Simcraft Thread - Legion/7.x

Warning: Theorycraft inside.

Moderators: Fridmarr, Worldie, Aergis, theckhd

Simcraft Thread - Legion/7.x

Postby Fundus » Mon Aug 08, 2016 6:56 pm

Alright, so I'm not sure if there's many out there still running Simcraft for Prot Paladins, especially with Theck being (mostly?) retired, but I thought we could at least have a thread for results and discussions on Simcraft. I've been mostly running Simcraft for my own information, especially since I'm not in the beta, but I'm going to post up my results for commentary or analysis. First though, I feel I should get this disclaimer out of the way. I'm not Theck or the math and simulation wiz that he was. I'm also not a mythic raider and probably more of a "casual" than many others here. I simply find it enjoyable to play the most optimally I personally can.

In addition, I know Liminara over on mmo-champ has been working with Simcraft a good bit and has a google docs spreadsheet of talent findings, as well as writing the WoWhead guide for prot pallies. If you really want the TLDR of all of this, I would just read Lim's wowhead guide, since it seems to be very well put together and well tested. If you want lots of nitpicky numbers and bar graphs, then keep on reading.

Curent Simcraft issues:
    1) It's in BETA still and could have bugs. Best we can do is stay up on the latest nightly builds and look for anomalous data, but remember all this could change if there's a significant bug fix.
    2) The TMI Standard bosses haven't been updated for Legion. This means that there's really no standard boss for us to fight, so we have to define one as part of the sim. Therefore, results may vary based on the damage profile of the boss defined.
    3) LotP/HotP are based on percent health, unlike all of our old heals. This means we have to at least somewhat accurately model our health changes/percent and external healing over time to get an idea of how effective these spells are. No more letting the simmed pally end up at a negative one million health by the end of the sim like in 6.x. So the sims now have periodic healing events defined which have to be adjusted based on incoming damage.

These were all run on the 8/8/16 build, number 22210.
I used the default gear set, which is heroic dungeon gear, except I corrected the back enchant from Agi to Str:
Code: Select all
head=casque_of_the_deep,id=134512,bonus_id=1726
neck=chain_of_scorched_bones,id=134529,bonus_id=1726,enchant_id=5439
shoulders=frozen_pendulum_shoulderguards,id=134518,bonus_id=1726
back=shadowfeather_shawl,id=136977,bonus_id=1726,enchant_id=5434
chest=inferno_breastplate,id=134503,bonus_id=1726
wrists=detention_wristclamps,id=134522,bonus_id=1726
hands=rumblestone_gauntlets,id=137355,bonus_id=1726
waist=roaring_shellbreaker_greatbelt,id=137361,bonus_id=1726
legs=tassets_of_ravenous_euphoria,id=134516,bonus_id=1726
feet=rockbound_sabatons,id=134141,bonus_id=1726
finger1=ring_of_contempt,id=134490,bonus_id=1726,enchant_id=5427
finger2=ring_of_looming_menace,id=134533,bonus_id=1726,enchant_id=5427
trinket1=horn_of_valor,id=133642,bonus_id=1726
trinket2=infernal_alchemist_stone,id=127842,bonus_id=598
main_hand=oathseeker,id=128867,ilevel=855
off_hand=truthguard,id=128866,ilevel=855

Truthguard's traits are maxed out, without any socketted bonus traits, and with the exception of Unrelenting Light (the end stacking bonus trait): http://beta.wowdb.com/artifact-calculat ... AAAAAAAAAA
Talents, when not being tested, are the default spec from Lim's guide: http://www.wowhead.com/talent-calc/pala ... ction/M5l0
I realize the gear being in heroic dungeon gear with a maxed out Truthguard may be a bit unrealistic, but I didn't want specific Truthguard traits impacting the talent and rotation sims. I'm hoping to do Truthguard trait sims a bit later, but still deciding how to best go about it.

I defined a boss that has an auto attack for 810000 every 1.5 seconds, a spell nuke for 990000 every 24 seconds, and a DoT ticking for 44000 that's applied every 20 seconds. The boss AoE's all the tanks in the Sim (aoe_tanks=1) WITHOUT splitting the damage.
The paladins in the sim are healed for 16000 every second to simulate a HoT and 390000 every 5 seconds to simulate direct healing.

These were based on the boss in the Legion default prot paladin profile and I simply adjusted the values to produce similar TMI and damage profile to the WoD sims I have still laying around. I have no logs from beta, so I'm not sure if my boss damage and/or healing values are anything like a real encounter. For reference, spell damage is approx. 21% of this boss's damage and auto-attacks make up the other approx. 79%. I also changed the default APL to cast LotP/HotP at a higher percent health (65%) while AW is active, keeping it 35% the rest of the time. Since I built my APL, the default has actually been changed to be more proactive with LotP/HotP based on incoming damage, so if I go back and include those changes, the amount of healing for LotP/HotP appears to go up even more. I can re-run with that later if needed.

Rotations
I ran several sims with a couple ability-rotations each to build up to this one. It doesn't have every single possible ability-rotation permutation, which seemed excessive, but I tried to include all of the ones that looked interesting.
Image
Full Results(you'll likely have to download it to view it)

A couple interesting things I notice:
1) Prioritizing Judgement appears to be the best in TMI, DTPS, and DPS. The bigger difference is whether BH or Cons/AS are next on the priority. J>Cons>AS>BH is a little over 3k DPS over J>BH>Cons>AS, and is only 660 TMI behind/worse. That seems like a reasonable trade-off to me. Thoughts?
2) Don't prioritize BH over Judgment. I don't think anyone was recommending this, but it's horrible for DPS and horrible for TMI and DTPS as well.
3) Cons>AS always wins for DPS by 1.2k to 1.5k and gives up less than 100 points on TMI. Again, possibly a reasonable trade-off?

Though it makes it more complicated, going forward I'm going to keep the 2 highest DPS and 2 lowest (best) TMI rotations. I'm dropping the 'J>' from the front of each one and the '>BL>HotR' from each one, since it's the same for all 4. That gives me the room to add the talent nomenclature to each one. So without further ado on to the Talents...
Last edited by Fundus on Mon Aug 08, 2016 8:10 pm, edited 1 time in total.
Fundus
 
Posts: 33
Joined: Mon Sep 29, 2014 10:19 pm

Re: Simcraft Thread - Legion/7.x

Postby Fundus » Mon Aug 08, 2016 6:57 pm

Talents
Gear, Artifact traits, Boss damage, and healing setup all remain the same from above. Unless otherwise stated, talents are still the default linked in the previous post.

First, we'll address Blessed Hammer (BH) vs Holy Shield (HS) vs Consecrated Hammer (CH). Why even bother with CH when the Sim isn't set for movement? Essentially I just wanted to be thorough. Also note that there's 4 ability-rotations for the BH ques but only 2 for the HS and CH ques, since the others would be redundant.
Image
Full Results
Interesting things:
1) Maybe a surprise to no one, but BH wins for DPS, DTPS, and TMI.
2) CH loses significantly. It's over 5.2k TMI worse than HS, which is an additional about 1k TMI worse than BH. The fight will need to have a lot of movement to make up for that amount of loss.

On the list of things to do, quantify exactly how much movement and time spent outside Cons it would take to level the field between CH and BH. Also on the list, how much magic damage does it take to bring HS equal to BH. More than 21% of the boss's damage is all I can say so far.

_____________________________
Second, looking at Bastion of Light (BoL) vs First Avenger (FA) vs Crusader's Judgment (CJ)
Image
Full Results
Interesting things:
1) As Lim posted on his spreadsheet, FA is a noticeable (3.6k-4.8k) increase in DPS over BoL on single target, especially in the ques that prioritize AS higher. It also produces a very slight increase in TMI benefit to putting AS above Cons in the que.
2) BoL still wins at DTPS and TMI (by around 2k-2.5k).
3) CJ produces slightly more HPS through more Grand Crusader procs, but otherwise is the worst choice in DPS and TMI.

_____________________________
Third, we're skipping ahead to the next interesting talents, the level 75 tier - Hand of the Protector (HotP) vs Knight Templar (KT) vs Final Stand (FS) - and including the new Cavalier (+Cav) from the level 60 tier only when KT is talented. Sorry this is a large image...
Image
Full Results
Intesting things:
1) Cavalier makes very little difference in TMI versus just KT alone. In fact it's so small it's within the margin of error and sometimes comes out worse than KT alone. Perhaps with very intelligent use of KT, Cav could be a benefit for survivability, but not by very much.
2) HotP unsurprisingly wins for TMI and HPS, though only by a couple hundred points over KT/KT+Cav. Perhaps more interestingly HotP also wins for DPS by 1.3k to 3k, probably due to being off the GCD and Divine Steed (DS) not. The sim uses DS during times of high damage ahead of the regular rotation, pushing back all of the abilities except SotR and reducing DPS. Perhaps more ammo for the suggestion that DS should be off the GCD, at least with KT talented?
3) FS isn't terrible for TMI or DPS, and wins at DTPS. In reality all of these are close enough for survival that encounter specifics will probably matter more.

_____________________________
Fourth, we're comparing Aegis of Light (AoL) vs Judgment of Light (JoL) vs Consecrated Ground (CG). Again, some may ask why even include Aegis of Light, but I'm curious how much personal survival we're giving up for an added raid cooldown.
Image
Full Results
So this is where things get really unexpected. CG is winning for TMI over JoL and AoL, and by about 2k versus JoL. In fact even AoL is beating JoL for TMI and HPS. Why is JoL underperforming? I thought perhaps the "raid healing" from the other paladins was overlapping and causing a problem, so I limited the sim down to only 3 paladins, one with each talent, each attacking their own boss. From the notes in the Simcraft file it appears AoL at least only applies to the paladin casting it, so we can even eliminate that paladin. I also guessed though that perhaps the artifact trait Consecration in Flame (CiF) was allowing more Cons up time and benefiting CG, so I added back in a paladin with CG but without CiF.
Image
Full Results
Even with 3 separate bosses, CG is still beating JoL. As far as I can tell that means that JoL isn't affecting the CG or CG+CiF paladins, but CG could still be healing the JoL paladin. Still, even without CiF, CG still beats JoL for TMI in this Sim.
1) Digging into the data, due to haste, CG is healing our paladin every 0.61 seconds without CiF and every 0.46 seconds with CiF. JoL by comparison is only healing our paladin every 1.32 seconds, probably due to the 1 second ICD. That contributes to the difference we're seeing. Still it appears that JoL is healing for a little over twice (5800 vs 2100) what CG is doing. So napkin math would have made me expect JoL to be almost identical or a little ahead of CG at least without CiF. Any ideas?
2) Why is CiF decreasing the average time between heals? Well, CiF allows Cons to overlap itself, so the sim is also giving heals from both Cons. I've read confirmation that Cons can overlap on beta, but I'd love to know from someone on beta if the heal overlaps as well?
3) The default gear has less than 26% haste. So since CG appears to scale with haste while JoL doesn't, at what haste point does CG start passing JoL?
4) Unrelated to the Sim, a question I haven't been able to answer. In a multiple boss/mob scenario, can you benefit from 2 different JoLs on 2 different mobs due to AoE abilities?

Granted JoL may still be the better heal for the raid since it's 40 charges will be much harder to cap than the 6 player limit on CG. And CG makes us even more dependent on not moving out of Cons, which may defeat it's benefit on any fight with movement.

_____________________________
Moving on, finally, we have Righteous Protector (RP) vs Seraphim (Sera) vs Last Defender (LD).
Image
Full Results
Interesting Things:
1) Sera wins DPS and loses TMI as expected, but isn't as bad for survival as I would have guessed. The top DPS que wins by over 15k dps vs RP, the current default, and only gives up about 5.7k TMI. For a fight where tank survival is less of a factor than DPS, that's a pretty easy choice.
2) LD actually beats RP across the board on this single target boss fight. It's typically over 1.5k DPS ahead and still 1k TMI better. Now in part this will depend on how often you're needing to cast LotP/HotP. If you're casting it on CD, I would expect RP to pull ahead, at least in TMI. This is where I need to test the new default APL that uses LotP/HotP more to see if we get a reversal. If not, RD may be a better default talent even for single target.

That's all the data I've got so far. I've reserved 1 more post below to try to follow-up on questions that I've raised in this post. If you have questions, comments, or find errors, please let me know.
*Edit: Corrected a couple typos. Also I forgot to link to the full results for each Sim, so I added those. Again, they're on Google drive, so you'll probably have to download them to view them properly. If anyone would prefer not to download and wants screenshots of specific parts, I could post those up as well.
Last edited by Fundus on Tue Aug 09, 2016 2:22 pm, edited 2 times in total.
Fundus
 
Posts: 33
Joined: Mon Sep 29, 2014 10:19 pm

Re: Simcraft Thread - Legion/7.x

Postby Fundus » Mon Aug 08, 2016 6:58 pm

Talents 2
Reserved...Post Under Construction...
Fundus
 
Posts: 33
Joined: Mon Sep 29, 2014 10:19 pm

Re: Simcraft Thread - Legion/7.x

Postby Dion » Wed Aug 10, 2016 10:11 am

JoL vs CG was extremely intresting, and raises intresting questions. I wouldn't thought CG was that good, even just by ourselves sitting on it.

But I have another question. Why are you simming Lotp and Hotp usage at 35%?
Dion
 
Posts: 264
Joined: Tue Jul 28, 2009 11:23 pm

Re: Simcraft Thread - Legion/7.x

Postby Fundus » Wed Aug 10, 2016 2:09 pm

But I have another question. Why are you simming Lotp and Hotp usage at 35%?

Good question. That was from the original default Action Priority List (APL) and I had just left it alone. It doesn't say who wrote original default APL in the change notes (though it was recently updated by "bitfluxhound"), but it's possible it was written or influenced by Liminara, since that 35% value was also mentioned over in the mmo-champ prot thread here. I've been reading that entire discussion, and saw all the discussion about how to properly use LotP/HotP. Assuming you're the same Dion from that thread, I tend to agree that more frequent use, especially of HotP with RP, might make more sense. I had actually tried moving it up to 70% health, similar to what Lim mentioned in that post, but it did result in higher/worse TMI, despite having increased HPS, so I had left it alone. I didn't save the results from those runs unfortunately. That was also before I added the "<65% during AW" LotP/HotP use line and on a previous version of the Sim before some significant bug fixes (like SotR was still only reducing physical damage until about a week ago).

Since I was already setting up another run to test the new 'smart' HotP/LotP use in the new default APL, I'll compare the current "<35% and <65% during AW implementation" vs "<65% and <75% during AW" vs the new default 'smart' use that's <40% always and then tiered <55% and <65% depending on incoming damage. Otherwise it's the same gear and "default" talents as above, but the boss damage in increased slightly across the board to stop TMI from getting too low.
Image
Full Results
Ok, so the new "smart" behavior is certainly beating my previous implementation on HPS and TMI, but the 65% is actually now winning on TMI by a little more (about 1k) over either of the other implementations. Partly that's because the boss, even with increased damage, is rarely getting the paladins in the sim below 40%. So that makes me think we really could sim using HotP at various (maybe every 5% or 10%?) health levels to see if there's an optimal health point to cast it. My only hesitation there is: 1) It's a lot of sims to run and 2) The results will vary depending on the amount of incoming damage. I'll put it down as something to run when I can, but I'm still hand coding all of these, so it'll take some time to get to.

Also, those results are all using HotP and RP. Let's see what happens to each one if we change to KT and LD for example.
Image
Full Results
Much more mixed and much closer together. Technically the 'smart' LotP behavior is now winning for TMI, though the differences are essentially all within the margin of error of the sim. The total difference from best to worst for the whole sim is only about 500 TMI. So considering different talent combinations that we might be testing for future sims, I think I'm going to start using the 'smart' behavior as my default for now. It's certainly better than "<35% health" and allows the sim to better adjust for the differing CDs of LotP/HotP more similarly to a player. It's also probably the closest we can get to a player who will use LotP/HotP when low on health or sooner when in heavy damage to ensure survival until the next external heal. Perhaps at some point we can switch to a more straightforward "always <X% health" use, but I think it will need more simming to decide how dependent that 'X' is on incoming damage.
Fundus
 
Posts: 33
Joined: Mon Sep 29, 2014 10:19 pm

Re: Simcraft Thread - Legion/7.x

Postby Decadere » Sun Aug 14, 2016 6:58 am

Thank you for the informative post. This is precisely what I need to know going into Legion.
Decadere
 
Posts: 22
Joined: Sun Jan 10, 2010 7:42 am

Re: Simcraft Thread - Legion/7.x

Postby Dion » Sun Aug 14, 2016 9:25 am

This is caused
Fundus wrote:I had actually tried moving it up to 70% health, similar to what Lim mentioned in that post, but it did result in higher/worse TMI, despite having increased HPS, so I had left it alone.

By this
Fundus wrote:I defined a boss that has an auto attack for 810000 every 1.5 seconds, a spell nuke for 990000 every 24 seconds, and a DoT ticking for 44000 that's applied every 20 seconds. The boss AoE's all the tanks in the Sim (aoe_tanks=1) WITHOUT splitting the damage.
The paladins in the sim are healed for 16000 every second to simulate a HoT and 390000 every 5 seconds to simulate direct healing.

And this is problematic also for Lotp/Hotp simulation.
Fundus wrote:Partly that's because the boss, even with increased damage, is rarely getting the paladins in the sim below 40%.

Why this is problematic with frequent usage of Lotp/Hotp? Because if lowest point of players HP pool is 40-35% then be all definations it should be used then. Because you want to use it as low health as possible and as often as possible. As well as after nuke but I get to that nuke part little later and why I still would advice to use it at cooldown instead of any % guideline. If I advice Lotp/Hotp usage on cooldown, in real raid encounter there is chance that it comes from cooldown at low health. But if sim health doesn't often drop below <40%, that means there won't be any those <20% casts which are possible in raid encounter

Why it should be used at 35% or lower? (Best would be at 1%) Because it saves your ass most likely at that moment by very big heal. Heal is extremely pleasent thing at low HP and postpones extremely efficiently death as 100 health recieved is not 100 damage need to dealt for death. And it absolutely sucks to be in spot with low HP and Lotp/Hotp on cooldown.

Why then Simulationcraft and TMI thinks cast only at <35% is worse tactic than <40% always and then tiered <55% and <65% depending on incoming damage? It's smaller heal and weren't you more likely to die at lower health than higher? One option is that drop to <35% is too rare so it's better to cast <40% than not at all. But this is unlikely. [In next guess I walk by very thin ice and might be totaly wrong. You have been warned.] This might be by nature of TMI, it is direct measurement of spike damage and attempt to "smooth" damage spikes. And it values bigger spikes more than smaller spikes. I might be wrong but for example spike that deals 130% of your health is 3 times more significant than spike that deals 120% of your health. And 120% is 3 times significant than spike that deals 110%. But how is Lotp/Hotp smoothing incoming damage? Well, simulation already considers big incoming hit more valuable than small, therefore I guess it values healing before Big Enough Hit more beneficial than big heal after big hit. Why is this? This is only my guess but more health leads better survival of said spike damage.

So, in light of these sims and theories, am I switching to cast Lotp/Hotp what is being called intellectual way? No. I would argue that casting it on cooldown is better choise. There are many reasons, as it's not simple matter.

First things first. If you happen to tie Lotp/Hotp usage before X amount of damage, you are demanding alot of data processing before cast. Internal absorbs? Internal damage reductions? Internal health gains? External ones? That's lot of variables to consider to cast which has cooldown 15sec max.

Now, Simulationcrafts standard TMI boss encounter is purely fictional and it heavily emphasises dangerous spike damage because that's what TMI tries to remedy. But if we look actual encounters like Mythic Archimonde, threat to die doesn't come from single spike but extended heavy damage phases lasting variety of times. For example, Mythic Archimondes Death Brand lasts awhile instead of single spiky attack.

It's important also note that when you cast Lotp/Hotp on cooldown, it's true that it might not be castable at 35%-40% situation. But as logs have shown, if you cast it on 35% it's might as well be on cooldown for next 35% situation. Infact, it's pretty common in logs to find out that once Lotp is cast and 7-10 seconds later there is Lay on Hands cast. On similar note, if cast on cooldown at 65% and at 20%, it's better than planned cast at 35%. Prediction of future and what-if games are messy business.

"But <40% always and then tiered <55% and <65% depending on incoming damage -rule works during long high damage phases too!" Yes, true. But also false. Because, that high damage phase lasts only fraction of whole encounter, trying to pick most optimal spot could easily shave casts away. There is extremely little room to pause and try to cast it optional way. If you cast it on cooldown, you end up likely more not-so-optional casts but get more casts. (Still, you could get optional casts too, predicting future is extremely tricky)

One of arguments I have heard is that HPS is somewhat meaningless thing. Padding the meters when it doesn't matter. This is rooted on view that for example boss auto-attack damage while 80% is less dangerous than boss auto-attack damage 35%. (I don't think TMI and Simulationcraft take health levels into account anyway, or those -1M health simulations would have been seriously distorted.) This is peculiar thought as nobody questions value of Blessed Hammer while it reduces auto-attack damage taken while 80% health, damage that was neemed not so worrysome. On the contrary, wowhead&icy-veins guides recommend trying to get high as possible uptime on debuff. So, funnily it's valuable mitigate damage but to heal from said damage is considered waste of cooldown or just padding the meters.

Then there is human nature. While in theoretical simulations show it would be optional to cast it under x,y,z conditions while juggling midgets juggling chainsaws, there is extremely much things going on simultaniously while tanking and trying constantly finding best spot for Lotp start to cost dearly. On hunch, missed ShotR and unintentional gaps very quickly add up. Not all encounters are Patchwerk/Butcher fights and I have already heard complaints that Tanks currently had more than their fair share on their plates in WoD.

There is also healers that really, really hate tanks dropping low health and start to freak out. That is situations where "Oh,fuck don't die!" buttons are pressed. And while I haven't really had this discussion with healers yet, I'm wondering how easy is it to convince them it's really good idea not to heal even when heal has pretty short cooldown.

And this wasn't aimed at Fundus, keep sims coming. I can speak only for myself, but I have many questions that are not answered yet. Is there encounter where Holy Shield is better? How much magic damage it's required before Holy Shield pulls ahead? When Consecrated Hammer pulls ahead of Blessed Hammer, just to name a few.
Worthy sidenote: Blizzards design philosophy has been at least lately that passive is worse than active ability. So if Last Defender is too close to Righteous Protector, I expect either nerfs or buffs.
Dion
 
Posts: 264
Joined: Tue Jul 28, 2009 11:23 pm

Re: Simcraft Thread - Legion/7.x

Postby Fundus » Sun Aug 14, 2016 10:26 pm

Thanks for more replies and interest. I'm doing my best to keep up with questions and still work on new data.

So first, it's come to my attention that not everyone looking at this may be familiar with TMI. In the briefest of terms, TMI is a metric defining how smooth a tank is taking damage over a predefined amount of time, typically 6 seconds. This means it's measuring both the size and frequency of "spike" damage taken during that 6 seconds. It's scaled such that a TMI of 88k is equivalent to taking a sinlge spike of 88% of your health in only 6 seconds, or possibly several smaller spikes around 80% of your health during the same 6 seconds. As an example of why both size and frequency of spikes might be important:
Dion wrote:But if we look actual encounters like Mythic Archimonde, threat to die doesn't come from single spike but extended heavy damage phases lasting variety of times. For example, Mythic Archimondes Death Brand lasts awhile instead of single spiky attack.

As I mentioned before, I'm not a hardcore player and I've never set foot in a mythic raid, so I can't speak to this or other mythic encounters from personal experience. However, the general reason for developing the TMI metric was that Theck and other theorycrafters at the time felt that most preventable tank deaths, meaning preventable mostly by the tank, came in fairly rapid spikes. The slow decay deaths generally meant a healer or healers were dead/asleep/out-of-range, because not receiving any heals for over 6 seconds on something like Mythic Atchimonde shouldn't happen. It's a long read, but if you want to know/recall the full thoughts and math behind it all, including the 6 second window, I recommend Theck's original TMI post, the 1st new TMI post, and the 2nd new TMI post.

On to other questions...
Dion wrote:
Fundus wrote:Partly that's because the boss, even with increased damage, is rarely getting the paladins in the sim below 40%.

And this is problematic also for Lotp/Hotp simulation.

I agree that not hitting hard enough may distort the sim, but it becomes difficult to hit our paladins in the sim too hard without killing them or hugely increasing the healing we're giving them, which also seems unrealistic. So I agree, it's going to be a challenge to accurately sim LotP/HotP. That doesn't mean though that we can't come to at least some informative conclusions. For example...
Dion wrote:If you cast it on cooldown, you end up likely more not-so-optional casts but get more casts.

Just to demonstrate, I'm adding an "on cooldown" que, instead of the "<35%" que, to that 2nd to last sim (using HotP and RP again for shortest CD), where LotP/HotP is simply cast on cooldown as you mention. I'm going to do 2 runs of it with different levels of damage. Increasing damage even more for the high damage sim, I defined a boss that has an auto attack for 830000 every 1.5 seconds, a spell nuke for 1300000 every 14 seconds, and a DoT ticking for 48000 that's applied every 20 seconds. That's moderately more auto-attack and DoT damage and much large and more frequent nukes. Ideally that should give bigger spike and get our paladins to lower life numbers more often.
Image
Full Results
Ok, so we can see that the "on cooldown" (_onCD) ques do very poorly. They're worse on both TMI and HPS than either the "smart" heal ques or the "<65%" ques, usually by upwards of 4k TMI. This is why you don't want to literally use LotP/HotP on cooldown or macro it into another ability. It also makes sense. Simply casting on cooldown will have you healing yourself may times when you're full or nearly full health.
And just in case anyone doesn't want to download the full results to see that tank's health profile, here's the BEST TMI tank's health over the course of the fight vs the same ability que but casting HotP on CD:
Best:
Image
On Cooldown:
Image
While both have a similar damage profile, since the boss is hitting both exactly the same, I think it's pretty visible how much closer to dying (hitting 0M health) the "on cooldown" version got. That's about the clearest demonstration of "4.5k TMI better" that I can come up with.

For the low damage sim, I defined a boss that has an auto attack for 750000 every 1.5 seconds, a spell nuke for 750000 every 24 seconds, and a DoT ticking for 30000 that's applied every 20 seconds. Let's see what happens now.
Image
Full Results
The "on cooldown" (_onCD) ques win the HPS chart, but still lose pretty significantly (around 3k) on TMI. This is a more extreme version of what I was describing seeing before:
Fundus wrote:I had actually tried moving it (HotP) up to 70% health, similar to what Lim mentioned in that post, but it did result in higher/worse TMI, despite having increased HPS...

THIS is what people are talking about when they mention padding the meters. On the meters it looks like you're doing better by healing more, but in reality you're actually increasing your chance of dying by making the damage you take more spiky. Again, this is why I would strongly recommend against literally casting LotP/HotP on cooldown.

One other thing you mentioned that I wanted to address:
Dion wrote:This is peculiar thought as nobody questions value of Blessed Hammer while it reduces auto-attack damage taken while 80% health, damage that was neemed not so worrysome. On the contrary, wowhead&icy-veins guides recommend trying to get high as possible uptime on debuff. So, funnily it's valuable mitigate damage but to heal from said damage is considered waste of cooldown or just padding the meters.

I know this wasn't really a question, and I don't don't want to nitpick, but perhaps offer an explanation/counterpoint. The reason for this difference in thinking is that you're comparing 2 different things. Blessed Hammer by it's nature reduces DTPS; as you said it's mitigation. Using LotP/HotP, no matter how we do it, doesn't reduce DTPS and isn't mitigation. So we try to have ideal uptimes on BH because we care about reducing DTPS, while deciding when to cast LotP/HotP can only effect TMI. We care about both metrics (DTPS and TMI), but they matter in different ways, hence the different thinking.

Ok, since I've already got a bunch of stuff set up on HotP, I'm going to run a true "% health to cast HotP" test. Despite using the "smart" que, I agree that the "smart" behavior is a little hard to explain to a beginning Protadin and having a % health guildeline might be easier at least for starters. Since this is a lot of ques, I'm going to let it go over night and hopefully I'll get results up tomorrow or Tuesday if everything works.

Dion wrote:Is there encounter where Holy Shield is better? How much magic damage it's required before Holy Shield pulls ahead? When Consecrated Hammer pulls ahead of Blessed Hammer(?)

These are on my list to test as well. I already have a plan to test HS vs BH, but it takes coding and testing a bunch of different bosses with different damage profiles. For the third question, in theory the sim can also test movement, though I've never used those commands before. I may not get to this one until we see what types of fights are coming in the Legion raids. If anyone else has more experience with that portion of the sim, feel free to give it a go.
Fundus
 
Posts: 33
Joined: Mon Sep 29, 2014 10:19 pm

Re: Simcraft Thread - Legion/7.x

Postby Fundus » Tue Aug 16, 2016 6:56 pm

Alright, so I just got everything run for the HotP % health sims for 5% all the way to 100%...and then I saw the hotfixes for today. In case anyone didn't see:
Blizzard wrote:Paladin
Light of the Protector and Hand of the Protector now heal for 35% of missing health (was 25%).
Bulwark of Order now grants an absorb for 40% of the damage done (was 20%).

That could certainly have an effect on these and previous results. So I think I'm just going to wait for Simcraft to be updated again (probably tomorrow) and re-run these...and maybe a bunch of the above runs if I can find the time.
Fundus
 
Posts: 33
Joined: Mon Sep 29, 2014 10:19 pm

Re: Simcraft Thread - Legion/7.x

Postby Piffzy » Fri Aug 19, 2016 4:49 am

I dont want to take up space but I want to mention that I love and support what you are doing! keep it up if you want to! ;)
Piffzy
 
Posts: 2
Joined: Thu Jul 28, 2016 7:10 pm

Re: Simcraft Thread - Legion/7.x

Postby Ayama » Fri Aug 19, 2016 9:13 am

Yeah keep it up ! Great work thanks !
Ayama
 
Posts: 1
Joined: Fri Aug 19, 2016 9:11 am

Re: Simcraft Thread - Legion/7.x

Postby Dion » Fri Aug 19, 2016 10:20 pm

Fundus wrote:Thanks for more replies and interest. I'm doing my best to keep up with questions and still work on new data.

No, I thank you. This is kind of conversation what I was looking for. And there are always more questions when one is answered.
Fundus wrote:As I mentioned before, I'm not a hardcore player and I've never set foot in a mythic raid, so I can't speak to this or other mythic encounters from personal experience.

I finished Legendary ring last sunday, I haven't properly raided since few BRF raids. Currently with sons sleeping what it is I can be either decent father or decent main tank. Can't have both.

We are in world of hurt if cancer patients can be treated only by doctors having cancer themselves. In other words, I believe firmly that "you don't run X, your opinion don't matter!" -argument is bullshit.
Fundus wrote:I know this wasn't really a question, and I don't don't want to nitpick, but perhaps offer an explanation/counterpoint. The reason for this difference in thinking is that you're comparing 2 different things. Blessed Hammer by it's nature reduces DTPS; as you said it's mitigation. Using LotP/HotP, no matter how we do it, doesn't reduce DTPS and isn't mitigation. So we try to have ideal uptimes on BH because we care about reducing DTPS, while deciding when to cast LotP/HotP can only effect TMI. We care about both metrics (DTPS and TMI), but they matter in different ways, hence the different thinking.

On more theoretical level, HPS is important and should be looked at also. DTPS doesn't care if you took 10k hit at 100% and lived or took 10k hit and died. TMI doesn't really value 10k hits when there are 100k hits even though it was 10k hit killed you. HPS doesn't care did you cast 1m self-heal at 10% or at 90%.

DTPS and HPS is very similiar actually. Neither metrics care about your current HP situation. And neither one can't exist without one other: to be dealt damage, you must have health. To be healed, you must take damage.

Imaginary example: Boss deals 10% of your max hp damage per second. But as you heal 11% of your max hp per second you will solo that encounter.

HPS nullifies DTPS effects. It doesn't make damage smaller, but it counters it. Say, you have 200k HP and cast mitigation 10% effect before 100k hit and arrive at 90k damage dealt. You are at 110k health. Or you cast 10k heal before or after hit and arrive at 110k health. Same destination, different route.

To maximize HPS and to minimize TMI and DTPS is very strong goal in my mind and in my mind leads to stronger survival.
Fundus wrote:However, the general reason for developing the TMI metric was that Theck and other theorycrafters at the time felt that most preventable tank deaths, meaning preventable mostly by the tank, came in fairly rapid spikes. The slow decay deaths generally meant a healer or healers were dead/asleep/out-of-range, because not receiving any heals for over 6 seconds on something like Mythic Atchimonde shouldn't happen.

Mythic Archimonde logs:
Gotowned
Liminara (OT)
Rëlëdzs
It's not that there isn't heals going on, it's that they aren't enough. For record, Mythic Archimonde has been regulary run less healers to bring more DPS. So, Tanks are getting healed but it's not enough and their health is dropping despite healers healing as much as they can.
Fundus wrote:Simply casting on cooldown will have you healing yourself may times when you're full or nearly full health.

Casting it always near or at full health is wrong way to go with that ability. If this proves to be case in actual real raids, I'll change my opinion.
Let's look logs.
Gotowned
Gotowned log, using on CD: First melee hit at 0:10, cast Lotp. 0:25, 0:40, 0:55, 1:10, 1:25, 1:40, 1:55, 2:10, 2:25, 2:40, 2:55.

Gotowned self casts:
Lay on Hands: 1:42
Lotp: 0:35, 0:42, 0:54, 1:04, 1:28, 1:38, 1:48, 1:58, 2:11, 2:24

Firts we notice that my rigid 15sec cooldown is only one cast ahead as he casts often. But his cast and my rigid casts pretty close each other. So, next we can look at damage graph.

On CD casts:
0:25 Not much damage.(~30k)
0:40 Spiked damage. (~70k)
0:55 Not much damage.(~30k) (Huge damage spike ~140k, starting at 0:56 and peaking at 1:02)
1:10 Downward damage curve (~50k)
1:25 Damage ramping up (~45k)
1:40 Big damage spike (~125k)
1:55 Downwards on damage curve (~75k)
2:10 Damage spike (~75k)
2:25 Damage spike (~75k)
2:40 Low damage (~25k)
2:55 Downward damage curve (~25k)

Actual casts:
0:35 Damage spike (~70k)
0:42 Downwards damage curve (~50k)
0:54 Not much damage (~30k)
1:04 Huge damage spike (~140k)
1:28 Damage ramping up (~50k)
1:38 Big damage spike (~125k)
[Lay on hands: 1:42] Second big damage spike(~120k)
1:48 Damage spike (~100k)
1:58 Not much damage (~25k)
2:11 Damage spike (~75k)
2:24 Damage spike (~75)

Looking at actual casts and my shadow on CD casts, I must say I think on CD lost this comparison. But I must say that in this log Lotp was used extremely often not delayed much.

But since argument was that on CD casts are inefficient, let's look at HP situations of those casts.

On CD casts:
0:25 Not much damage.(~30k)
HP:97%
0:40 Spiked damage. (~70k)
HP:66%
0:55 Not much damage.(~30k) (Huge damage spike ~140k, starting at 0:56 and peaking at 1:02)
HP:100%
1:10 Downward damage curve (~50k)
100%
1:25 Damage ramping up (~45k)
HP:78%
1:40 Big damage spike (~125k)
HP:81%
1:55 Downwards on damage curve (~75k)
HP:72,5%
2:10 Damage spike (~75k)
HP:49%
2:25 Damage spike (~75k)
HP:44%
2:40 Low damage (~25k)
HP:96%
2:55 Downward damage curve (~25k)
90%

Actual casts:
0:35 Damage spike (~70k)
HP:79%
0:42 Downwards damage curve (~50k)
HP:55%
0:54 Not much damage (~30k)
HP:94%
1:04 Huge damage spike (~140k)
HP:60%
1:28 Damage ramping up (~50k)
HP:80%
1:38 Big damage spike (~125k)
HP:86%
[Lay on hands: 1:42] Second big damage spike(~120k)
HP:14%
1:48 Damage spike (~100k)
HP:85%
(1:54 HP dropped to 6%)
1:58 Not much damage (~25k)
HP:59%
2:11 Damage spike (~75k)
HP:62%
2:24 Damage spike (~75)
HP:90%

Frequent casts at high health on CD casts and actual casts of player.
Rëlëdzs

Casts by Rëlëdzs:
0:42 Lotp Damage spike (~80k) HP:64%
1:00 Lay on Hands Upwards damage curve (~75k) HP:33%
1:07 Versatility potion+healing tonic and
1:07 Lotp Huge damage spike (~160k) HP:24%
3:24 Lotp Low damage spike (~25k) HP:82%

On CD casts would be:
0:19 Melee hit so Lotp cast. Steady damage (~50k) HP:92%
0:34 Lotp Steady damage (~75k) HP:98%
0:49 Lotp Steady damage (~80k) HP:89%
1:04 Lotp Huge damage spike (~160k) HP:94,5%
1:19 Lotp No damage HP:88%
1:34 Lotp No damage HP:100%
1:49 Lotp No damage HP:100%
2:04 Lotp Small damage (~30k) HP:97%
2:19 Lotp Damage spike (~75k) HP:93%
2:34 Lotp No damage HP:100%
2:49 Lotp No damage HP:100%
3:04 Lotp Small damage (~20k) HP:91%
3:19 Lotp Damage spike (~50k) HP:59%

On CD cast lost. Many casts but high health. This is ineffective use of ability, just as Fundus sims have shown.
0:31 Lotp. Huge damage spike (~150k) HP:30%

My on CD casts
0:00 Melee hit so Lotp cast. Small damage (~15k) HP:100%
0:15 Lotp Small damage (~25k) HP:100%
0:30 Lotp Huge damage spike (~125k) HP:22%
0:45 Lotp Small damage (~15k) HP:100%
1:00 Lotp Damage spike (~50k) HP:95%
1:15 Lotp No damage HP:100%
1:30 Lotp No damage HP:100%
1:45 Lotp Big damage spike(~100k) HP:76%
2:00 Lotp No damage HP:100%
2:15 Lotp No damage HP:100%
2:30 Lotp Small damage HP:100%
2:45 Lotp No damage HP:100%
3:00 Lotp Damage spike (~55k) HP:100%
3:15 Lotp Damage spike (~55k) HP:81%
3:30 Lotp No damage HP:100

On CD casts managed to one bigger and one smaller heal compared to planned use at 30%, 22% and 76%.

Why on CD fared so badly? There are many reasons, and one significant reason is this.

When Tank, may it be OT sits over 98% of encounter over 50% health, it's extremely bad situation for Lotp on CD usage. If only 2 cast on encounter is meaningful, you should save Lotp to those situations.

So lights out on Lotp/Hotp on CD usage? Well.. since everybody likes logs, I have one more.
But let's look at that final log.
Morally

Morallys casts
0:33 Lotp Damage spike (~80k) HP:55%
0:36 Lay on hands Damage spike (~100k) HP:17%
1:02 Lotp Damage spike (~60k) HP:24%
3:12 Lotp Damage spike (~75k) HP:33%
3:28 Lotp Damage spike (~60k) HP:48%

My on CD casts:
0:17 First hit so Lotp cast. Small damage (~20k)HP:99%
0:32 Lotp Damage spike (~85k) HP:47%
0:47 Lotp Small damage (~15k) HP:50%
1:02 Lotp Damage spike (~60k) HP:24%
1:17 Lotp Damage spike (~40k) HP:48%
1:32 Lotp Small damage (~20k) HP:88%
1:47 Lotp No damage HP:91%
2:02 Lotp Damage spike (~40k) HP: 100%
2:17 Lotp No damage HP:61%
2:32 Lotp Big damage spike (~90k)HP:46%
2:47 Lotp No damage HP:100%
3:02 Lotp No damage HP:100%
3:17 Lotp Small damage HP:53%
3:32 Lotp Big damage spike (~70k) HP:61%

Now on CD cast pretty clearly got ahead of planned casts.

My conclusions:
If there isn't enough damage or when healers are overgearing encounter, you should probably save cooldown for situation. It's tough trying to hit 1-2sec window even with planning already and proven ineffective with on CD usage.

But, if it's progress situation and you are injured state most of the fight, you should use it on cooldown. But I would suggest avoiding casting it very high health 85+%. Because haste gear and Righteous Protector+Hand of The Protector talent are pushing CD of Hotp really low, nearing ~7-8sec. (Current gear)

You remember when I said that this was the final log? Well, I lied.
The (seriously) final log:
Minibat
In this log there is in my mind well used Hotp. It's used very often when there is damage and less when you are at high health. Short CD is around 7sec because Hotp and RD, so during damage phases it's very often available.

Now, as Legion is closer so is also first new raid tier. During firts new raid tier, historicaly this has hardest part of expansion (excluding Wotlk and Naxx) when our characters are at weakest. Healers can't keep you up as efficiently as they are just gathering gear ad are tanks and dps. You are far more likely to be injured state for extended periods of time than now. If tanks are requlary missing health and spending time, say, at 60%-40% we want get as much as possible casts during that time because healers are struggling to heal you far more than they are now. Comparison: At launch of WoD, Grim Rail Depot had some seriously nasty trash packs that required cooldowns. If you now enter there with 735+ gear, I would guess that there is 99% chanse that you aren't in danger to die to them.
Fundus wrote:For the third question, in theory the sim can also test movement, though I've never used those commands before.

It's strange talent, really. It offers non-gap filled rotation but you really need to sit on button, it's not likely offer mitigation so strong as Holy Shield and Blessed Hammer. It's more like: "Dislike these aspects of Prot paladin? Well here is talent for you but you aren't going to be as effective if master those aspects of class."
Dion
 
Posts: 264
Joined: Tue Jul 28, 2009 11:23 pm

Re: Simcraft Thread - Legion/7.x

Postby Fundus » Sun Aug 21, 2016 12:36 pm

Dion wrote:My conclusions:
If there isn't enough damage or when healers are overgearing encounter, you should probably save cooldown for situation. It's tough trying to hit 1-2sec window even with planning already and proven ineffective with on CD usage.

Very nice logs, I just got a chance to look over them and I agree with what you're seeing. I'm also re-running the several of the sims right now and I'll let you know that so far the sim mostly agrees as well. Hopefully I have time to screenshot and post all of them in the coming days. Sorry for the short reply, busy weekend here.
Fundus
 
Posts: 33
Joined: Mon Sep 29, 2014 10:19 pm

Re: Simcraft Thread - Legion/7.x

Postby Baeldin » Mon Aug 22, 2016 7:46 am

Hi Fundus,

Those are indeed very interesting results. This is my first post here, so hello everyone! :)

I have recently started to use Simcraft for 7.0.3 but I am having some issues with getting started. I have managed to sim a series of talent combinations using 840 dungeon gear and an unmodified 840 Truthseeker. My questions concern mainly the artefact and how to implement it and the paladin module itself, so I guess this is the right place to ask.

How do I implement and modify the artefact and the relics?
Even though I downloaded the latest version from GitHub, my paladin module is out of date (first two talent rows switched, the latest divine steed changes not present). Where can I get a new version of that, since you and Liminara seem to be using somewhat custom versions?

If I continue with my sims, and I would like to, I guess it makes sense to use the same version of Simcraf as you and Liminara, so we can compare the results. Feel free to post something along the lines of RTFM, if there is one, but since this is mostly beta stuff I am having a hard time finding an up-to-date manual or a standard download version.

Baeldin

//edit: OK, using premade templates and searching through the SimCraf source code I was able to unravel how artefacts work. It seems to me that I can chose the relics but then I have to set the ilvl and the bonus traits separately, is that correct?
Baeldin
 
Posts: 7
Joined: Mon Aug 22, 2016 5:52 am

Re: Simcraft Thread - Legion/7.x

Postby Fundus » Wed Aug 24, 2016 9:45 pm

Baeldin wrote:OK, using premade templates and searching through the SimCraft source code I was able to unravel how artefacts work. It seems to me that I can chose the relics but then I have to set the ilvl and the bonus traits separately, is that correct?

Yes, it seems that is correct. So as you saw from the default profile the artifact looks something like this:
Code: Select all
main_hand=oathseeker,id=128867,gem_id=137548/136778/137547,relic_id=3412/3412/3412
off_hand=truthguard,id=128866,ilevel=870

The artifact traits are defined separately like talents with ID then # of points in it:
Code: Select all
artifact=49:137548:136778:137547:0:1120:1:1121:3:1124:3:1125:1:1126:3:1128:6:1129:3:1133:1:1135:1:1343:1

However, instead of looking up the ID for each trait or keeping a list, you can use a link to the wowdb.com calculator, which I was I do:
Code: Select all
artifact=http://beta.wowdb.com/artifact-calculator#KZbbbbJJJIAAAAAAAAAA

I think it's easier for someone to copy that out of the report and see what setup we're using than trying to decode the IDs in the default setup.

Baeldin wrote:Even though I downloaded the latest version from GitHub, my paladin module is out of date (first two talent rows switched, the latest divine steed changes not present).

I'm not an editor on Simcraft though I think Lim is. I'm using build level 22423, git build 87d83dd (which should show at the top of the results when you sim). The first 2 talent rows are switched, but Cavalier is present on mine. It appears to work, though now that I look at it, the APL may be using it twice in a row and wasting it. I'll have to check that. I also just noticed another random oddity in the way the SotR line is written using mostly '&' instead of '|' like the other lines. I'll have to test that too. Anyhow, the talent lines being switched isn't a huge deal as long as you remember to account for it when specifying the talents, so I've just been working around it. The other bug Lim mentioned that I haven't been able to test yet is that addition artifact points weren't giving bonus to damage and stamina. Oh and I haven't tried blessing of spellwarding recently, since it kept crashing the sim when I added it to the APL.

*Edit: Ok, I just re-ran the KT vs KT+Cav sims, making sure that Divine Steed (DS) couldn't be cast if it was already up. Didn't make any difference in how many times the sim cast DS. So it is possible that the whole lvl 60 talent row (BoSW, KT, and RA) isn't implemented properly yet. I'm also running those SotR tests right now, because if I'm right I'm going to have to edit every single APL and re-run all of those tests again. I'll try to throw up results below and explain.
Fundus
 
Posts: 33
Joined: Mon Sep 29, 2014 10:19 pm

Re: Simcraft Thread - Legion/7.x

Postby Fundus » Wed Aug 24, 2016 10:30 pm

Ok, so in looking over the default APL I noticed that most of the CD usage calls looked something like this:
Code: Select all
actions.prot+=/eye_of_tyr,if=incoming_damage_2500ms>health.max*0.4&!(debuff.eye_of_tyr.up|buff.aegis_of_light.up|buff.ardent_defender.up|buff.guardian_of_ancient_kings.up|buff.divine_shield.up|buff.potion.up)

For those who don't love code, that tells the sim to use Eye of Tyr whenever it's available and the paladin takes over 40% of max health damage in 2.5 seconds and NONE of the other cooldowns (Ardent Defender, Aegis of Light, etc) are up. Basically the idea is that we want to chain our CDs for smooth damage instead of overlap them all.

Now oddly the default SotR line looks a little different:
Code: Select all
actions.prot+=/shield_of_the_righteous,if=(!talent.seraphim.enabled|action.shield_of_the_righteous.charges>2)&!(debuff.eye_of_tyr.up&buff.aegis_of_light.up&buff.ardent_defender.up&buff.guardian_of_ancient_kings.up&buff.divine_shield.up&buff.potion.up)

Ok, so in the first part it's checking to make sure we don't need to save charges for Seraphim, but it's also going to cast when available unless ALL of the CDs are up (notice the '&' instead of '|'). That means the sim will still cast SotR overlapping other CDs, if there's only 1 or even 2 (which shouldn't happen anyway) up. So I changed it to read:
Code: Select all
actions.prot+=/shield_of_the_righteous,if=(!talent.seraphim.enabled|action.shield_of_the_righteous.charges>2)&!(debuff.eye_of_tyr.up|buff.aegis_of_light.up|buff.ardent_defender.up|buff.guardian_of_ancient_kings.up|buff.divine_shield.up|buff.potion.up|buff.divine_steed.up)

So in the modified version the sim will only cast SotR if NONE of the other CDs are up. And here's the results I get for my modified version (_Mod) vs the default (_Default) for 2 different rotations, using the default talents and gear from above:
Image
Yup, it's a problem, though maybe only a slight one. The modified versions are an improvement on both TMI and DTPS, though they lose a tiny, tiny bit of DPS, possibly from using SotR slight less frequently. At least the two different rotations stay in the same order, but I did pick very different rotations. I know everyone's anxious for more updated data, but I think to be safe I still need to fix this in my APLs. I don't know if Lim or anyone from Simcraft reads this, but ideally it should probably be changed in the default APL as well.
Fundus
 
Posts: 33
Joined: Mon Sep 29, 2014 10:19 pm

Re: Simcraft Thread - Legion/7.x

Postby Baeldin » Thu Aug 25, 2016 2:21 am

//edit: Images fixed

//edit2: THESE ARE PRELIMINARY RESULTS AND THE SIMULATIONS ARE STILL IN THE TESTING PHASE. This means that there will likely be errors, potentially severe ones. I will be back with newer sims this weekend, hopefully more reliable ones.

Disclaimer: For some reason I started by using artifact while others use artefact instead. I am trying to sort this out but I might be using them alternatingly when I don't pay attention. Sorry if that irritates you but for a non-native speaker AE and BE can be confusing at times. :roll:

Since I am on the wrong computer now, I will add later:
- exact Simcraft version
- talent setup

Hi and thanks for showing me the trick with the armory. I found such a line but it was commented and didn't think of just uncommenting it to see if SimCraft is able to use it. That is indeed much simpler. However, for the thing I am trying to test it turned out to be easier to decode the line using the IDs. The reason is that I tried to sim the artifact progression path and trace it in the 2D-Space of DPS and TMI. So basically I chose a recommended path and checked how well it increases defense and offense. The recommended path was chosen almost purely based on common sense and not maths or testing. This is the path I took (German graphic, as found in the German prot paladin guide by Ayende):
Image
ATTENTION: Tyr's Enforcer appears to be purely offensive, today's testing indicates that Bulwark of Order does not benefit from the AoE of Tyr's Enforcer! Also, the sims I am about to show will put this recommendation into question.

I simmed the artefact progression using the same gear as you did, only unlocking the traits one after another. Here is what I got:
Image
Description: Truthguard's Light and Eye of Tyr are active for all because for some reason deactivating them either causes the paladins to do 2 billion DPS or they crash SimulationCraft as soon as the results are displayed. More traits are added until the artifact is maxed out. The dots represent the increase in DPS and decrease in TMI as the artifact grows in power. The X-error bars are symmetric using the DPS-Range value while the Y-error bars use the minimum and maximum TMI. These are 53 paladins using 53 different artifact versions, each consecutive one having one further trait activated, 1000 iterations each. The Y-axis is reversed, such that higher survivability is at the upper end of the scale instead of the lower one.

Comparison to your results:
- My DPS values seem higher than yours (34 Traits would be the data point of Forbearant Faithful, about 145k DPS single target in my sims vs. around 130k in yours)

For the traits as they are in my version (single target!):
- The best DPS-traits are Painful Truths, Consecration in Flame, and Tyr's Enforcer. They are way ahead of everything else. Stern Judgment appears next in line, then Righteous Defender
- The best TMI-traits are Faith's Armor, Light of the Titans, and the first rank of Unrelenting Light
- More targets will favor Consecration in Flame, Painful Truths, Tyr's Enforcer and Hammer Time, because they all scale with the number of targets, but more targets should have no impact on Stern Judgment
- bear in mind that the later a trait is chosen the more effective it can be because different effects often are multiplicative if the enhance the same thing and they scale with the flat damage boost that we get from Truthguard. For example, Painful Truths alone is worth over 4k DPS in the above example. If we get it as quickly as possible it will be a bit less than that, but it will increase the flat damage boost that we get from consecutive points.
- A few words on Painful Truths: This trait modifies one of our defensive CDs, Ardent Defender, to add a strong offensive component. The sims presented above do not take this into account and the damage gain results purely from using it as defensive CD. It's likely the dot that ticks once a second and the relatively high attack speed (1.5) of the custom boss which makes this trait look so strong here. Together, they can damage the paladins up to 12 times within 8 seconds, not counting in the nuke that comes every 24 seconds and might line up as well. However, in the hands of a skilled player this can become a powerful weapon, when combined with offensive CDs on large trash packs. The potential DPS gain is much higher than indicated in the chart, if one is willing to use it as offensive rather than defensive CD. It heavily benefits from Unflinching Defense, which reduces the CD of Ardent Defender by up to 30 Seconds (up to 60 with relics). On the other hand, in you average dotless single target fight its DPS component is probably negligible, hitting anywhere from 3-5 times with 80% of your attack power, making it as weak as one or two uses of Blessed Hammer.

Some discoveries:
- Increasing Stamina and Damage seems to work. In fact, it seems to work beyond where it should. If you check the last 19 ranks of Unrelenting Light, which should only increase armor (producing a vertical line of dots), they still push DPS higher and higher. Likewise, all purely offensive traits also reduce TMI.
- Going for the three golden traits as quickly as possible and then putting one point into Unrelenting Light is a large boost in survivability (instant 10% armor!), the other minor traits should be maxed after that
- Even though in my sims Scatter the Shadows seems really weak, a player using Light/Hand of the Protector intelligently will go a long way in increasing its defensive value. This is something that the sims do not account for.
- Unlocking more traits appears to shorten the upward (down in the chart!) Y-error (TMI), which means that fights with much higher than average TMI become increasingly more rare and the maximum variability upward is reduced as the artefact gains power. This is a welcome change as it seems to flatten damage spikes the further we progress. I am not sure yet why this is happening or if it is even correct though!

The basic idea was to get a measure for the offense/defense ratio of the different traits. Those which move us up increase survival, those which move us right increase DPS. Based on that I could come up with recommendations for paths through our traits depending on whether a player wants to first boost his or her defense or offense. The differences during progression are quite substantial (around 10% more DPS at the cost of about 10k TMI). Eventually, of course, they will all end up in the same point. I also thought about simulating the traits each for themselves, but that's not realistic since most of them can not be reached without maxing several others.

Based on the above, one might consider the following path, due to the fact that the outer circle links the three major traits with only 7 points required between them:
Faith's Armor
Righteous Crusader
Hammer Time
Consecration in Flame
Tyr's Enforcer
Painful Truths (purely offensive until here)
Resolve of Truth (go grab the second major trait, it is close)
Bulwark of Order
Stern Judgment (best DPS option available)
Forbearant Faithful (continue toward the third major trait along the shortest path)
Bastion of Truth
Light of the Titans (third major trait)

At this point you have only spent 27 Point, you have unlocked all major traits and you can unlock rank 1 of Unrelenting Light, instantly granting you 10% Armor. You could even do it with with only 24 points if you skipped Hammer Time and went straight for Resolve of Truth. Hammer Time doesn't look overwhelming anyway.

All that being said, you instantly see that it takes several points to move out of the space enclosed by the error bars of each specific point. This means that each individual trait unlocked will have much less effect than fight-to-fight variation, even for very skilled players and static fights (of which there are practically none). Therefore, the precise path along our artifact tree is somewhat less important than I made it look above. While there certainly are preferable options depending on the goals in game, the difference will rarely exceed single digit percentages and average players will not gimp themselves by taking a different route. However, cutting edge progress players might want to consider this.

Fundus wrote:Yup, it's a problem, though maybe only a slight one. The modified versions are an improvement on both TMI and DTPS, though they lose a tiny, tiny bit of DPS, possibly from using SotR slight less frequently. At least the two different rotations stay in the same order, but I did pick very different rotations. I know everyone's anxious for more updated data, but I think to be safe I still need to fix this in my APLs. I don't know if Lim or anyone from Simcraft reads this, but ideally it should probably be changed in the default APL as well.

What does it do if you arrive at 3 charges? It shouldn't be using fewer SotR unless it starts sitting at 3 charges if any other CD is currently up.
Code: Select all
actions.prot+=/shield_of_the_righteous,if=(!talent.seraphim.enabled|action.shield_of_the_righteous.charges>2)&!(debuff.eye_of_tyr.up|buff.aegis_of_light.up|buff.ardent_defender.up|buff.guardian_of_ancient_kings.up|buff.divine_shield.up|buff.potion.up|buff.divine_steed.up)

So from what I can tell, the above uses SotR if you do not have Seraphim enabled or if you have more than two charges of SotR. Does the "&!" mean "and the following is not true"? In that case, could you add another line below which simply says
Code: Select all
actions.prot+=/shield_of_the_righteous,if=(action.shield_of_the_righteous.charges=3)

Would this use SotR instantly if you have full charges regardless of any CD currently active? This should only happen if SotR is not used directly above, because the condition you added prevents SotR from being used even though it's fully charged? I could be wrong though... still trying to get the hang of this. :)

//edit: For some reason I can not find the info about when Unrelenting Light is unlocked. I am quite sure that I read it is available after unlocking the third major trait, but other sources stat that it only becomes available once all 34 others are unlocked.
Last edited by Baeldin on Thu Aug 25, 2016 10:41 am, edited 2 times in total.
Baeldin
 
Posts: 7
Joined: Mon Aug 22, 2016 5:52 am

Re: Simcraft Thread - Legion/7.x

Postby Dion » Thu Aug 25, 2016 9:25 am

Baeldin, that is intresting actually. Liminaras (or celinamuna in MMO-champions forums, also @bit_hound on twitter) have differing results, taken from wowhead guide (that is written by him).

Damage Rankings for Minor (3/3) Traits : Consecration in Flame > Hammer Time (if Blessed Hammer) > Stern Judgment > Righteous Crusader > Hammer Time (non Blessed Hammer)

TMI Rankings for Minor (3/3) Traits : Resolve of Truth > Bastion of Truth > Righteous Crusader > Scatter the Shadows > Unflinching Defense > Faith's Armor > Stern Judgment

DTPS Rankings for Minor (3/3) Traits : Bastion of Truth > Righteous Crusader > Faith's Armor > Unflinching Defense > Stern Judgment

Faith's Armor is one at that directly leaped on my eyes. Also, are you simming buffed version of Bulwark of Order as I would have thought that it would have pulled ahead of Light of the Titans easily.

Second thing, at least WoWheads artifact talent calculator lets you put points on Unrelenting Light only after when you have maxed on everything else.

Good artifact power spreadsheet So, when mythic raids unlock max artifact trait is ~21 (rares and treasures uncounted in spreadsheet) so it would be good to have somekind of plan of pick up order. For comparison, 21 traits is unlocked in 29days, all talents but no Unrelenting Light bonus trait 101 days and maxing bonus trait 194 days.

Also, this was thrown in Legion thread. I think it contains theorycrafting.
honorshammer wrote:
Sabindeus wrote:
honorshammer wrote:
In addition to Maintankadin, I've been following the Prot Paladin theorycrafting in this Discord: https://discordapp.com/channels/1152880 ... 2499838986


I don't think that link works, you need to generate an invite link

I didn't realize we had theorycrafting or paladin discords? I'm interested!


Sorry about the link not working. Please try with this one: https://discord.gg/rAysD25

There are plenty of players on that discord. And Liminara is there too, using @gray_hound tag.

Fundus, I forgot to answer your earlier questions. Theck has indeed retired, I like to think that he got eaten by wererabbit by amount of bunny pictures he posts on Twitter. I am also same Dion whose argument you saw in MMO-champion thread.
Dion
 
Posts: 264
Joined: Tue Jul 28, 2009 11:23 pm

Re: Simcraft Thread - Legion/7.x

Postby Baeldin » Thu Aug 25, 2016 10:40 am

Dion wrote:Baeldin, that is intresting actually. Liminaras (or celinamuna in MMO-champions forums, also @bit_hound on twitter) have differing results, taken from wowhead guide (that is written by him).

Damage Rankings for Minor (3/3) Traits : Consecration in Flame > Hammer Time (if Blessed Hammer) > Stern Judgment > Righteous Crusader > Hammer Time (non Blessed Hammer)

TMI Rankings for Minor (3/3) Traits : Resolve of Truth > Bastion of Truth > Righteous Crusader > Scatter the Shadows > Unflinching Defense > Faith's Armor > Stern Judgment

DTPS Rankings for Minor (3/3) Traits : Bastion of Truth > Righteous Crusader > Faith's Armor > Unflinching Defense > Stern Judgment

I would have expected Hammer Time to be much better than what my sims show, especially with Blessed Hammer. That's one of the things that I am still looking into. I noticed a few other oddities in my sims which I am working on, but I figured I'd throw the post out anyway. I'll likely have an updated version of my sims this weekend, hopefully with some more information. By the way, I once saw the spreadsheed where Liminara ranked the traits, but I couldn't find it again, do you have a link by chance?

I think one reason why Bastion of Truth ranks so low in my sims is because the Boss does quite some damage via unblockable spells (dot and nuke), which devalues blocking. Since I'm still working on the basic parameters of my sims I didn't bother running them for all talent setups and artifact paths, but once I am confident that my results are not garbage, I'll automate them a bit more and produce more plots and rankings. Right now the priority is to make sure that I am using the same SimCraft version, gear set and enemy as others, and especially, that I don't have any stupid mistakes in my sims, like picking up Unrelenting Light way ahead of time...

This first post was mostly to start a discussion and get some attention from the other theorycrafters out there. After all, there is no quicker way of getting the right answer than posting the wrong one on the internet. :lol:

Dion wrote:Faith's Armor is one at that directly leaped on my eyes. Also, are you simming buffed version of Bulwark of Order as I would have thought that it would have pulled ahead of Light of the Titans easily.

Faith's armor seems weird to me too, especially since rank 2 seems to lower DPS, contrary to what all of the others do. Regarding Bulwark, no, unfortunately not. I tried to correct Bulwark and LotP but changing it in the source code but it did not change their numbers in the sims. That's one of the reasons why I was asking Liminara for an up-to-date version on MMO-Champion (yes, the Baeldin there is also me).

Dion wrote:Second thing, at least WoWheads artifact talent calculator lets you put points on Unrelenting Light only after when you have maxed on everything else.
Good artifact power spreadsheet So, when mythic raids unlock max artifact trait is ~21 (rares and treasures uncounted in spreadsheet) so it would be good to have some kind of plan of pick up order. For comparison, 21 traits is unlocked in 29days, all talents but no Unrelenting Light bonus trait 101 days and maxing bonus trait 194 days.

Yeah, I found that out in the meantime. I guess grabbing the 10% armor that quickly would have been too good to be true.

I briefly joined Discord yesterday evening but I felt kinda lost. I'll revisit it and have a look later this week.
Baeldin
 
Posts: 7
Joined: Mon Aug 22, 2016 5:52 am

Re: Simcraft Thread - Legion/7.x

Postby Baeldin » Fri Aug 26, 2016 3:58 pm

I have had some time to check my resuls further. Stamina increases with each trait, including Unrelenting Light, even though it should not.
Baeldin
 
Posts: 7
Joined: Mon Aug 22, 2016 5:52 am

Re: Simcraft Thread - Legion/7.x

Postby Fundus » Fri Aug 26, 2016 8:38 pm

Fundus wrote:The modified versions are an improvement on both TMI and DTPS, though they lose a tiny, tiny bit of DPS, possibly from using SotR slight less frequently.

Baeldin wrote:What does it do if you arrive at 3 charges? It shouldn't be using fewer SotR unless it starts sitting at 3 charges if any other CD is currently up.

So I'd like to clarify what I meant and what I was trying to do. First off, here's the Full Results from that little test. So the difference in DPS is only about 300, well within the margin of error, but the only reason I see for that difference is that SotR was cast an average of 58.6 times or every 7.85s in the modified version vs an average of 59 times or every 7.8s in the default. That's hardly any difference . Since SotR is first in the APL, I don't think the sim should ever be sitting at 3 charges, at least not with this talent build. I do worry about that effecting a Seraphim build but that's another discussion.
Baeldin wrote:So from what I can tell, the above uses SotR if you do not have Seraphim enabled or if you have more than two charges of SotR. Does the "&!" mean "and the following is not true"?

So first some syntax, since you asked. Sorry if this is over-explained. In simcraft the Action Priority List (APL) runs top to bottom with the sim testing if each action can be performed, due to cooldowns (from spell data) and conditionals ("if"). In writting conditionals "&" is "and" meaning both conditions must be true for the action to be performed, "|" is "or" meaning either of the conditions can be true for the action to be performed, and "!" is "not" meaning that the condition must not be true for the action to be performed.
Ok, so as an example the default SotR line in the APL is:
Code: Select all
actions.prot+=/shield_of_the_righteous,if=(!talent.seraphim.enabled|action.shield_of_the_righteous.charges>2)&!(debuff.eye_of_tyr.up&buff.aegis_of_light.up&buff.ardent_defender.up&buff.guardian_of_ancient_kings.up&buff.divine_shield.up&buff.potion.up)

So breaking down what that says: Cast SotR if we 1)do not have seraphim enabled OR (if we do have Seraphim) whenever we have more than 2 charges of SotR AND 2)do not have Eye of Tyr and AoL and AD and GoAK and DS and potion and DS up.
The first part is why the sim shouldn't be sitting on 3 charges. With Seraphim talented is the only time that the ">2" charges should come into play. Otherwise it'll just cast until we're out of charges, as long as the 2nd part of the condition is met. The second part is what I was questioning, because we're never going to have ALL of those abilities up at once, so the 2nd part of the condition will always be false, making the "not" always activate. Essentially that means the sim will still cast SotR during GoAK or even during Divine Shield (with Final Stand talented, because that's the only time DS is cast), which I think we can agree doesn't make sense. Ideally the sim should be trying to chain all the CDs, which is what I was trying to do in modifying it.

I think the SotR line should be (and I changed it in the modified version to):
Code: Select all
actions.prot+=/shield_of_the_righteous,if=(!talent.seraphim.enabled|action.shield_of_the_righteous.charges>2)&!(debuff.eye_of_tyr.up|buff.aegis_of_light.up|buff.ardent_defender.up|buff.guardian_of_ancient_kings.up|buff.divine_shield.up|buff.potion.up|buff.divine_steed.up)

So breaking down what that says: Cast SotR if we 1)do not have seraphim enabled OR (if we do have Seraphim) whenever we have more than 2 charges of SotR AND 2)do not have Eye of Tyr or AoL or AD or GoAK or DS or potion or DS up.
So basically without seraphim talented the sim will cast SotR whenever it's available, unless there's another cooldown up. That seems like it should be better and, as I posted above, it does in fact lead to a better TMI.
Baeldin wrote:In that case, could you add another line below which simply says
Code: Select all
actions.prot+=/shield_of_the_righteous,if=(action.shield_of_the_righteous.charges=3)

Would this use SotR instantly if you have full charges regardless of any CD currently active?

You are correct, that would use SotR whenever you hit 3 charges. Again, that seems unlikely, but let's test. So I'm running a version with my fix and your line directly underneath it in the APL (Mod2).
Image
Ok, so Mod2 (my change + your change) is the overall winner for TMI on both rotations, though we're right around the margin of error with differences of 50 to 270 TMI. It's also better than the first Mod for DPS on both, though again by less than 100 DPS. So you must be right that the sim is hitting 3 charges somehow at some point and wasting refresh time. I'll try including that line in my APLs just in case. Thanks.

Baeldin wrote:Comparison to your results:
- My DPS values seem higher than yours (34 Traits would be the data point of Forbearant Faithful, about 145k DPS single target in my sims vs. around 130k in yours)

That would make sense if you're using the newer default gear set, since it's using a higher ilvl. I really need to update my gearset I guess, but I was keeping it for the moment for consistency with previous results. If you save the results output file (put a file name in the white bar at the bottom and hit the save button on the right) and upload it somewhere, I'll look and confirm. Or upload a screenshot of the gear output from the sim (it's near the bottom of each result). Should look like this:
Image

Baeldin wrote:I have had some time to check my resuls further. Stamina increases with each trait, including Unrelenting Light, even though it should not.

So they over-fixed it. It used to be not increasing at all, now it's too high for some configurations. Maybe let Lim know over on mmo-champ?

Awesome graph of artifact traits by the way. Any way we could get the raw TMI and DPS numbers to mess around with some more?
And one other comment on that data, mostly for future reference:
Baeldin wrote:These are 53 paladins using 53 different artifact versions, each consecutive one having one further trait activated, 1000 iterations each.

I know that's a lot of paladins and takes a while to run, but I'd recommend doing more iterations to reduce the variability and margin of error. For most things 10000 is probably fine, but when we get into stat plots and priorities, using 25000 is preferable. I know this makes the sims take much longer to run, but it generally gets the TMI error under 100. In case you're wondering about where you can find the error, it's in the tables under "Results, Specs, and Gear" before you get to the "Charts" section.
Fundus
 
Posts: 33
Joined: Mon Sep 29, 2014 10:19 pm

Re: Simcraft Thread - Legion/7.x

Postby Dion » Sat Aug 27, 2016 12:57 am

Baeldin wrote:By the way, I once saw the spreadsheed where Liminara ranked the traits, but I couldn't find it again, do you have a link by chance?

I remember seeing said spreadsheet but I really can't remember where or seem to find it. Shame.
Baeldin wrote:Regarding Bulwark, no, unfortunately not. I tried to correct Bulwark and LotP but changing it in the source code but it did not change their numbers in the sims.

For the record, there is twist in Bulwark. You see, hotfix notes say it's 40% of damage yet people in Beta have stated that it would actually be 50%. So if it's actually 50% compared to earlier 20% buff is so heavy that it probably pulls ahead of Light of the Titans. And on single target fight, First Avenger could be intresting.
Baeldin wrote:Faith's armor seems weird to me too, especially since rank 2 seems to lower DPS, contrary to what all of the others do.

Why? Most powerful question I can pull top of my head. Why gravity works? Why we can see stars? Why sky is blue? Why 1+1 is 2?

This is what in my mind defines good theorycrafting, answering never ending why questions instead of simply stating because "sims say so". This complainment was on general level and not aimed to you guys.

So, why is Faith's Armor, 30% armor at 40% HP, is so much worse than 10% armor all the time? Or why it's so good at Baeldis sims?
Fundus simmed Consecrated Ground to be better than Judgement of Light, how big difference it gets when we start getting that haste+versatility gear because I think while Judgement of Light only benefits from versatility with bigger heals, Consecrated Ground benefits from both with faster heal ticks and bigger heals.

Second thought: While our Consecration reaches melee group if boss has hitbox of Millhouse Manastorm, it's probably not happening. But it's not really hard to imagine our co-tank managing to stand our Consecration. So, while Judgement of Light heals whole raid and by no small amount, Consecrated Ground might heal our co-tank on the side.

I'm also wondering how big DPS gain is Tyr's Enforcer? When everyone says that Light of the Titans is last pick, I wonder what are sacrificing for Tyr's Enforcer. Because, we gain very, very much defensive traits by going Light of the Titans -> Bulwark of Order. Tyr's Enforcer doesn't work with Bulwark of Order so it's purely DPS trait.

Traits gained:
Scatter the Shadows (Lotp heal boost)
Unflinching Defense (reduces cooldown of Ardent Defender)
Bastion of Truth (6%more block chanse)
Faith's Armor (increased armor on low HP)
Righteous Crusader (increased damage reduction and increases damage we deal with ShotR)
Resolve of Truth (HP increase)

We lose Consecration in Flame trait for not picking Tyr's Enforcer. Trait has synergy with ShotR and LotP and Consecrated Ground so it's not purely DPS trait but Stern Judgement I count pretty much DPS trait as 6% increased Judgement chanse is very minor survival increase. And Hammer Time is pure DPS trait, so we gain DPS and sacrifice survivability.
Dion
 
Posts: 264
Joined: Tue Jul 28, 2009 11:23 pm

Re: Simcraft Thread - Legion/7.x

Postby Baeldin » Sat Aug 27, 2016 4:23 am

OK, so I understood correctly what your plan with SotR was. The line, as it was before your fix, did indeed not make sense. The line I proposed is just some last resort at the bottom of the list, and I'm actually happy that it works. :mrgreen:

Fundus wrote:That would make sense if you're using the newer default gear set, since it's using a higher ilvl. I really need to update my gearset I guess, but I was keeping it for the moment for consistency with previous results. If you save the results output file (put a file name in the white bar at the bottom and hit the save button on the right) and upload it somewhere, I'll look and confirm. Or upload a screenshot of the gear output from the sim (it's near the bottom of each result). Should look like this:

//edit: My set is actually 825/855, not sure what causes the difference in DPS. I'll post the gear below, also the stats of my paladins.

Fundus wrote:I know that's a lot of paladins and takes a while to run, but I'd recommend doing more iterations to reduce the variability and margin of error. For most things 10000 is probably fine, but when we get into stat plots and priorities, using 25000 is preferable. I know this makes the sims take much longer to run, but it generally gets the TMI error under 100. In case you're wondering about where you can find the error, it's in the tables under "Results, Specs, and Gear" before you get to the "Charts" section.

I know that I should run more of them, I was just trying to save time. I was certain that there would be at least some kind of problem with my sims and so I didn't bother to do a very long run. I'll set the itereations to 10k or even 100k once I'm confident that I fixed all there is to fix.

For the chart, I'm using a little shell script to extract the TMI and DPS values. If you are running your sims on Linux, I can send you the script. It basically extracts all the DPS and TMI lines from a text output and clears the columns such that they can easily be imported into Excel or OpenOffice Calc, which I use to make the chart that I posted.

Dion wrote:For the record, there is twist in Bulwark. You see, hotfix notes say it's 40% of damage yet people in Beta have stated that it would actually be 50%. So if it's actually 50% compared to earlier 20% buff is so heavy that it probably pulls ahead of Light of the Titans. And on single target fight, First Avenger could be intresting.

I can confirm that it is at 50%, but there's more:
- while at some point it seemed to benefit from Enforcer, it does not do so any more
- it appears to benefit from versatility (which would effectively be double-dipping, something that blizzard traditionally tries to avoid)

Dion wrote:Why? Most powerful question I can pull top of my head. Why gravity works? Why we can see stars? Why sky is blue? Why 1+1 is 2?

This is what in my mind defines good theorycrafting, answering never ending why questions instead of simply stating because "sims say so". This complainment was on general level and not aimed to you guys.

I think I didn't make myself clear. I think that it's an error of some kind and I have been looking for the cause.

With Liminaras help I was already able to find the reason for a weird behavior I had with my first two paladins. When using relics but spending fewer points on the artifact than there are relics, SimCraft suffers from and underflow and has everything hit in the 2 billion damage range and sims 4 trillion DPS. Removing the relics solves this problem Considering that the relics do nothing but changing the ilvl and add a point to a minor trait, I see no use in specifying them right now. Everything they do is specified elswhere (ilvl, trait lvls) and they have not other stats anyway, so it seems to me that they cause more problems than they solve.

Dion wrote:I'm also wondering how big DPS gain is Tyr's Enforcer? When everyone says that Light of the Titans is last pick, I wonder what are sacrificing for Tyr's Enforcer. Because, we gain very, very much defensive traits by going Light of the Titans -> Bulwark of Order. Tyr's Enforcer doesn't work with Bulwark of Order so it's purely DPS trait.

The latest Bulwark change will probably bring it up on par with Light of the Titans even in single target fights, and make it pull ahead for more than one target, at least that would be my guess. Going for Bulwark AND LotT first lets us miss out on most of our best DPS-traits: Hammer Time, Consecration in Flame, and Enforcer. Unless it somehow turns out in the next two days that we need more survivability that soon into the addon, I would still not recommend doing that, unless you have the mentality that you want to be as sturdy as possible at all costs.

Dion wrote:We lose Consecration in Flame trait for not picking Tyr's Enforcer. Trait has synergy with ShotR and LotP and Consecrated Ground so it's not purely DPS trait but Stern Judgement I count pretty much DPS trait as 6% increased Judgement chanse is very minor survival increase. And Hammer Time is pure DPS trait, so we gain DPS and sacrifice survivability.

I did some back-of-the-envelop calculations on SotR uptime and came up with a formula that reproduces the results of Liminara quite accurately (within 0.01% of uptime). According to that formula, Stern Judgment is very weak defensively. Three ranks increase the SotR uptime by around 2.5 percentage points. Here's the spreadsheet:
https://docs.google.com/spreadsheets/d/ ... sp=sharing

@Fundus: Since I was able to fix a few errors, I'l now run 10k iterations of an offensive, defensive and intermediate route through our artifact and redo the chart I made above. To save some time I will stop adding Unrelenting Light. It's bugged and it will be the same for everyone. What's interesting is whatever happens before.
Baeldin
 
Posts: 7
Joined: Mon Aug 22, 2016 5:52 am

Re: Simcraft Thread - Legion/7.x

Postby Dion » Sat Aug 27, 2016 8:52 pm

Baeldin wrote:I still not recommend doing that, unless you have the mentality that you want to be as sturdy as possible at all costs.

It will depend on the cost. Would I be trading 20% DPS to 20% survivability? 50% of DPS to 5% survivability? 5% of DPS to 50% survivability? Numbers makes or breaks the trade.

This is not accurate but as 21 traits are earned around 29 days of launch. All 34 traits are earned around 101 days from launch. So we need to make that trade thinking what do we need for future as we can't change minds instantly.
Blizzard wrote:5.They won't keep the game imbalanced for a long period of time, but in the short term there is a lot of disruption that comes from players feeling like the time they spent powering up their artifact was a mistake.

So, this means a lot players are not using guides (possible) or it's as clear cut as guides make it.
Dion
 
Posts: 264
Joined: Tue Jul 28, 2009 11:23 pm

Re: Simcraft Thread - Legion/7.x

Postby Baeldin » Sun Aug 28, 2016 6:40 am

Damn. I was actually planning to publish new sims today but my Laptops speaker - only the right one - stats buzzing violently as soon as I boot into Linux (not in Windows) and I have no idea how to fix it. It gets a LOT worse as soon as I start the simulation, so it might be a heat/fan problem but it's certainly coming from the speaker. It's driving me nuts and since I can't build simcraft in Windows there's no way around it. Pretty fucking annoying. Visual Studio would cost me 500€ whereas the compilers on Linux are completely free and installed entering one line in the shell...
Baeldin
 
Posts: 7
Joined: Mon Aug 22, 2016 5:52 am

Next

Return to Advanced Theorycraft and Calculations

Who is online

Users browsing this forum: No registered users and 1 guest

Who is online

In total there is 1 user online :: 0 registered, 0 hidden and 1 guest (based on users active over the past 5 minutes)
Most users ever online was 380 on Tue Oct 14, 2008 6:28 pm

Users browsing this forum: No registered users and 1 guest