Simulationcraft results for WoD

[elfporn]

did you put together the new bis list yet aethys?

saw you mention it at some point

[Feali]

Did anyone sim the new Heirloom trinket from Mythic Dungeons yet?

[fierydemise]

It wasn't very clear that that's what you meant. I would also like to see a source/argument for why you think so. Especially when you make such an absolute statement. According to the BiS list in the first post Fang is considered the best. When I put my character into Shadowcraft it tells me the same.

You might be right, but you're not making a compelling argument.

I don't want to turn this into an ShC thread, that is why I spun that off in the first place but I feel its worth noting that ShC for some complex reasons doesn't do a very good job with current sub, especially not on very short fights like most people are experiencing and SimC is modeling (SimC website uses 4 minutes for stack ranks). Basically ShC tries to get at steady state dps, a lot of the assumptions don't hold or don't approximate as well over shorter timescales. Sub is particularly bad because with T18 4pc and prep it takes a really long time to settle into steady state dps. On a 4 minute fight when you are spending probably 30+% outside of steady state things don't work as well. Then you have the ring which makes things even worse, the ring heavily amplifies damage during these non-steady state periods so if we're already undervaluing those it will only make things worse. In particular where this manifests is stuff like OH dps and haste which are actually pretty ok during those steady state periods but highly devalued during periods where you are primarily using MH weapon damage with big ambushs and/or pooling heavily so energy doesn't really matter.

These problems are solvable, and I'd like to solve them, however I don't plan to do it in the current engine. I know we've probably got 4-6 months ahead of us but I basically consider WoD over. There are a lot of potentially interesting directions for ShC to go that I believe can solve some of these issues but trying to graft them into old models will be complicated especially if instead I can design the new Legion spec models with these new structures in mind.

This also seems like a good place to plug ShC development. Right now I work on the engine, Tamen works on the UI and thats about it. Getting new features or improved modeling in is really hard when we spend most of our time doing maintenance on what we have now. If we had a couple more folks working on ShC we could probably work on more cool stuff. If you are interested shoot me a PM here or I'm almost always in #Ravenholdt IRC on quakenet.

[Aethys256]

These problems are solvable, and I'd like to solve them, however I don't plan to do it in the current engine. I know we've probably got 4-6 months ahead of us but I basically consider WoD over. There are a lot of potentially interesting directions for ShC to go that I believe can solve some of these issues but trying to graft them into old models will be complicated especially if instead I can design the new Legion spec models with these new structures in mind.

And I follow this approach.
I won't make intense modifications to WoD Rogue SimC -myself- (so others can still do/suggest) because beta might be soon released and then I will be focused onto new specs.
About Sub APL, there is still room to improve it.
You can find 2 big (relatively lol) and 2 minor things that make DPS down :
- Vanish may be delayed under some circumstances (while it's a dps loss)
- Shadow Dance may be used while you're already at 10 cp( in fact 9 + 1 HaT incoming the GCD you will use SD) leading to a loose of at least one Ambush).
- Sometimes Rupture uptime may be a little off.
- Sometimes S&D uptime may be a little off.

These issues are not quick to fix because they do not occurs everytime and then would take me a couple of hours that at this moment I do not have.

This also seems like a good place to plug ShC development. Right now I work on the engine, Tamen works on the UI and thats about it. Getting new features or improved modeling in is really hard when we spend most of our time doing maintenance on what we have now. If we had a couple more folks working on ShC we could probably work on more cool stuff. If you are interested shoot me a PM here or I'm almost always in #Ravenholdt IRC on quakenet.

May take a look if you don't mind
Will pm you.

[Deleted]

Hello fellow rogues,

so I got "lucky" and whilst farming mythic dungeons with my guild received the following: http://www.wowhead.com/item=109998/g...&bonus=642:760

I was wondering if it's worth using this over SoulCap WF or Class trinket at all (I only have HC versions of those so far). I'm already passing 100% multistrike using the spike, so I haven't upgraded it yet. I'm around 725 ilvl in sub.
Probably not many have looked at those trinkets at all since the drop chance is so low. At first thought I can only see it being useful in fights where SoulCap could potentially ruin the fight IF you don't pay attention (farming Gorefiend, whatever). However comparing trinkets requires a little bit more than my maths skills - was hoping to find some answers here.

Cheers!

[Deleted]

Hello fellow rogues,

so I got "lucky" and whilst farming mythic dungeons with my guild received the following: http://www.wowhead.com/item=109998/g...&bonus=642:760

I was wondering if it's worth using this over SoulCap WF or Class trinket at all (I only have HC versions of those so far). I'm already passing 100% multistrike using the spike, so I haven't upgraded it yet. I'm around 725 ilvl in sub.
Probably not many have looked at those trinkets at all since the drop chance is so low. At first thought I can only see it being useful in fights where SoulCap could potentially ruin the fight IF you don't pay attention (farming Gorefiend, whatever). However comparing trinkets requires a little bit more than my maths skills - was hoping to find some answers here.

Cheers!

I might be wrong on this, but I believe the mythic dungeon trinkets are underbudgetted. As you can see, the 725 trinket only has only 257 agility. A 725 Soul Capacitor has 502 agility. The Soul Capacitor proc is also a lot stronger when it comes to pure dmg output. It is probably better to compare this trinket to Malicious Censer, to see which one is the better SC replacement for certain fights. Hard to estimate the value of the on use effect, but the base stats give it a big disadvantage.

[Deleted]

Hi Guys,

First off thanks for the amazing work, been visiting this thread from time to time and its my first source for information around new patches.

So what i dont really understand is why now after 6.2.3 the Zakun offhand (Demonblade Evicerator) is better than the one from Zul (Hammer of something something)?

So far as i can understand it the one from Zul has more MS and Crit instead of haste? Did the Stat balance change somehow with the new patch and haste is now not garbage anymore?

Appreciate your help!

Edit: Talking about Sub sorry if that wasnt clear

[Deleted]

I might be wrong on this, but I believe the mythic dungeon trinkets are underbudgetted. As you can see, the 725 trinket only has only 257 agility. A 725 Soul Capacitor has 502 agility. The Soul Capacitor proc is also a lot stronger when it comes to pure dmg output. It is probably better to compare this trinket to Malicious Censer, to see which one is the better SC replacement for certain fights. Hard to estimate the value of the on use effect, but the base stats give it a big disadvantage.

That is pretty much the conclusion I arrived at after pondering on it during downtime at work today. Thank you for your time man

[Deleted]

Why am I simming 1.2k higher using warsong on my offhand over frostwolf simming as Sub

[maddnesskhaz]

Why am I simming 1.2k higher using warsong on my offhand over frostwolf simming as Sub

Some kind of MS cap with the higher ilvls perhaps? I assume the clever mathsy folk might need to know the gear you were simming.

[Deleted]

Some kind of MS cap with the higher ilvls perhaps? I assume the clever mathsy folk might need to know the gear you were simming.

I was simming 739 ilevel

head=,id=124263,bonus_id=567
neck=,id=124210,bonus_id=567,enchant_id=5320
shoulder=,id=124274,bonus_id=565/567,upgrade=2,gem_id=127763
back=,id=124134,bonus_id=567,enchant_id=5313
chest=,id=124248,bonus_id=567
shirt=,id=98091
tabard=,id=128450
wrist=,id=124280,bonus_id=562/567,upgrade=1
hands=,id=124257,bonus_id=562/567,upgrade=2
waist=,id=124276,bonus_id=567
legs=,id=124265,bonus_id=562/565/567,upgrade=2,gem_id=127763
feet=,id=124252,bonus_id=565/567,upgrade=2,gem_id=127763
finger1=,id=124189,bonus_id=567,enchant_id=5327
finger2=,id=124636,bonus_id=650/638,enchant_id=5327
trinket1=,id=124520,bonus_id=567,upgrade=2
trinket2=,id=124225,bonus_id=562/565/567,upgrade=2,gem_id=127763
main_hand=,id=124366,bonus_id=562/567,upgrade=2,enchant_id=5334
off_hand=,id=124371,bonus_id=565/567,upgrade=2,gem_id=127763,enchant_id=5337

# Gear Summary
# gear_ilvl=733.81
# gear_agility=4786
# gear_stamina=6384
# gear_crit_rating=1872
# gear_haste_rating=969
# gear_mastery_rating=565
# gear_multistrike_rating=2655
# gear_versatility_rating=248
# gear_armor=1389
# set_bonus=tier18_2pc=1
# set_bonus=tier18_4pc=1

[Aethys256]

I didn't checked your gear, but it is probably because you have almost 0 haste (I mean, not that much), and haste weight value the more you have others. See below.


@Woni, I explained earlier in the thread but I will try to put an example.

DISCLAIMER : I know that TCers will have a heart-attack to read this, but it's the best image I can give to understand how stats works "basically".

Imagine that every secondary stats are a multiplicative factor of your damage.
Then, imagine that you have 100 multistrike, 60 mastery, 40 haste, 20 crit, 10 versa and stats weight are equal initially (do not want to add too much complexity, it's not the real case but it will explain the phenomenon we're talking about).

So, with this hypothesis in mind, it would end up at something like this (1 is your damage) :
A = 1 * 0.5 * 0.3 * 0.2 * 0.1 * 0.05 = 0.00015
Then you may intuit that having more versa or crit in our example would increase the damage.

So we have 230stats points, let's say that now we have 60 ms, 50 mastery, 50 haste, 40 crit, 30 versa.
B = 1 * 0.3 * 0.25 * 0.25 * 0.2 * 0.15 = 0.00056 (> A)

And I know that now you can find yourself that having same stats repartition (STILL IN OUR HYPOTHESIS, IT'S NOT HOW IT WORKS REALLY) is the best compromise :
C = 1 * 0.23 * 0.23 * 0.23 * 0.23 * 0.23 = 0.00064 (> B)

Of course it is way more complicated and it is what we call Theorycrafting to determine how each stats vary our dmg. Because some passive, mechanics or just how stats (for ex Crit) are designed amplify some stats value (for Sub, MS is really high due to Sinister Calling + how it works) BUT you will ALWAYS end up with a stats that skyrocket if you put 0 point there even if it's the worst one.

@TCers, it's not intended (we aren't on EJ ) to explain in details everything because it's for players who don't know how stats works in dmg calculation. I may someday (let's say for Legion) make proper explanations on how it works "really".

PS : You may remember that Blizzard increased the effectiveness of Haste early in WoD by increasing the % rate per point. They wanted to increase this stat value, increasing the rate is applying a scalar factor to your value, so in our example haste value would end at original_value*1.1 (But then would again be modified by how it affects each spec, everything is relative and "linked")
Exacts calculations are way more complicated because every stats doesn't give you "% dmg" but rather chance to make XX things so they aren't directly multiplicative factor to your dmg, that's why Theorycrafting exists.


Edit : In 6.2.3, we have +10 ilvl, so we have more of some stats that is why others grow up. (Depends of your gear)

[Deleted]

Aha so the fact that i completely ignored haste up until now makes it more valuable than multistrike (in case of the offhand) which i have stacked to a very high amount thus is the zakun offhand better. Thanks man I got a little grasp on the situation now not saying i understand everything on stat weight but that definitely helped, ur awesome.

[Deleted]

Statvalue and DPS with the current simprofile, Xhulorac OH: http://i.imgur.com/hkYKrJI.png

Statvalue and DPS with the current simprofile, Zakuun OH: http://i.imgur.com/lz0Prjs.png

So here is the riddle, how is it possible, that the Zakuun OH outperforms the other alternatives, despite Haste being the lowest secondary stat in stat values?
Well, it's because of how many actions can be performed due to the limitation of the APL. In this case the ~200 Haste of the Zakuun OH ensure a third Ambush use during Vanish, while the Xhul/Manno OH is only able to perform 2 Ambushes most of the time, due to the energy pooling limitation set by the APL.

Some examples, 25k Iterations, 450 fight length(no variation):

APL: actions+=/pool_resource,for_next=1,extra_amount=100
actions+=/vanish,if=talent.subterfuge.enabled&energy>=100&combo_points<4-talent.anticipation.enabled&buff.shadow_dance.down

Ambush Interval for Zakuun OH: 6.18sec
Ambush Interval for Xhulorac OH: 6.41sec

APL: actions+=/pool_resource,for_next=1,extra_amount=110
actions+=/vanish,if=talent.subterfuge.enabled&energy>=110&combo_points<4-talent.anticipation.enabled&buff.shadow_dance.down

Ambush Interval for Zakuun OH: 6.15sec
Ambush Interval for Xhulorac OH: 6.31sec(at this point the Xhulorac/Manno OH is superior DPS-wise to the Zakuun OH)

APL: actions+=/pool_resource,for_next=1,extra_amount=115
actions+=/vanish,if=talent.subterfuge.enabled&energy>=115&combo_points<4-talent.anticipation.enabled&buff.shadow_dance.down

Ambush Interval for Zakuun OH: 6.13sec, DPS: 133219.3
Ambush Interval for Xhulorac OH: 6.27sec, DPS: 133486.2

There are still some cases, where the sim isn't able to dish out 3 Ambushes during Vanish, but I guess people shouldn't worry about it too much. As long as you are able to perform 3 Ambushes during Vanish in "real life" Manno/Xhul OH will still be your choice as offhand.
The difference between those two is very minimal and will most likely change depending on how many sockets and which WF items you have. In the vanilla BiS gear though, Hammer of Wicked Infusion seems to be slightly superior to Fang of the Pit by about 50-60 DPS.

Thanks a lot to Aethys for improving and maintaining the profiles as well describing the concept of stat weights, though in this case the latter wasn't the reason for the problem, since Haste never overtook any secondary stat in its value.

[Aethys256]

Yes as we spoke on IRC yesterday, I agree.
I will try to look into the pooling issue during the week-end

[Aethys256]

So I tested further thing, and here are the results :

Every sims are 100k iterations each, 0.2% dps error, the duration go from 60s to 480s each +-20%.
The table is :
060 - 090 - 120
150 - 180 - 210
240 - 270 - 300
330 - 360 - 390
420 - 450 - 480

I did 3 changes :

- Prep : Fix a minor thing with Preparation wich occurs very rarely and then not "visible" here (reduced the cd timer of vanish to use prep)
- V-Pooling : Incresed the pooling to 115 for Vanish (was the best value I found by an increment of 5, can still find a better between 110 and 120 but that will vary depending on gear)
- SM-Pooling : Only for Night Elf, will modify how it uses Shadowmeld by increasing his pooling value. Here, 115 is also the better. Why ? Because when we ShadowMeld it is to re-apply FW, so there is no point using it at 60 energy because you won't be able to use others abilities after that.

Overall, what it changes the most is we sync better with Maalus (gaining 1-3s of uptime), and as nowadays we use it instantly when CD is off it fit better the reality.

Finally, about Xhul/Zakuun debate, it is incomplete there. I will run these with the 500 gearsets I've choosen for Sub tonight to see if another piece to swap fit better or not.

Enjoy !



As you can see, the shadowmeld edit is better if we go over around 120s, else it is a bit lower.

[mitre27]

Mirror of the blademaster boosted by deep insight?

[Coldkil]

Jumped in only to clarify the sticky/unstick thing.

I recognize the usefulness of this thread, and i praise the work of many rogues in here that made everything in their power to provide our class community with useful information. The thread is so famous that is still being bumped and discussion is still going on here - even if it's not stickied, i don't see it ever going down lower than the first 5 in the page.

On the contrary, as soon as Legion news started coming out, people started flooding the forum with all kind of threads. There was a Legion thread sticky already which was pretty active, but this didn't stop people form creating countless threads about the same stuff over and over.
I sticked the threads in hope of making them actually used, but in the end the "activity" has been mostly me merging and redirecting other threads into the stickies nearly every day.

And all this ended into having a huge amount of sticked threads that didn't do what they was made for.

I made an evaluation error here, and i'm sorry for this. The huge work some people has done here didn't deserve this - thread restickied.

And i assure you it won't be taken down until Legion release.

[Aethys256]

Here it is, we got our winner, it is the n°386 guy !
Iter : 100k, DPS Error : less than 0.01%, mostly 0.006% ~= 60-80dps, Fight Duration = 240s +-20%


And it is with Xhul OH
Also, note that fang isn't "far" behind nor Zakuun.
Finally, the other piece that is not that bad is the Zakuun Bracers VS Killrog one, but Killrog still better.

It is probably the last update of Sub until Legion (or a new patch, ahaha ) except maybe for T18B (gear entirely socketed/wfed) or an APL cleanup.

https://github.com/simulationcraft/s...c6a32cadda047b

Also, this episode was a great example of how important APL are and how they need to be maintained to let us compare efficiently different gear or different talents using simcraft (that's why we support "useless" talent in APLs like Shadow Focus for sub, to let you see how it is bad in comparison of Subterfuge, and that's the aim of Simulationcraft, find the best setup in a given context).

I'm pretty sure simc will be again useful at Legion, because with all these talents/combinations of talents, we might be surprised about a setup that we didn't thought at a first glance !
Also, if you want to improve your skills using the tool or simply contribute to the community (by doing Pull Request on Github), do not hesitate to read the wiki.
https://github.com/simulationcraft/simc/wiki

I may give a shot to combat & ass, but with the next legion build coming with all the infos needed to update rogue engine on simc (+ doing first APL), it is unlikely that I will update these (if update there is). If someone want to take a look, again, do not hesitate

[Todesbote]

Jumped in only to clarify the sticky/unstick thing.

I recognize the usefulness of this thread, and i praise the work of many rogues in here that made everything in their power to provide our class community with useful information. The thread is so famous that is still being bumped and discussion is still going on here - even if it's not stickied, i don't see it ever going down lower than the first 5 in the page.

On the contrary, as soon as Legion news started coming out, people started flooding the forum with all kind of threads. There was a Legion thread sticky already which was pretty active, but this didn't stop people form creating countless threads about the same stuff over and over.
I sticked the threads in hope of making them actually used, but in the end the "activity" has been mostly me merging and redirecting other threads into the stickies nearly every day.

And all this ended into having a huge amount of sticked threads that didn't do what they was made for.

I made an evaluation error here, and i'm sorry for this. The huge work some people has done here didn't deserve this - thread restickied.

And i assure you it won't be taken down until Legion release.

Thank you for explaining that & resticky-ing the thread. Also thank you for putting in the work to moderate this forum. Without that work this forum would be a clusterfuck of multiple threads on the same issues where no meaningful conversation could ever take place. So thank you!