig-sticker, then sure dps tells you that (well... sometimes, unless you care about agro, damage absorbtion, or the host of other parameters in weapon selection).

But if you want to know if you have the dps for a particular event, you have to count ALL of the damage soruces including indirect by debuffing the mob's melee stats (and thus its mitigation of your damage), interupting heals, efficiency with which mez can be broken (this is why bards are a huge win in LDON for example), positioning (so that people don't have to back off mid-fight to re-position and so that you can get close to theoretical dps for classes like rogue).... there are tons of examples of how dps isn't a simple, single number. It's actually possible to do 1000 times a mob's HP in damage per fight, and if you do, it doesn't matter what your dps was, you've done that fight wrong.

I was describing the need for log and/or packet-stream utilities that look at the larger picture. That try to determine statistical corolations that we might not even be aware of (just as a for instance, if I told you that slowing a mob made it harder to hit, you'd scoff, but would you be SURE I wasn't right? Just an example, but one that illustrates the point).

The original question was couched in terms that made it clear the poster was assuming a quanta of damage over the duration of a single fight was wholy representitive, and I strongly disagree.

[S_B_R]

To me it seems what you are looking for is really outside the scope of what a simple log parser/packet sniffer could do.

To truely get all the detailed information you'd need to make the kinds of corrolations you describe you'd need complete logs from every person involved in the encounter(s) in question. Even then you'd have to parse the encounter many times, increasing the complexity of the calculations requires a larger data set.

My view maybe simplistic but it is well within the realm of what's available from the data at hand. And the more complex your statics become the less meaningful they become. Atleast in my book if you start walking that thin line to maximize your efficiency you'll undoubtedly faulter at some aspect. Perfection is unattainable when it depends on factors out of your direct control.

Thats the way I see it anyway. ;)

[Freakyuno]

You make some excellent points based on the "ability to complete a specific task" but I have to argue your wording and description as way more complicated than it needs to be, and purposly so for some reason.

I must preface this by saying as an expert EQ player, I understand everything your saying, and can invision the examples your after, but the conclusions your comming to are overly complicated in the Real game of EQ, or in a log parser, or lastly, in the term of DPS.


Ok, so how much damage did Joe do? In practice, probably next to none.

I think you left part of that example out, which is, that when Joe blasted, the monster only had 20 hp left. I understand what your saying, if it only had 20 hp left, and he spent the mana, the the monster may as well have had 3600 more HP. But in Real EQ, it doesnt matter, when the mob is dead, the mob is dead, and no longer doing damage to you. Unless you only have enough mana for 1 single spell per fight, then nothing is wasted or lost, you didnt "increase" the mobs HP. And for all intents and purposes, you can "count" on that much dmg in any given time.


if I told you that slowing a mob made it harder to hit, you'd scoff, but would you be SURE I wasn't right?

Again, I understand what your meaning to say, but your "ability" to hit a mob has nothing to do with how much or often it can hit you. If what you really mean is, did you know that slowing a mob reduces the amount of damage in a given time a melee can do to said mob, then you are absolutly correct, factoring in riposts, and damage shields, but you will "hit" a slowed mob just as much as an unslowed mob.



It's actually possible to do 1000 times a mob's HP in damage per fight, and if you do, it doesn't matter what your dps was, you've done that fight wrong.

Couldnt agree more, but this is exactly where making an issue over complicated is a problem. In Real EQ, your DPS has everything to do with this., as well as "tactics" that have absolutly nothing to do with DPS, like pullling mobs away from others, stuns, killing healers first...ect.

Example:
Mob A has 100 hp
You do 1 dmg per second
Mob B begins complete healing Mob A at 20 HP
Complete heal finishes with mob A at 10 HP, Mob A now has 100hp

Now, change your "dps" to 30 and see what happens to the example above.


In conclusion, Pearlmonkey, it's obvious you have an excellent excellent working knowledge of the game, and how everything effects everything else. Your type of understand is what makes great players, and raids won. But you cant rope everything into the DPS category and call it a false figure.

DPS is simply the amount of damage you do in a certain amount of time, in this case (per second). This is obviously a highly variable number. I used to parse in real time with my char using a "realtime average" of DPS using the last 15 seconds of data compared for average to all other 15 second segments, and over time, you can accuratly predict what you can expect to do in "DPS" to a mob at a given time, and that is what people are looking for.

I must say again for the people that read only the first and the last sentance, if your only speaking on calculation regarding sucessfull completion of a single target, your examples hold true, but do not disprove DPS.

[S_B_R]

Yeah what Freakyuno said. ;)

[perlmonkey]

Thanks for the detailed reply. I do appreciate the input.

Originally posted by Freakyuno
You make some excellent points based on the "ability to complete a specific task" but I have to argue your wording and description as way more complicated than it needs to be, and purposly so for some reason.

Ok, I'll accept the criticism, that's good. But, don't presume to know that your perception that I'm making this more complex than it needs to be is a) something I would agree with or b) something I intend.

Not flaming you, just want to keep those two concepts separate.


I think you left part of that example out, which is, that when Joe blasted, the monster only had 20 hp left.

Actually you're making my point there. You simply don't know. You can estimate the mob's HP and say "I see 9980 dmg being done before Joe's nuke and this mob usually has 10k HP."

My point was that if you look at a log file and say "we put out 13k damage in that fight", you're simply wrong and all of the log parsers I've seen will tell you exactly that. Why are you wrong? Because Joe nuked late and that meant that someone ELSE had to make up the damage that he could have done, had he thrown that nuke early. Now, if he was not hurting for mana AND he'd done everything that he could so far that fight without getting agro, then this does not matter... that makes it even HARDER to figure out what the numbers in a log mean!

It's messy as heck to think about dps in terms of both mechanics AND skill, but just as a tank who doesn't step out of melee range holds agro better than a tank who does, "dps classes" who apply large amounts of damage at the wrong times aren't really doing as much damage as they (or log parsers) think they are.


But in Real EQ, it doesnt matter, when the mob is dead, the mob is dead, and no longer doing damage to you.

True! I absolutely agree with you there. I have to admit that part of my logic here is based on having done WAY too much LDoN recently, and I do think in terms kills per unit time and the like now. I have to do more outside of LDoN so that I stop skewing everything that way. However, unless you're just doing easy trash punctuated by large bosses all the time, thinking of killing a single mob as the only objective is not terribly useful.

Also, in terms of damage, there are many ways to apply damage and edge conditions like wasting extra damage on a nuke can help you to choose between them. I've actually been taking advantage of that in a recent project. I wrote a simulator that walks through all of the possible combinations of spells given a "fight duration", "mob hp", "caster class" and "caster level". One very important thing that I do in there is to try to optimize how much overall damage you can fit into the span of time that a mob is going to be alive. To that end, I even simulate other damage sources (it was a revelation to me that if everyone in my party uses this tool, its results became incorrect because the fight time dropped so much!)


Unless you only have enough mana for 1 single spell per fight, then nothing is wasted or lost, you didnt "increase" the mobs HP.

Ok, here I have to disagree. If you fight a mob that has 10k HP and by the time you're done you've done 100k damage, you have effectively increased its HP in the only sense that dps calculations care about, e.g. how much damage must you do to kill the mob. There are many reasons that might happen: the mob mitigated some of the damage through healing or other means; the mob's regen rate was almost equal to your rate of damage; you "wasted" damage such as I described above; or my favorite while soloing... the old "no agro super-regen" that happens when you get removed from the mob's hate list for some reason (note: this last one has become a big problem since the last patch, as there seem to be more cases where this mode is triggered and I've seen this happen several times in the middle of a long fight, especially while agro-kiting).


In conclusion, Pearlmonkey, it's obvious you have an excellent excellent working knowledge of the game, and how everything effects everything else. Your type of understand is what makes great players, and raids won. But you cant rope everything into the DPS category and call it a false figure.

Perhaps that is what I'm doing, but I'm just so tired of seeing people think in the limited sense of dps as if you could sustain that damage for ever, or maintain it in the face of a shorter fight.


if your only speaking on calculation regarding sucessfull completion of a single target, your examples hold true, but do not disprove DPS.

I don't think I was trying to disprove dps. Perhaps I was just knee-jerking and should step back. Damage is a useful statistic to track. Tracking it with respect to time is all well and good. Calculus ensues and we all get out our SOE-approved Plane of Math quested Slide Rule of the Clear Mind.

But when "dps" becomes the predominent (often ONLY) tool used in analyzing the effeciency of a fight, I get my hackles seriously up. Sorry if I was overbearing in doing so.

[Freakyuno]

Without drawing this thread out too much, since it's headed off on a tangent...

I think you make great points Pearlmonkey, and it became very obvious with your last post, that we are talking about the same thing, just describing it very differently. Our views on efficency differ somewhat, but here is what my opinion leads me to...

DPS for any skilled player is a cumulative and hindsite stat, it shouldnt be used as a sole means of ability to complete an objective, rather it should be used after said objective has been completed as a general mesaure of what you can expect out of your toon on similar objectives.

If I fight Spider X and it has 50 HP, and My DPS vs It's DPS leaves me at 90% HP when it is dead, then if I decide I want to heal up and fight Spider Y with 100 hp, everything else being similar, then I can "assume" that I will complete that fight with 45% hp when the spider Y dies. Of course, this is not ALWAYS true, but it gives us a reasonable basis, with just enough mystery to give the game a sembalance of excitment.

Now, back to the original topic of this thread, the actual SEQ DPS system. I personally dont use it. I use a realtime log parser when playing Melee, and could care less when playing a caster. But thoes that do use the SEQ Combat window would like to know that the information is as "accurate as possible" given the circumstance and variables. If they want to take it as gospel, and it gets them killed, their fault. If they want to use it as a guideline, to their advantage, then I think it would be nice to know the information isnt totally skewed by an unworking system.

My advice, people like you Pearlmonkey need to review the DPS code and revamp it! =P Because you obviously know what your doing. The best point of reference I could give you would be Y.A.L.P (yet another log parser). One of the best windows based parsers I have seen, with very accurate information.

edit: **
If you need a log to test with BTW...In skyshrine, the quest mobs have like 2 million hp, incredible regen, and dont turn around if your non KoS to Skyshrine, so you can step up behind them, and start attacking and go AFK for hours and get an incredible log with enough data.

[S_B_R]

I have to say that my point of view on the SEQ DPS system probably comes the fact that I don't really care all that much about absolute efficiency. After all I've been playing 1 character since Feburary 1999 and I'm still not 65 yet (just hit 62).

I just want to know that I am contributing as muc has I possibly can based on Previous experience with my person DPS. Also there is a level of one-up'smanship between myself and the rest of my groupmates. So added complexity really wouldn't gain me anything. The group I'm usually a part of seems very much overpowered for the content we are presented with anyway. I really wouldn't want to be anymore efficient at it...

About "Fixing" the SEQ Combat window. I'd like to be able to remove a selection of mobs (or mob) from the list, and some method of sorting combat would be nice. A wish of mine would be to have a graphical representation of DPS :D

[perlmonkey]

All good stuff. Thanks both of you for helping me to clarify my thoughts here. Right now, I'm working on cataloging ZEM numbers (they've changed since the last patch, slightly) and mapping those against some stats from each zone. I have a theory that the planes are not the best place to solo post-60 for exp, but I'm working on proving that.....

When I'm done with that, I'll look into what SEQ is doing and what YALP does and see if I can resurect some of my old Perl code and see what it did. Perhaps some sort of hybrid can be cobbled together as a library in C that I can distribute as a generic tool for parsing logs, packet streams or whatever. Then SEQ can use it, and I can mock it up into Perl for my command-line log parsing.

[Freakyuno]

Good stuff, I look forward to seeing what you come up with. I program not even a little bit in C++, but my logical mind usually can put together formula in short order, if you need any help with this, let me know.

I can tell you already without any tests at all, that for Solo XP, the planes arent very good XP at all, for several factors. It may be that you gain more XP per kill, but each kill takes 2 sometimes 3 times as long as a luclin or earlier zone.
If your in a full group, you easily make up for this.

The exception being elemental planes. Once you get to this level, a Solo or Duo is AMAZING xp, few of my friends quad kite in plane of fire for 57% AA xp per kite, and have enough mana with KEI to do 2 kites consectutivly. Each kite takes roughly 16 mins. So thats an AA point roughtly every half hour minus downtime. Thats pretty amazing.

[fester]

Originally posted by Freakyuno
The exception being elemental planes. Once you get to this level, a Solo or Duo is AMAZING xp, few of my friends quad kite in plane of fire for 57% AA xp per kite

This I just don't understand. I have long witnessed a cap of 11 percent exp to level. If you can obtain 57 percent AA, then AA exp would not be capped like level exp. Level exp cap of 11 times 4 is only 44 percent. That is 13 percent more than the cap or around 14 percent per kill in a quad. I have heard numbers of 75 percent for a quad before the ZEM nerf a while back (145 to 119 etc.)

[Cryonic]

Getting off thread, but... I wonder if they cap the level exp and either use that amount as the cap for AA or have no exp/kill cap for AA.