DPS Calculation of 2.0 relics & C++simulation code gratis!

[oofalong]

I think you used the term "Power's effect" for power´s coefficient which is ofc a constant. Keep in mind Stats-DPS ratio also depends on specs, rotations ect. Once stacked up each ability´s Stats to DPS ratio with all kinds of exponential functions, the final DPS function will become more complicated than y = a*x+b to analyse. So stats-dps ratio is actually a very strong averaged value.

 

Holding all else constant - which is what happens when we consider the benefits of Serendipitous Assault Relics - power absolutely has a linear effect on Bonus Damage and thus DPS. Do we agree on this?

 

Further - and again holding all else constant - Primary Stat affects Bonus Damage linearly, but it also affects critical chance. However, I am specifically suggesting that the impact to critical chance is minimal. For example, my Marauder is in a mix of 72s, 75s, and 78s and his strength is 3,328. This contributes 8.63% to critical chance. An additional 225 Strength will contribute less than 1% more critical chance.

 

Let me ask again, how did you arrive at a Primary Stat to DPS factor of 0.42? And, more importantly how did you calculate that is is higher than the Power to DPS factor?

[oofalong]

I appreciate that four replies in a row is bad form, but I wanted to post something unrelated to others' comments.

 

The observed behavior for Focused Retribution boggles my mind. I admittedly didn't believe Sindariel's logs so I bought my own relic and tested it. I dumped the results in this sheet. Anyway, I too observed proc refreshes as short as 20s and durations ranging from 12s to 32s...

 

This made me question the relics which existed pre-2.4 such so I re-tested my Kell Dragon SA. Fortunately, it is behaving as I expected - which is still broken. I saw proc durations of greater than 6s and refreshes more frequent than 20s.

 

Anyway, my conclusion is that the SA relics behave as I had previously modeled them, and I have not determined how the new relics in 2.4 behave.

 

Once the new patch comes out on Tuesday, I expect to check this again.

[X-Boson]

Holding all else constant - which is what happens when we consider the benefits of Serendipitous Assault Relics - power absolutely has a linear effect on Bonus Damage and thus DPS. Do we agree on this?

 

Further - and again holding all else constant - Primary Stat affects Bonus Damage linearly, but it also affects critical chance. However, I am specifically suggesting that the impact to critical chance is minimal. For example, my Marauder is in a mix of 72s, 75s, and 78s and his strength is 3,328. This contributes 8.63% to critical chance. An additional 225 Strength will contribute less than 1% more critical chance.

 

Well, the point I made was that DPS does not only consider "the impact to critical chance" of one ability which is minimal, but the sum-up impact of all used abilities in your rotation which will be no more minimal. Ratio stats-dps can be roughly considered as derivative of DPS function with respect to stats as variables. Let me illustrate it with some equations

 

 

DPS function: [ (1+crit*surge)*(A+B+C+D+...)*power + (1+crit*surge)*(a+b+c+d+...)*pri_stats + constant_factors]/T

 

Ratio power-dps: [A+B+C+D+... + crit*surge*(A+B+C+D+...) ]/T

 

Ratio pri stats-dps: [a+b+c+d+... + crit*surge*(a+b+c+d+...) + crit_gain_from_pri_stats*surge*(a+b+c+d+...)]/T

 

 

 

Whereby A,B,...a,b... are coefficients for power or pri stat which are constant. Notice the red coloured terms are not constant, but highly non-linear. As we all know, pri stats contribute minimal on crit-chance (3rd term in 3rd equation). But once u sum up all the coefficients A,B,C... in those red coloured terms, this contribution can become quite considerable. Same as if u increase your crit, surge in second term of both last equations. Because of that, the higher the DPS, the higher the Stats-Ratio and vice versa. It doesnt mean the more abilities u use, the higher the DPS. Because the time used for your rotation T will diminish your DPS at the same time in the formulae above. Actually we are not allowed to use derivation. Because crit, surge, accuracy are not "linear" functions. Their derivations are not constant. We otherwise would have wrongly become a smaller ratio value. Therefore we must calculate ratio 1 stat to dps primitively as:

 

ratio stat-dps = DPS with (X+1) stats - DPS with X stats

 

Thats how I calculate this ratio. As mentioned before, I even have cunning = 0.51dps as Eng sniper on op dummy. Btw you are welcome to check my spreadsheets linked on my sig for more details.

 

Let me ask again, how did you arrive at a Primary Stat to DPS factor of 0.42? And, more importantly how did you calculate that is is higher than the Power to DPS factor?

 

Well, 1 power is indeed better than 1 pri stat. But each 1 "raw" pri stat u put into your gear will be converted into 1.14 due to buff&skills. Thats why 1 "raw" pri stat will always contribute more on dps than 1 power. For example, proc 225 cunnings will become proc 225*1.14 cunnings.

 

 

 

 

The observed behavior for Focused Retribution boggles my mind. I admittedly didn't believe Sindariel's logs so I bought my own relic and tested it.

 

We discovered there was no reasonable behaviour. Thats why an empirical formula was fitted on outcoming data in the end. At least that works. However as u can see, Pri stats proc is at least as good as powerproc.

Edited November 8, 2013 by X-Boson

[oofalong]

Well, the point I made was that DPS does not only consider "the impact to critical chance"

...

Thats how I calculate this ratio. As mentioned before, I even have cunning = 0.51dps as Eng sniper on op dummy. Btw you are welcome to check my spreadsheets linked on my sig for more details.

 

I'll be honest I didn't understand any of what you said. The question/concern I had tried to pose was that a Primary Stat-to-DPS ratio can be misleading as it depends on knowing the current Primary Stat in my opinion. While I too created a Power-to-DPS ratio, I created a Critical Chance-to-DPS number in order to evaluate the new Primary Stat relics.

 

My number was 1% = 16.5 DPS. Ultimately I believe this is close to your Primary Stat-to-DPS number assuming ~3,500 Primary Stat.

 

We discovered there was no reasonable behaviour. Thats why an empirical formula was fitted on outcoming data in the end. At least that works. However as u can see, Pri stats proc is at least as good as powerproc.

 

I don't understand; you state there is no reasonable - I'll say expected - behavior, but you think a formula based on empirical evidence will work? As far as I know, there are three data points - Sindariel's two logs and my one - totaling less than 20mins. This does not seem like enough data to draw conclusions.

 

This is your thread and I sense I haven't really helped you in your quest so I will bow out. Cheers!.

 

Edit: I apologize if this sounded negative. I appreciate the math you are doing here. My focus has always been much narrower and my methods seem to have worked for that purpose.

Edited November 8, 2013 by oofalong

[X-Boson]

I'll be honest I didn't understand any of what you said. The question/concern I had tried to pose was that a Primary Stat-to-DPS ratio can be misleading as it depends on knowing the current Primary Stat in my opinion. While I too created a Power-to-DPS ratio, I created a Critical Chance-to-DPS number in order to evaluate the new Primary Stat relics.

 

My number was 1% = 16.5 DPS. Ultimately I believe this is close to your Primary Stat-to-DPS number assuming ~3,500 Primary Stat.

 

Ratio Pri Stat-to-DPS depends strongly on knowing the current Primary Stat, surge rating, accuracy rating and crit rating, but not on power. Thats what I tried to tell, since I had the feeling u didnt agree on that. Because u doubted how 1pri stat = 0.42 can be such high and I told u I even have 1pri stat = 0.51. After that showed u all the math behind it.

 

Btw I dont understand your concept "to created a Critical Chance-to-DPS number in order to evaluate the new Primary Stat relics". If we say 1 pri stat = 0.51dps for example, the crit-buff should be reasonably included in that ratio.

 

I don't understand; you state there is no reasonable - I'll say expected - behavior, but you think a formula based on empirical evidence will work? As far as I know, there are three data points - Sindariel's two logs and my one - totaling less than 20mins. This does not seem like enough data to draw conclusions.

 

This is your thread and I sense I haven't really helped you in your quest so I will bow out. Cheers!.

 

The numbers u see in my latest spreadsheet were from my own test btw, so one source more. You are always welcome to provide more data logs. Im actually always happy to see someone can find out my formula doesnt work well. Just like ppl did before. Its nothing about "my thread". Sry but I dont feel like that.

 

As for "empirical evidence", most formulae in nature science are acutally pure empirical and work very well, much better than theoretical expected.

Edited November 8, 2013 by X-Boson

[oofalong]

Ratio Pri Stat-to-DPS depends strongly on knowing the current Primary Stat, surge rating, accuracy rating and crit rating, but not on power

...

Btw I dont understand your concept "to created a Critical Chance-to-DPS number in order to evaluate the new Primary Stat relics". If we say 1 pri stat = 0.51dps for example, the crit-buff should be reasonably included in that ratio.

 

I never meant to imply that Pri Stat-to-DPS required knowledge of one's Power; I am sorry if it came across that way. Still, Critical Rating, Surge and Accuracy are held constant when comparing SA and FR relics and you do not need knowledge of them to compare the relics. Generally speaking, you need:

 

1. Incremental Damage from X Power (Factor A)
   
2. Incremental Bonus Damage from x Primary Stat (Factor B) and the incremental Damage, inclusive of this higher Bonus Damage, due to a higher Crit Chance from x Pri Stat (Factor C).
   

 

We both used the same method to estimate #1. To estimate #2 you presented an aggregate number inclusive of all of Factors B & C. Whereas I kept B & C separate. I did this because I wanted to maintain a direct comparison between Factor A & B. My Factor C essentially calculates the expected Damage from X% of Critical Chance or a Crit%-to-DPS ratio.

 

Make sense?

 

Thats what I tried to tell, since I had the feeling u didnt agree on that. Because u doubted how 1pri stat = 0.42 can be such high and I told u I even have 1pri stat = 0.51. After that show u all the math behind it.

 

I was doubting that 1pri stat was greater than 1pow to the extant you listed it - 0.42 vs. 0.39. I primarily analyze Marauders/Sentinels which do not have a talent to boost Pri Stat at all. So as I said the numbers felt off to me - specifically that Pri Stat was ~10% more DPS. I was simply asking for clarification on them.

 

 

As for "empirical evidence", most formulae in nature science are acutally pure empirical and work very well, much better than expected.

 

Yeah, I understand this. My concern with it is we know behind the scenes there is actually a formula being used to drive this relic's behavior so it is not some natural phenomenon that can not be entirely explained. I am very interested in find out what it is, and I do not think <20mins of logs is enough information to figure this out. I am going to run more tests over the weekend and I will keep adding to that Google spreadsheet. Of course, we might end up redoing all of this after Tuesday's patch.

[X-Boson]

I consider this "bug" simply as a wrong code in programming, which is actually very easy to fix. I rlly cant think of anything easier than that. Basically only copy-paste-edit from programming code of a functional relic. This should only take a few minutes of a dev. Anyway im not complaining here, but only telling the truth.

 

Btw, In case u want to check my formula, the assumptions I made in pri stats relic:

 

a) with only 1 proc effect, dmg or heal, lockout time is not 40s as claimed in tooltip, but 20s

b) Mutual lockout time of dmg and heal effect is fitted at 8.1s

Edited November 13, 2013 by X-Boson

[oofalong]

I consider this "bug" simply as a wrong code in programming, which is actually very easy to fix. I rlly cant think of anything easier than that. Basically only copy-paste-edit from programming code of a functional relic. This should only take a few minutes of a dev. Anyway im not complaining here, but only telling the truth.

 

Btw, In case u want to check my formula, the assumptions I made in pri stats relic:

 

a) with only 1 proc effect, dmg or heal, lockout time is not 40s as claimed in tooltip, but 20s

b) Mutual lockout time of dmg and heal effect is fitted at 8.1s

 

I definitely agree with the first point based on tests, but I am not sure I agree with the second. I am interested in seeing more data on that point.

 

As an aside, the longest duration we have seen is ~32s. Let's assume a damaging attack proc'ed at t=0 and created a 12s buff, then a heal refreshed the buff at t=8s causing the new end time to be 20s. Next, another damaging attack would be eligible to re-proc at t=20 and if it did and refreshed the buff we would have a buff of 32s in total, right?

 

In theory a heal could cause this to go longer than 32s, but that has not been observed yet...

[leto_cleon]

I definitely agree with the first point based on tests, but I am not sure I agree with the second. I am interested in seeing more data on that point.

 

As an aside, the longest duration we have seen is ~32s. Let's assume a damaging attack proc'ed at t=0 and created a 12s buff, then a heal refreshed the buff at t=8s causing the new end time to be 20s. Next, another damaging attack would be eligible to re-proc at t=20 and if it did and refreshed the buff we would have a buff of 32s in total, right?

 

In theory a heal could cause this to go longer than 32s, but that has not been observed yet...

 

In terms of long procs longer than the stated 12s on the tooltip, the longest two are 32.6s and 34.8s (http://www.swtor.com/community/showpost.php?p=6871408&postcount=93). Both damage and heals are causing the proc to refresh.

 

Personally, I holding further parse analysis until 2.4.3 where the Ephemeral Mending, Fortunate Redoubt, Reactive Warding, Shield Amplification, Devastating Vengeance, Serendipitous Assault relic "effects no longer trigger simultaneously when the following relics equipped at the same time". Strangely enough the Focused Retribution relic (power proc) is not on the list.

[Sindariel]

Quick Update:

 

Stacking relics of the same kind (e.g. two SAs) is indeed fixed. The other "features" still exist, though.

[oofalong]

Quick Update:

 

Stacking relics of the same kind (e.g. two SAs) is indeed fixed. The other "features" still exist, though.

 

lol, I still wonder what their intent really is/was...

[KeyboardNinja]

lol, I still wonder what their intent really is/was...

 

I don't know if Bioware has any idea how to build a working relic slot. I think Dread Guard was the only tier where we haven't had something weird going on there.

 

In any case, the primary stat relic ICD issue doesn't seem anywhere near as OP as the passive healing problem. It at least makes the primary stat relic worth looking at. If they could fix the heal issue, I'd probably call it a day.

[Typeslice]

Apologies if I missed this in the thread, but why is UW SA listed as BIS for passive healing classes (over DF)?

[oofalong]

Apologies if I missed this in the thread, but why is UW SA listed as BIS for passive healing classes (over DF)?

 

Prior to today's patch, the Kell Dragon and Underworld relics would proc independently from heals and damage. This would manifest itself as buffs that lasted longer than 6s and procs more frequent than every 20s. The Dread Forged relic did not seem to exhibit this dual nature; at most it would give a 6s buff every 20s. Thus, the pre-2.4 relics provided greater benefit.

 

All of this could have changed with today's patch, but the early reports are that it has not.

[Typeslice]

Prior to today's patch, the Kell Dragon and Underworld relics would proc independently from heals and damage. This would manifest itself as buffs that lasted longer than 6s and procs more frequent than every 20s. The Dread Forged relic did not seem to exhibit this dual nature; at most it would give a 6s buff every 20s. Thus, the pre-2.4 relics provided greater benefit.

 

All of this could have changed with today's patch, but the early reports are that it has not.

 

Got it - thanks for clearing that up. Looks like it is still working as it was prior to patch. Also, Dread Forged main stat proc is also proc'ing independently for heals/dmg

[colemanron]

I don't know if Bioware has any idea how to build a working relic slot. I think Dread Guard was the only tier where we haven't had something weird going on there.

 

In any case, the primary stat relic ICD issue doesn't seem anywhere near as OP as the passive healing problem. It at least makes the primary stat relic worth looking at. If they could fix the heal issue, I'd probably call it a day.

 

We had weird stuff going on back then too. Remember internal relics using tech crit and elemental using force, while it was'nt stated in the tooltip? Relics have always been a mess, let's hope they will either fix it soon, ot at the very least get it right in future tiers.

[X-Boson]

Apologies if I missed this in the thread, but why is UW SA listed as BIS for passive healing classes (over DF)?

 

UW powerproc benefits from dmg and heal simultaneously with a mutual lockout much shorter than 20s, while DF powerproc simply doesnt(so 20s lockout after a proc is triggered either from dmg or heal).

Edited November 13, 2013 by X-Boson

[Kgaroo]

Are we including the Obroan SA (PVP) with the DF? In other words UW/KD is better than Obroan as well? (As in it doesn't proc separately with passive heals) Edited November 13, 2013 by Kgaroo

[recsa]

Prior to today's patch, the Kell Dragon and Underworld relics would proc independently from heals and damage. This would manifest itself as buffs that lasted longer than 6s and procs more frequent than every 20s. The Dread Forged relic did not seem to exhibit this dual nature; at most it would give a 6s buff every 20s. Thus, the pre-2.4 relics provided greater benefit.

 

All of this could have changed with today's patch, but the early reports are that it has not.

 

mmm if both UW and KD benefit from same "bug" why is UW listed as BiS? shouldnt KD be BiS above UW? or is there another bug involved with UW SA relic?

 

Also, for the "healing class" category, would a gunslinger with 1 point in self healing while in cover (1 tick every 3 secs) count as that? i have noticed an increased amount of procs, but im not sure if its enough to go for the "self healing" relics.

Edited November 13, 2013 by recsa

[oofalong]

mmm if both UW and KD benefit from same "bug" why is UW listed as BiS? shouldnt KD be BiS above UW? or is there another bug involved with UW SA relic?

 

I agree Kell Dragon is better than Underworld - maybe X-Boson will update his post(s).

 

Also, for the "healing class" category, would a gunslinger with 1 point in self healing while in cover (1 tick every 3 secs) count as that? i have noticed an increased amount of procs, but im not sure if its enough to go for the "self healing" relics.

 

Based on 0.33 heals/second, you should experience a 15% higher uptime for the relic than a pure DPS class. So yes you count as a "healing class" . Welcome to the club we are getting jackets made

[X-Boson]

I agree Kell Dragon is better than Underworld - maybe X-Boson will update his post(s):

 

done, thanks for the advice. I just didnt have any KD for a test.

 

Are we including the Obroan SA (PVP) with the DF? In other words UW/KD is better than Obroan as well? (As in it doesn't proc separately with passive heals)

 

Obroan SA proc as intended, just like DF SA. It seems only UW and KD are "overpowered" (Arkanian works as intended as I know).

 

Obroan pri stats relic procs as "overpowered" as DF pri stats (20s lockout instead of 40s). So take that as your 2nd BiS in PvP. Its as good as SA (at the moment), no matter if dd or healer.

Edited November 13, 2013 by X-Boson

[leto_cleon]

I was just cross-checking the procs for the Arkanian power proc and crit proc relics on my Assualt Specialist Commando. The proc rate is really consistent.

 

Log 1: Power Proc in slot 1, Crit Proc in Slot 2

 

21 Crit Procs, 28% Uptime, 21s avg btwn procs, 1st proc 12:08:01

21 Power Procs, 29% Uptimes, 21s avg btwn procs, 1st proc 12:08:02

 

Log 2: Power Proc in slot 2, Crit Proc in Slot 1

 

19 Crit Procs, 29% Uptime, 21s avg btwn procs, 1st proc 12:16:41

18 Power Procs, 29% Uptimes, 21s avg btwn procs, 1st proc 12:16:46

 

So Ark crit proc and power proc don't lock each other out, and the relic in slot 2 can proc before the relic in slot 1.

 

https://docs.google.com/spreadsheet/ccc?key=0Ap394Q8NB51gdHdhMlVScWhfR2pwd2gwYklEZkJHcXc&usp=sharing

[Melfalmaion]

As a Watchman spekked Sentinell (dot spek) would an Underworld Relic of Dark Radiance not outdps an Underworld Relic of Boundless Ages?

 

As the other relic I have an Underworld Relic of Serendipitous Assault..

Advice would be welcome

[oofalong]

As a Watchman spekked Sentinell (dot spek) would an Underworld Relic of Dark Radiance not outdps an Underworld Relic of Boundless Ages?

 

As the other relic I have an Underworld Relic of Serendipitous Assault..

Advice would be welcome

 

On the Training Dummy, the Dark Radiance will provide higher DPS. In a boss fight, the Boundless Ages may be better due to its on-demand damage. I would personally recommend the Dark Radiance as there are rare instances when the extra burst from Boundless Ages will be significant.

[Melfalmaion]

On the Training Dummy, the Dark Radiance will provide higher DPS. In a boss fight, the Boundless Ages may be better due to its on-demand damage. I would personally recommend the Dark Radiance as there are rare instances when the extra burst from Boundless Ages will be significant.

 

Ok thought so as well thanks for the help.

was just wondering if there were any other Underworld top choice for my spek than that, so wanted to ask.

 

Btw i was reading your "Annihilation | Watchman Compendium" post a little while back.. and has this been updated to latest patch or is there still some issues with the dmg and power proc relics working together so it should be SA and BA instead of SA and DR?

Or has this issue finally been resolved in latest patch and thus making the SA/DR combo better as you state?

We are still only talking Underworld relics even though I'm getting the quest mainstat proc one soon.

 

*Edit*

Ps. great work on the compendium btw

Edited November 21, 2013 by Melfalmaion