Please upgrade your browser for the best possible experience.

Chrome Firefox Internet Explorer
×

BiS Relics for Healers PvP/PvE

STAR WARS: The Old Republic > English > Classes
BiS Relics for Healers PvP/PvE

Darth_Dreselus's Avatar


Darth_Dreselus
01.15.2014 , 04:22 PM | #21
Quote: Originally Posted by cs_zoltan View Post
Did they change something? oO
Nope, by updated I pretty much just meant checked and as a result found a mistake.

Also its better than just bumping, there was already a thread asking about relics (because scrolling down 5 threads is hard)
No One Important Died Tavernus - Sorc / Taverna - Mara / Pintus - Op / Gintonica - PT
The Red Eclipse. Sweet Dreams to all my Harbinger children.

oofalong's Avatar


oofalong
01.17.2014 , 04:15 PM | #22
Quote: Originally Posted by Orderken View Post
Though it's not meant for healers, the ranking of relics is the same for healers who maintain a DOT.

It's accurate for ranking relics for DPS classes, which is what matters most. That said --

1. The spreadsheet's formula are rough approximations.
Why do you call them rough approximations? I am pretty confident in the math, and the inputs - Power-to-DPS, Crit-to-DPS, Heals/Sec and Hits/Sec - are based on Annihilation Marauder.

Quote: Originally Posted by Orderken View Post
2. The DPS increase from a "double" proccing relic will have a different average and variance based on the frequency of self-healing for each particular class and specialization.
To expand upon this, the frequency affects the uptime of the relic. Again, the numbers are specific to Annihilation Marauders, but as you can see 0.75 heals or hits per second would result in ~25% uptime, while 2.2 heals or hits per second results in ~28% - a difference of only 3%. Even the fewest hits or heals per second - ie one every other GCD - should still result in 20%+ uptime.

Quote: Originally Posted by Orderken View Post
2. If you copied the above quoted DPS from the spreadsheet, there are errors. For a DPS in BIS 78s, the claim that the average increase in DPS from substituting KD SA for DF SA is 4.48% is incredible. In contrast, for a Sorcerer|Sage healer in BIS 78s, the maximum increase in healing from substituting KD SA for DF SA has an upper bound of 1.29%, which is three-and-a-half times less.
I don't think this comment relates to my spreadsheet at all. Still, an observed 4.5% difference in DPS is not due to relics it is due to varying crit rates/rotation.

@ Darth_Dreselus - While Merc/Commandos are unable to maintain a DoT. There basic attack hits many many times - 10 for a Merc and 7 for a Mando. Thus, by simply using their basic attack on the boss once every 20s that can significantly increase the uptime on a UW/KD SA relic as well as any FR relic.
Oofalong | Oofawrong | Incendiari | Uriyah | Litlah & More
Repoofalong | Tuamara | Liolami & More
Quote: Originally Posted by EricMusco View Post
You are correct Oofalong.

Darth_Dreselus's Avatar


Darth_Dreselus
01.18.2014 , 05:48 AM | #23
Quote: Originally Posted by oofalong View Post
@ Darth_Dreselus - While Merc/Commandos are unable to maintain a DoT. There basic attack hits many many times - 10 for a Merc and 7 for a Mando. Thus, by simply using their basic attack on the boss once every 20s that can significantly increase the uptime on a UW/KD SA relic as well as any FR relic.
Are you sure (and more importantly do you have data to show) each tick has its own separate chance to proc the relic and it is not the attack as a whole? As a result it would be even better than Sage/Sorc 6 hits per GCD used on Weaken Mind/Affliction or Sco/Op 5 Hits per GCD used on Vital Shot/Corrosive Dart assuming healing builds without any boosts to the DoTs, which a Sco/Op could take. In fact by taking Open Wound/Lethal Injectors the duration and ticks of the DoT equal that of a Sage/Sorc and with 2.6 Open Wound/Lethal Injectors will surpass Sage/Sorc. Mortal Wound/Corrosive Microbes also affects this but it is lot less calculable and TBH there are only 2 spare points in the classic Sco/Op healing build.

Scoundrels' Flurry of Bolts only logs 2 hits. Not sure about Rifle Shots, my Sniper gets 2 hits logged but not sure if this applies to Operatives as well. If yes Vital Shot/Corrosive Dart is by far the better choice.
No One Important Died Tavernus - Sorc / Taverna - Mara / Pintus - Op / Gintonica - PT
The Red Eclipse. Sweet Dreams to all my Harbinger children.

oofalong's Avatar


oofalong
01.18.2014 , 10:41 AM | #24
First off I cringed when I discovered I used the wrong form of 'their' in my previous post. Based on the math and analysis I have done for Marauders, each hits or line item in the combat log has a chance to trigger the relic. I will parse on my Merc with just Rapid Shots to confirm and post the link here.
Oofalong | Oofawrong | Incendiari | Uriyah | Litlah & More
Repoofalong | Tuamara | Liolami & More
Quote: Originally Posted by EricMusco View Post
You are correct Oofalong.

Orderken's Avatar


Orderken
01.18.2014 , 05:38 PM | #25
Quote: Originally Posted by oofalong View Post
Why do you call them rough approximations?
For this phenomena, analysis of combat logs or simulations will provide more reliable results than formula.

I've referred to using formula as approximating in this case, because
  1. The formula for the uptime of procs from hits, and the formula for the uptime of procs from heals, are averages (of hits or heals per second) of averages (treating a relic's CD as divisible without remainder by hit or heals per second) of averages (of the probability distribution for time-to-proc, which has a long tail).
  2. The formula for combined uptimes for "double" proccing relics is yet another average (of the probability distribution for the amounts of overlap between procs from damage and procs from heals, which distribution has a substantial variance).

These formula are a clean summary of these several Procrustean steps. Your analysis of combat logs doesn't show that you've used the "correct" formula, but rather that using formula in this first place hasn't cast doubt on your results.

I like, use, and recommend to others your work on relics "as is". I have no reason to doubt its ranking of relics, nor do I believe that you should note in it any of what I've discussed here. As a mathematician I prefer formulaic approaches to quantitative questions, and I would use formula similar to yours if I were to choose this approach. But I wouldn't model relics with formula; I'd analyze data from combat logs or simulations, instead.
Orderken, <Hates You>, Prophecy of the Five

Guides and Model for Healing
Sorcerer / Sage / Mercenary / Commando / Operative / Scoundrel

oofalong's Avatar


oofalong
01.19.2014 , 09:15 AM | #26
Quote: Originally Posted by Orderken View Post
For this phenomena, analysis of combat logs or simulations will provide more reliable results than formula.

I've referred to using formula as approximating in this case, because
  1. The formula for the uptime of procs from hits, and the formula for the uptime of procs from heals, are averages (of hits or heals per second) of averages (treating a relic's CD as divisible without remainder by hit or heals per second) of averages (of the probability distribution for time-to-proc, which has a long tail).
  2. The formula for combined uptimes for "double" proccing relics is yet another average (of the probability distribution for the amounts of overlap between procs from damage and procs from heals, which distribution has a substantial variance).

These formula are a clean summary of these several Procrustean steps. Your analysis of combat logs doesn't show that you've used the "correct" formula, but rather that using formula in this first place hasn't cast doubt on your results.

I like, use, and recommend to others your work on relics "as is". I have no reason to doubt its ranking of relics, nor do I believe that you should note in it any of what I've discussed here. As a mathematician I prefer formulaic approaches to quantitative questions, and I would use formula similar to yours if I were to choose this approach. But I wouldn't model relics with formula; I'd analyze data from combat logs or simulations, instead.
All that makes sense - and I agree the formula is an estimate (although I wouldn't call it rough ). To me the important takeaway here is the relative rank of each relic. I don't think anyone could perfectly model the true DPS benefit of each relic as by design models make certain assumptions. Anyway, I just was sensitive to the 'rough approximation' and wondered if you felt there was something I could do to improve it.
Oofalong | Oofawrong | Incendiari | Uriyah | Litlah & More
Repoofalong | Tuamara | Liolami & More
Quote: Originally Posted by EricMusco View Post
You are correct Oofalong.

Darth_Dreselus's Avatar


Darth_Dreselus
02.04.2014 , 01:44 PM | #27
Updated for 2.6
Change log:
- Slightly changed Merc/Mando section to reflect changes to TP/KP
- Corrected links in light of mmomechanics dying

Still no data on the HS/RS for Merc/Mando
No One Important Died Tavernus - Sorc / Taverna - Mara / Pintus - Op / Gintonica - PT
The Red Eclipse. Sweet Dreams to all my Harbinger children.

oofalong's Avatar


oofalong
02.04.2014 , 02:49 PM | #28
Quote: Originally Posted by Darth_Dreselus View Post
...
- Corrected links in light of mmomechanics dying
...
Oofalong | Oofawrong | Incendiari | Uriyah | Litlah & More
Repoofalong | Tuamara | Liolami & More
Quote: Originally Posted by EricMusco View Post
You are correct Oofalong.

Orderken's Avatar


Orderken
02.06.2014 , 12:36 AM | #29
Quote: Originally Posted by oofalong View Post
I just was sensitive to the 'rough approximation'.
I shared this with you in Mumble, Oofalong, but I wanted to share it here, too, where our discussion began.

Thanks to the surprisingly thorough help files for Visual Basic that come with Excel, I simulated a "double" proccing relic. Using the same hits (2.2) and heals (0.75) per second as Oofalong's spreadsheet, 1 million simulations of a 300-second fight had an average uptime of 46.5%. This average is only 2% higher than the 45.6% uptime predicted by Oofalong's formulaic approach.

So, the uptimes calculated in Oofalong's spreadsheet are not "rough" approximates, and my concerns have been shown to be immaterial.
Orderken, <Hates You>, Prophecy of the Five

Guides and Model for Healing
Sorcerer / Sage / Mercenary / Commando / Operative / Scoundrel

JustBLaxin's Avatar


JustBLaxin
02.11.2014 , 10:25 AM | #30
So I gotta ask because I've been doing my own research and testing, how exactly is the KD SA > DF SA? They both have the exact same affects, same cool down, same proc base, and the DF SA has over 100 more power per proc. And between the two I've noticed a much better output from the DF SA than the KD SA. So I'm going to go ahead and call shenanigans on that assumption. Otherwise I do enjoy having the dual proc of SA + FR and not having a clicky. Just to put my credentials up, 55 Op Healer 10/10 in the newer HMs.