New savant SLA DPS numbers

[Monkey_Archer]

Hmm... seems the thread was hijacked

Anyway, as promised some more realistic numbers for typical trash nuking, or multiple sorcs in raids:

An 8 second, 5 spell cycle that does not require quicken and has time gaps for clicky uses. It can be thought of as the sorc's typical single target trash nuking cycle, or the typical dps a 2nd sorc can do in a raid setting assuming the first sorc does all the DOTS.

1 Polar ray
2 Niacs
3 Frost lance
4
5 Polar ray
6 Niacs
7
8
<repeat>

Damage per cycle:
2077 polar ray x2
511.3 niacs x2
575.2 frost lance
------
3163.5 damage / 8 seconds = 395.4 dps

sp cost per cycle:
92 polar ray x2
4 niacs x2
6 frost lance
---
102 / 8 seconds = 12.75 sp/sec

Sustainable for 235 seconds (3.9 minutes) with 3000 sp

Replacing 1 polar ray with snowball swarm would decrease to 288.1 dps, but allow it to be sustained for 393 seconds (6.6 minutes)


A 6 second AOE nuking cycle.

1 <clicky/ice storm>
2
3 ottilukes
4 cone of cold
5 snowball swarm
6
<repeat>

Damage per cycle:
563.2 ice storm x 3 ticks
689 ottilukes
539.2 cone of cold
179.7 snowball swarm
-----
1971 damage / 6 seconds = 328.5 dps

sp cost per cycle:
10.2 icestorm /5
63 ottilukes
62 cone of cold
5 snowball swarm
---
140.2 / 6 seconds = 23.4 sp/sec

Sustainable for 128 seconds (2.1 minutes)


Mana dump without DOTS (assuming another sorc does that already) with eardweller and quicken.

1 <clicky/curse>
2 Polar ray
3 Frost lance
4 snowball swarm
5 niacs
6 polar ray
7 ottilukes
8 cone of cold

Damage per cycle:
2307.8 polar ray x2
639.1 frost lance
199.7 snowball swarm
284 niacs
765.6 ottilukes
599.1 cone of cold
-----
4795.3 / 8 seconds = 599.4 dps

sp per cycle:
112 polar ray x2
6 frost lance
5 snowball swarm
2 niacs
73 ottilukes
72 cone of cold
---
270 / 8 seconds = 33.75 sp/sec

sustainable for 88.9 seconds (1.5 minutes)

[mystafyi]

We had 60+ players all fully attacking epic raid bosses like Velah, Arraetrikos, etc.

How much scaling was there with so many people in the zone though?

[steelblueskies]

Hmm... seems the thread was hijacked

Anyway, as promised some more realistic numbers for typical trash nuking, or multiple sorcs in raids:

An 8 second, 5 spell cycle that does not require quicken and has time gaps for clicky uses. It can be thought of as the sorc's typical single target trash nuking cycle, or the typical dps a 2nd sorc can do in a raid setting assuming the first sorc does all the DOTS.

1 Polar ray
2 Niacs
3 Frost lance
4
5 Polar ray
6 Niacs
7
8
<repeat>

Damage per cycle:
2077 polar ray x2
511.3 niacs x2
575.2 frost lance
------
3163.5 damage / 8 seconds = 395.4 dps

sp cost per cycle:
92 polar ray x2
4 niacs x2
6 frost lance
---
102 / 8 seconds = 12.75 sp/sec

Sustainable for 235 seconds (3.9 minutes) with 3000 sp

Replacing 1 polar ray with snowball swarm would decrease to 288.1 dps, but allow it to be sustained for 393 seconds (6.6 minutes)


A 6 second AOE nuking cycle.

1 <clicky/ice storm>
2
3 ottilukes
4 cone of cold
5 snowball swarm
6
<repeat>

Damage per cycle:
563.2 ice storm x 3 ticks
689 ottilukes
539.2 cone of cold
179.7 snowball swarm
-----
1971 damage / 6 seconds = 328.5 dps

sp cost per cycle:
10.2 icestorm /5
63 ottilukes
62 cone of cold
5 snowball swarm
---
140.2 / 6 seconds = 23.4 sp/sec

Sustainable for 128 seconds (2.1 minutes)


Mana dump without DOTS (assuming another sorc does that already) with eardweller and quicken.

1 <clicky/curse>
2 Polar ray
3 Frost lance
4 snowball swarm
5 niacs
6 polar ray
7 ottilukes
8 cone of cold

Damage per cycle:
2307.8 polar ray x2
639.1 frost lance
199.7 snowball swarm
284 niacs
765.6 ottilukes
599.1 cone of cold
-----
4795.3 / 8 seconds = 599.4 dps

sp per cycle:
112 polar ray x2
6 frost lance
5 snowball swarm
2 niacs
73 ottilukes
72 cone of cold
---
270 / 8 seconds = 33.75 sp/sec

sustainable for 88.9 seconds (1.5 minutes)

for trash cycling you have to remember you need to align the actual dps calculation to the frame of per mob hp.

if you are pouring your first rotation into mobs with 2000hp each and a neutral(ie 0) resistance level, then you are wasting tons of damage by not simply waiting to cast the second polar ray. overkill is not dps. to call it dps is to muddy the issue. if a mob has 1 hp and you strike for 1000, you have dealt 1(!) damage.if this takes 1 second your potential 1000dps strike is now 1 dps. there you go, now all of you fix your math.

this advents a better approach being ray on target 1, niacs+lance on target two, ray on target 1(dead with small overkill), niacs and lance on target two again (dead with slightly larger overkill).

big numbers or spamming to spam is waste. especially when you are using a giant nuke on a sliver of health.

of course this also means one needs to seriously look at misses, or mobs meatshielding one another and spreading that damage unintended places compared to actual target.

----------------

the dragon etc shade was referring to were EPIC unreleased and likely unfinished versions, placed in a non scaling , temporarily modified explorer area.

think smugglers rest.

[Cartheron]

I personaly think that calculating single target caster dps on trash is pointless - if you, as a caster, are dpsing single trash mobs (competing with melees ) you are doing this wrong. Gather 20 of them, set up a killing zone (solid fog for -5 reflex save, web to reduce incoming damage, maybe waves of exhaustion to completly negate any danger and -3 reflex save) and then blast them with aoe cycle, achieving dps unreachable for any other class, with maybe exception of bb kiting.

With that strategy it is not important how long you can do this. Question is how many times you can do it per shrine and adjust number of mobs dragged into every kill zone to this number.

And btw, it is funny how most of poeple criticizing dps calculations for casters as unrealistic, perfect scenario, unreachable in game etc. are using in their argumentation values of max melee dps, which is completly unreachable in game too, calculated on calculator made by the same people that made caster dps spreadsheet.

[Arctigis]

I love doing math too and eagerly await your videos showing these theories in action.

I wouldn't. You'll get neg. rep for saying something so egregiously confrontational...

[Jaid314]

And btw, it is funny how most of poeple criticizing dps calculations for casters as unrealistic, perfect scenario, unreachable in game etc. are using in their argumentation values of max melee dps, which is completly unreachable in game too, calculated on calculator made by the same people that made caster dps spreadsheet.

it is a lot more likely to be able to deal close to your theoretical damage with a melee. while you do have to use the occasional buff, most of them are going to be used once every 20 seconds at most. everything else just comes from click and hold, for the most part (i suppose if you want to talk about chasing things down, well, casters have that problem too. except that usually there's a lot more time in between us casting and hitting the mob than there is between a melee swinging and hitting the mob.

[Arctigis]

How much scaling was there with so many people in the zone though?

Arraetrikos was CR46 I think. He was being nailed and still stayed up for ~3 minutes. I had curse, fully
stack Niac's, fully stacked Eladar's the whole time and emptied my mana bar with Frost Lance (SLA), Polar
Ray and Otiluke's. I was quite pleased with my DPS. I have to report a bug as he didn't stay still
the whole time (some of my spells missed) then got p*ssed and DBF'ed me in the face

[steelblueskies]

I personaly think that calculating single target caster dps on trash is pointless - if you, as a caster, are dpsing single trash mobs (competing with melees ) you are doing this wrong. Gather 20 of them, set up a killing zone (solid fog for -5 reflex save, web to reduce incoming damage, maybe waves of exhaustion to completly negate any danger and -3 reflex save) and then blast them with aoe cycle, achieving dps unreachable for any other class, with maybe exception of bb kiting.

With that strategy it is not important how long you can do this. Question is how many times you can do it per shrine and adjust number of mobs dragged into every kill zone to this number.

And btw, it is funny how most of poeple criticizing dps calculations for casters as unrealistic, perfect scenario, unreachable in game etc. are using in their argumentation values of max melee dps, which is completly unreachable in game too, calculated on calculator made by the same people that made caster dps spreadsheet.

nothing stopping you from doing your own work on this, and or writing a handy little Bayesian optimization or genetic optimization algorithm to sort through several interdependent variables, applied to exclusive per time slice options from a large list, with dependent constraints, to result in either optimization of the damage and sp usage, the damage and time usage, or sp and time usage... then you can add working in per mob hp, resistance, and saves, account for evasion, map out the frequency of all these variables across the entire game and per quest, and let it tell us whats generally best, whats best per quest, per mob, etc.

should only be slightly nastier and longer to write and do well then putting together a brute force password guesser. maybe just one or two orders of magnitude if poor constraints are used.

that is why you throw kudos for effort, while pointing out flaws, and acknowledging the tool for what it is- a model, good for a basis and to give a general sense, but incapable of reflecting things perfectly.

that is why you have people test. reality is the best simulator there is. for a game like this, ingame is reality for testing purposes. you wanna write a full simulator simulator to proof your math , like i said above. have at it.

[krogyy]

nothing stopping you from doing your own work on this, and or writing a handy little Bayesian optimization or genetic optimization algorithm to sort through several interdependent variables, applied to exclusive per time slice options from a large list, with dependent constraints, to result in either optimization of the damage and sp usage, the damage and time usage, or sp and time usage... then you can add working in per mob hp, resistance, and saves, account for evasion, map out the frequency of all these variables across the entire game and per quest, and let it tell us whats generally best, whats best per quest, per mob, etc.

should only be slightly nastier and longer to write and do well then putting together a brute force password guesser. maybe just one or two orders of magnitude if poor constraints are used.

that is why you throw kudos for effort, while pointing out flaws, and acknowledging the tool for what it is- a model, good for a basis and to give a general sense, but incapable of reflecting things perfectly.

that is why you have people test. reality is the best simulator there is. for a game like this, ingame is reality for testing purposes. you wanna write a full simulator simulator to proof your math , like i said above. have at it.

i consider bad form to throwing around terminology that most of your audience don't understand without an explanation.

besides your discussion does not makes sense anyway... after all, you need a mathematical model to perform the optimization upon.

[Cartheron]

nothing stopping you from doing your own work on this, and or writing a handy little Bayesian optimization or genetic optimization algorithm to sort through several interdependent variables, applied to exclusive per time slice options from a large list, with dependent constraints, to result in either optimization of the damage and sp usage, the damage and time usage, or sp and time usage... then you can add working in per mob hp, resistance, and saves, account for evasion, map out the frequency of all these variables across the entire game and per quest, and let it tell us whats generally best, whats best per quest, per mob, etc.

should only be slightly nastier and longer to write and do well then putting together a brute force password guesser. maybe just one or two orders of magnitude if poor constraints are used.

that is why you throw kudos for effort, while pointing out flaws, and acknowledging the tool for what it is- a model, good for a basis and to give a general sense, but incapable of reflecting things perfectly.

that is why you have people test. reality is the best simulator there is. for a game like this, ingame is reality for testing purposes. you wanna write a full simulator simulator to proof your math , like i said above. have at it.

I think we have a misunderstanding here, so I will try to clarify my point:

1. I am full of respect for everyone who is doing actual calcuations and test how things work.

2. Using single target trash arcane dps as proof that sorc are gimped because they cannot sustain it for all dungeon is the same as claiming that single target melee trash dps is irrelevant because arcanes can gather 20 mobs and nuke them at the same time, achieving few thousands dps and every melee that will try to help will be killed by multiple cleaves (for example) thus melees are useless and only piking. Both statements are not true.

3. Melee way of dealing with trash is incompatible with arcane way of dealing with trash - both methods are good, but arcanes are inneficient in single target trash dps, melees are inneficient in aoe trash dps.

4. Single target dps for arcanes is important on bosses, not on trash.

5. Single target arcane trash dps has NOT changed in last build, so all posts "sorc cannot achieve or sustain trash dps after changes, so they are gimped and unplayable now" are a bit odd in my opinion.

6. Calculating max dps for arcanes is difficult because of cooldowns and spell cycles, but it is only a matter of time before someone will find a perfect cycle. In the mean time, we can use some simple cycles that gives lower dps but is sustainable for a longer period of time.

7. In my opinion, single target arcane dps with 5 stacking curses was way to high. Currently, with only 1 curse, I am not sure. If nerfed to the ground acid spells will be buffed again (as Thoon posted) I think it can stay as it is. If not, maybe value of curse debuff should be higher, or lower but stackable 2-3 times.

[steelblueskies]

I think we have a misunderstanding here, so I will try to clarify my point:

1. I am full of respect for everyone who is doing actual calcuations and test how things work.

2. Using single target trash arcane dps as proof that sorc are gimped because they cannot sustain it for all dungeon is the same as claiming that single target melee trash dps is irrelevant because arcanes can gather 20 mobs and nuke them at the same time, achieving few thousands dps and every melee that will try to help will be killed by multiple cleaves (for example) thus melees are useless and only piking. Both statements are not true.

3. Melee way of dealing with trash is incompatible with arcane way of dealing with trash - both methods are good, but arcanes are inneficient in single target trash dps, melees are inneficient in aoe trash dps.

4. Single target dps for arcanes is important on bosses, not on trash.

5. Single target arcane trash dps has NOT changed in last build, so all posts "sorc cannot achieve or sustain trash dps after changes, so they are gimped and unplayable now" are a bit odd in my opinion.

6. Calculating max dps for arcanes is difficult because of cooldowns and spell cycles, but it is only a matter of time before someone will find a perfect cycle. In the mean time, we can use some simple cycles that gives lower dps but is sustainable for a longer period of time.

7. In my opinion, single target arcane dps with 5 stacking curses was way to high. Currently, with only 1 curse, I am not sure. If nerfed to the ground acid spells will be buffed again (as Thoon posted) I think it can stay as it is. If not, maybe value of curse debuff should be higher, or lower but stackable 2-3 times.

thanks for the clarification then, and on these points as presented i agree more or less.

i consider bad form to throwing around terminology that most of your audience don't understand without an explanation.

besides your discussion does not makes sense anyway... after all, you need a mathematical model to perform the optimization upon.

this whole thread is centered on a spreadsheet that is equivalent to napkin math in a physics problem.
and largely on people who want it to essentially reproduce the base model( ie the game itself), intermixed with those of us who see it for what it is- a very rough ballparking, and a place to start. see the discussion on a trip to mars via car for clarification on what napkin math works out to.

for terminology you do not understand, you clearly have internet access, please utilize it. might i suggest wikipedia for a starting place if researching in depth is not your cup of tea?

that said what we have is a problem involving multiple functions. spell @ cost *modifiers @ cost, with other elements such as time(cooldown, cast time) and we are trying to optimize one of a few things or find a point of intersection between multiple things at once(ie optimal solution)- sp pool is a constraint for example.

given that we have multiple variables that are themselves functions, it starts looking like a multivariate calculus problem, but given the mutual exclusivity of many of the functions in the time domain, this scenario falls more in line with a combinatorial ( or combination) optimization problem. see travelling salesman problem for a nice treatment. working through a system to run through all possible combinations to find one that meets our constraints (best damage/sp, best damage/time, best damage/sp/time, etc) without taking egregious amounts of time is where we get into methods outlined.
there are after all at least 25 spells capable of being loaded at once and potentially used in any order, and depending on cooldown of used spells, we may or may not restrict that list for any given time t.
then multiply that 25 possible combinations with all the possible modifications to base damage, critical damage, dc, and save, metamagic, and more and you get a very very large number of permutations just for a chain of two spells, much less more.

and bear in mind all of this is still in the realm of "napkin math" as we are still not accounting for variate mob hp, resistances, overkills,player execution errors, alternate gearing levels and so much more.

we are fixed on examining an approximate minimum and maximum, and some special alternate cases, with regards to a single monster.

to build a full model would be essentially to rewrite the game over again to test with.

[krogyy]

this whole thread is centered on a spreadsheet that is equivalent to napkin math in a physics problem.
and largely on people who want it to essentially reproduce the base model( ie the game itself), intermixed with those of us who see it for what it is- a very rough ballparking, and a place to start. see the discussion on a trip to mars via car for clarification on what napkin math works out to.

for terminology you do not understand, you clearly have internet access, please utilize it. might i suggest wikipedia for a starting place if researching in depth is not your cup of tea?

that said what we have is a problem involving multiple functions. spell @ cost *modifiers @ cost, with other elements such as time(cooldown, cast time) and we are trying to optimize one of a few things or find a point of intersection between multiple things at once(ie optimal solution)- sp pool is a constraint for example.

given that we have multiple variables that are themselves functions, it starts looking like a multivariate calculus problem, but given the mutual exclusivity of many of the functions in the time domain, this scenario falls more in line with a combinatorial ( or combination) optimization problem. see travelling salesman problem for a nice treatment. working through a system to run through all possible combinations to find one that meets our constraints (best damage/sp, best damage/time, best damage/sp/time, etc) without taking egregious amounts of time is where we get into methods outlined.
there are after all at least 25 spells capable of being loaded at once and potentially used in any order, and depending on cooldown of used spells, we may or may not restrict that list for any given time t.
then multiply that 25 possible combinations with all the possible modifications to base damage, critical damage, dc, and save, metamagic, and more and you get a very very large number of permutations just for a chain of two spells, much less more.

and bear in mind all of this is still in the realm of "napkin math" as we are still not accounting for variate mob hp, resistances, overkills,player execution errors, alternate gearing levels and so much more.

we are fixed on examining an approximate minimum and maximum, and some special alternate cases, with regards to a single monster.

to build a full model would be essentially to rewrite the game over again to test with.

way to miss my point.

i know quite a lot about optimization so you don't need to point me to wikipedia or google. i was just pointing out that using very specialized terminology in this contest is a cheap way to argue.

[steelblueskies]

way to miss my point.

i know quite a lot about optimization so you don't need to point me to wikipedia or google. i was just pointing out that using very specialized terminology in this contest is a cheap way to argue.

yes in the same sense that having a nuke, sharing the plans with people using thrown rocks, and then entering a war with them would be cheap. there's the right way, then there's the long way, then there's the bickering around the point for ages. the read, research, improve or get out of the way, holds for everyone.(and sometimes they actually build their own or improve on it. humans are adaptable and we are dealing with far better then rock chucking levels of ability)

but in fairness i responded remembering that many will read this, not just you, and it *might* make the reasoning a little easier to follow going forward if they got a 60 seconds of cliff notes flyby.

given we know now that the devs are looking to our work for how things stand and should be balanced, the bickering in circles to no effect has to stop, and actually calling the calc here what it is, and presenting its flaws for improvement, while acknowledging other options to get better solutions becomes far more pressing.

after all this now becomes a fre to play extension option. spend a little time plying skills to help them balance it properly, or wind up with a mess that may cost 20$ something to reincarnate around.

from torc - "We are following the folks crunching the numbers right now with much interest." from here http://forums.ddo.com/showthread.php...49#post3728949

[sirgog]

Arraetrikos was CR46 I think. He was being nailed and still stayed up for ~3 minutes. I had curse, fully
stack Niac's, fully stacked Eladar's the whole time and emptied my mana bar with Frost Lance (SLA), Polar
Ray and Otiluke's. I was quite pleased with my DPS. I have to report a bug as he didn't stay still
the whole time (some of my spells missed) then got p*ssed and DBF'ed me in the face

That's the version of Harry from part 5 of Epic Shroud that was in the game files a while back (and has never been accessible to players). According to the datamined info that went around a while back, he has 329 resistance to all elements except electricity, a 40 Reflex save and 320k HP. I'm not surprised he took a while to zap down.

Interestingly, no Evasion and less HP than Elite, but he makes up for that with Horoth-style Disintegrate.

[Noelemahc]

Real ingame experience from tonights even shows his purposely biased projections are so off base its not even funy.

We had 60+ players all fully attacking epic raid bosses like Velah, Arraetrikos, etc.. None of them died anywhere close to 1 min. All took 2-5 min.

A quick look at there bio showed as many as 20+ DoTs on them at once! I could see no less then 15+ casters spamming spells too. Was an insane array of colorful effects

It didnt break the game.

Nor can 12 sorcs.

The projections are wrong, the numbers are wrong, the nerfs were done for the wrong reasons. Theoretical math doesnt happen in ddo. period.

Yeah I know it. Biased projections are biased. I was there for the event today as well on my wizard. Was getting some terrible fps dips though :/ I really need a new pc. At any rate, it still took us a good while to bring down the epic bosses Thoon threw out at us even with casters throwing up DoTs and the melee swinging away. Granted I wasn't on my Sorc, but still the grand majority of the dps in these "calculations" came from usage of Niac's and Eladar's, Polar Ray, and Ice Storm.

Not only was the dps not nearly as impressive as it was made out to be, it wasn't sustainable as long as the theoretical charts say either. You still blow sp on heals, other emergency buffs, cloudkills to help keep melee up, etc. During that 2 hour challenge I was in with you guys, I burned through an entire stack of store pots in addition to using the shrines Thoon threw up. Just keeping up basic stuff like Niac's, Ice Storm, and using my PM SLAs I was still running out of sp pretty fast using other spells on a practical basis.

I should have brought my Sorcerer instead of my Wizard though to be honest, but still the numbers wouldn't have been that far off anyway. Either way real testing disproves most of the theoretical math.

Shade, if you're free I'd like to get a few guys together to run some epic raids and get some realistic and practical numbers in from epic raids we have currently. Would even be awesome if we could hammer it out some time with a dev sitting in on the raid and observing actual real world testing of these changes, whether it be Thoon, Eladrin, or another dev.