Mists of Pandaria: All Specs - Page 19

Berthold · 09/10/12, 4:35 AM

fixed the reporting bug for the boss dps and some others. The boss is now hitting harder (values from Sha of Fear 25 heroic), so do not be surprised by the warrior's dps.

New runs for Prot Warrior:

Shield Block + additional rage into Shield Barrier: Simulationcraft Results

This is also using 5.0.5 spell data. The profile now looks good to me and even gives reasonable hit/exp scalefactors. If you find anything odd, please mention it.

p.s. Unfortunately, the latest simc release 5.05-2 does not include the fix for SB (but anything else). Still, you should play around with it a bit.

xstratax · 09/12/12, 2:58 PM

Can anyone point me to a good source for Base Strength and Stamina values for the Warrior races (Preferably at 90)? Searching WoWWiki and WoWpedia only gives me the old character creation values, and then class bonuses. Those sites are in need of massive updates, sadly.

Caltiom · 09/13/12, 5:13 PM

Originally Posted by Dots

That makes sense. Using it ASAP may yield a better overall result though. The shield only lasts 6 seconds, so there is a chance that part of it will be wasted. That is less likely to happen if you just use Shield Barrier whenever possible. Probably not a huge difference, but worth looking at.

You should be able to observe the difference in the detailed Shield Barrier results between the actual amount ( the actual amount of absorbtion used by the warrior to mitigate physical hits on him ) to the total amount ( the theoretical maximal amount when applying the shield ). Looking at Simulationcraft Results that's 3116825 to 4486644 or 70%.
But you'd have to compare death counts ( with a lot of iterations ) to see whether using it more rarely with 60 rage but only when needed is better than using it more often with less rage.

Malevolencia · 09/14/12, 9:40 AM

Originally Posted by Berthold

fixed the reporting bug for the boss dps and some others. The boss is now hitting harder (values from Sha of Fear 25 heroic), so do not be surprised by the warrior's dps.

New runs for Prot Warrior:

Shield Block + additional rage into Shield Barrier: Simulationcraft Results

This is also using 5.0.5 spell data. The profile now looks good to me and even gives reasonable hit/exp scalefactors. If you find anything odd, please mention it.

p.s. Unfortunately, the latest simc release 5.05-2 does not include the fix for SB (but anything else). Still, you should play around with it a bit.

Hmm, one thing seem odd to me.

This is a single target test afaik, yet the number of times Thunder Clap is being used is ridiculously high, at every 7.32seconds. Even if your concern is maximising dps, it is lower damage than devastate (both for hits, crits and average, and I am assuming "average" is taking account of misses too) so it should really be being used far less than it is (just enough to keep up the debuff)? Doing devastate more often will yield more S&B procs giving more shield slams/bonus rage and thus more rage to defend yourself with too (which may alter exp/hit weightings further).

Looking at the conditions for it, it seems to only be meant to TC every time the debuff falls off... so the Weakened Blows debuff isn't being simmed properly?

reighnman · 09/14/12, 4:44 PM

Originally Posted by Berthold

fixed the reporting bug for the boss dps and some others. The boss is now hitting harder (values from Sha of Fear 25 heroic), so do not be surprised by the warrior's dps.

New runs for Prot Warrior:

Shield Block + additional rage into Shield Barrier: Simulationcraft Results

This is also using 5.0.5 spell data. The profile now looks good to me and even gives reasonable hit/exp scalefactors. If you find anything odd, please mention it.

p.s. Unfortunately, the latest simc release 5.05-2 does not include the fix for SB (but anything else). Still, you should play around with it a bit.

Would it be possible to add a timeline for vengeance?

Berthold · 09/15/12, 8:51 AM

Fixed TC issues. This gave hit/exp yet another boost, now making stats
Hit== Exp == Mastery > Dodge/Parry
Dodge and parry scalefactors depend heavily on the gear setup due to DR.

Simulationcraft Results

With the next release, we can focus on getting a BiS list together, assuming that minimal dtps is our goal, as #deaths heavily depends on the kind of damage we take and heal support we have.
Although double elixirs are going to be the better choice dtps-wise, for now I sticked with the stamina flask as the other choice gets... expensive.

Glurak · 09/16/12, 11:37 PM

Have some Stat wheigts for t13 Fury now ? Scales Crit now more than strengh ? So i can Gem Only Crit and miss the Socket bonuses without dps loss ? I have at the moment Str/Crit mixed gems and Crit/hit gem.
so Ho are the Scale factors now ?

and one big question that i have is how important is Hit after 7,5 % hit

Berthold · 09/17/12, 4:28 AM

reighnamn, it is possible, the question is more, does it warrant the effort? It's going to be a rather boring curve that goes up for the first 20 secs and basically flatlines on that level.

p.s. challenge mode scaling is in the works. This should gives the right tool to see how we want to gear for them. I am expecting something like Profiler - Wowhead to be close to optimal. (gems/enchants/reforges arent' final)

Krennick · 09/18/12, 4:53 AM

Regarding Vengeance timeline.

Vengeance is no longer capped, so it will not a case of a curve going up for 20 seconds and then staying at one level. It will rise and fall with special ability use. For instance on heroic madness I've noticed my vengeance go above 180,000 AP (and shield barrier absorb above 110,000 per 20 rage as a result).

If simming with a boss doing regular dmg without spikes, a vengeance timeline is indeed pretty flat (post-20 seconds) - but all bosses have some spikey elements - and during spikes shield barrier increases dramatically in value. In some cases beating out shield block. Morchok heroic stomp is a time for shield block ofc - Yor'Sahj after 3 debuffs is shield barrier. But there may be times (impale on madness is arguably one) where the vengeance climbs to a point where you would use shield barrier over shield block even on blockable physical dmg.

It just underlines the challenge in simming for tanks rather than dps. For dps you mainly care about the player - for tanking you care a lot about the boss and what it does.

Berthold · 09/18/12, 12:15 PM

Considering spiky damage, the vengeance timeline will indeed look a bit more interesting. And I agree that we need to take a close look at what the boss does to figure out how we should user our abilities.

But for finding out what works best, we don't need to know the vengeance timeline:

If we model a boss that nukes with X every so often, we can just simulate the use of Shield Barrier or Block and what works best. Additionally, given the amount of rage we gain in T14H gear, we can easily time the use of BOTH for spike damage: The gear I put up uses SBlock on CD and additionally can fit in a 60 rage Shield barrier every 30 seconds. (impale, e.g., is 35 seconds)

This might sound like a lame excuse for me not building the timeline, but I'd rather spend my time on correct models for every T14H boss than draw an interesting but not really decision-making timeline.

But I will put it up as a suggestion for the fellow simcraft developers, maybe somebody else will pick it up.

Krennick · 09/18/12, 4:28 PM

I agree that correct models for T14 bosses is more valuable - but the obvious corrolary is that these models are used to decide tank ability usage. Tank ability usage should never be a simple priority list, but rather poll situational factors (including vengeance and rage ofc, but also considering expected boss spikes in the imminent future) and based on those decide which ability to use. This particularly applies to cooldown usage (shield wall, last stand etc) but also very much to whether to use shield block or shield barrier.

What I'm trying to say is that every step taken towards simulating tanking in SimulationCraft is appreciated, but I am not sure there are few and simple steps to go before it becomes a truly meaningful picture.

Having said that, I haven't really spent enough time with SimulationCraft even when dpsing and must admit I have some trouble reading the tanking results posted. I would love to understand why it is set up in the way that it is set up, how to set SimulationCraft up in a similar way myself and how to read the outputs. I'll maybe spend a little time this weekend trying to figure it out, but if there's a ready explanation for how to use SimulationCraft for tanking I'd appreciate a reference to it.

Berthold · 09/19/12, 1:25 PM

krennick, I am in the process of updating the starter guide at
https://code.google.com/p/simulation.../StartersGuide
to also talk about tank stuff. Should be finished by the weekend.

reighnman · 09/19/12, 4:03 PM

Originally Posted by Berthold

Considering spiky damage, the vengeance timeline will indeed look a bit more interesting. And I agree that we need to take a close look at what the boss does to figure out how we should user our abilities.

But for finding out what works best, we don't need to know the vengeance timeline:

If we model a boss that nukes with X every so often, we can just simulate the use of Shield Barrier or Block and what works best. Additionally, given the amount of rage we gain in T14H gear, we can easily time the use of BOTH for spike damage: The gear I put up uses SBlock on CD and additionally can fit in a 60 rage Shield barrier every 30 seconds. (impale, e.g., is 35 seconds)

This might sound like a lame excuse for me not building the timeline, but I'd rather spend my time on correct models for every T14H boss than draw an interesting but not really decision-making timeline.

But I will put it up as a suggestion for the fellow simcraft developers, maybe somebody else will pick it up.

I would appreciate it even as a feature request. I may have under estimated the amount of effort it would require to implement, thought it might be something you could just slap on quick.

Berthold · 09/20/12, 4:06 AM

Back to the Shield Barrier vs Block debate. I am seeing strange results in the sims (Barrier way OP), but even my napkin math can't disprove them. Can somebody at least point me to the flaw of the latter?

First: Shield Block.
with Bis gear we get a full hit ~%30 of the time and have 70% cb. This means that if we use Shield Block all hits will be turned in to block and 70% of those blocks will be critical blocks: 9% * 30% + 21%* 60 % = 2.7 % + 12.6% = 15.3% dmg reduction. If we take 3 hits over the course of 6 seconds, this is a reduction by ~45 % of one white hit.

Second; Shield Barrier:
Boss does hit for X every 2 seconds, i.e., 10 times over 20 seconds. This means our vengeance will be at 2% of the damage taken over the course of this time, i.e., AP= 10*X*0.02%=0.2*X
Shield Barrier gives us max(2*AP(minus AP from STR), 2.5*Stamina), i.e. 0.4*X at 60 rage. However, if the boss hits for X, we will mitigate it via defstance/armor down to ~ 0.4X. This means that with a 60 rage Shield Barrier we absorb 100% of one hit.

Sims are saying the same. So please, tell me that I am the one who is stupid or else the whole barrier/block discussion is pointless as shield block is then only going to be used in situations where our vengeance level is not appropriate for the incoming damage, i.e., at the beginning of the fight or to reduce rare spikes.

p.s. it looks very similar for other gear levels, e.g., 463 it is 56% of a white hit for Shield Block
p.p.s. Vengeance timeline/buff modeling was added to the feature Requests.

Malevolencia · 09/20/12, 10:00 AM

Originally Posted by Berthold

Back to the Shield Barrier vs Block debate. I am seeing strange results in the sims (Barrier way OP), but even my napkin math can't disprove them. Can somebody at least point me to the flaw of the latter?

First: Shield Block.
with Bis gear we get a full hit ~%30 of the time and have 70% cb. This means that if we use Shield Block all hits will be turned in to block and 70% of those blocks will be critical blocks: 9% * 30% + 21%* 60 % = 2.7 % + 12.6% = 15.3% dmg reduction. If we take 3 hits over the course of 6 seconds, this is a reduction by ~45 % of one white hit.

Second; Shield Barrier:
Boss does hit for X every 2 seconds, i.e., 10 times over 20 seconds. This means our vengeance will be at 2% of the damage taken over the course of this time, i.e., AP= 10*X*0.02%=0.2*X
Shield Barrier gives us max(2*AP(minus AP from STR), 2.5*Stamina), i.e. 0.4*X at 60 rage. However, if the boss hits for X, we will mitigate it via defstance/armor down to ~ 0.4X. This means that with a 60 rage Shield Barrier we absorb 100% of one hit.

Sims are saying the same. So please, tell me that I am the one who is stupid or else the whole barrier/block discussion is pointless as shield block is then only going to be used in situations where our vengeance level is not appropriate for the incoming damage, i.e., at the beginning of the fight or to reduce rare spikes.

p.s. it looks very similar for other gear levels, e.g., 463 it is 56% of a white hit for Shield Block
p.p.s. Vengeance timeline/buff modeling was added to the feature Requests.

I think you're comparing very slightly different things in the napkin math, you're comparing the best case for shield barrier (it absorbs the full value of the shield - admittedly this should happen very frequently, and in cases when it isn't reached shield block will always be worse than it anyway) to the average case performance of shield block. The best case performance of shield block is that it converts what would of been 3 full hits into 3 critical blocks, which would mean it reduces damage by 3 * 60% = 180% of a white hit in the best case. Average of shield barrier, should be less than 100% but I imagine still quite high up there..

So that won't change things much, and I do agree that it would seem likely that shield barrier might be outperforming shield block in normal circumstances (after the first few seconds of a fight). But bosses that do special extra big hits infrequently should still probably promote the usage of block over barrier, such as Gara'jal and his Shadowy Attacks, or Sha of Anger and his Thrash (well, probably you want both of them up for it rather than only one).

Is it really true we have only ~30% chance of a full hit in BiS from the first tier? On live, in full heroic DS gear (admittedly reforged for hit/exp with mastery after that, and with 2 str trinkets) I have only 62.86% combined dodge+parry+block unbuffed. Iirc the 5% chance to miss for bosses was removed or reduced (can't remember) too.. so having only a 30% chance of taking a full hit sounds really low to me...

Edit: hmm I added dodge+parry+block from habit.. but the block chance (27.32% for me) should actually be a % of the hits that aren't dodged/parried now instead to calculate the actual amount of full hits you take shouldn't it.. So against a mob who hits me 100% of the time if I am naked, I dodge+parry 35.54% of the time, so he hits 64.46% of the time, and then I block 27.32% of those or 17.59% of all hits. And thus my total coverage is only: 53.13%... Or does the block % listed ingame/on armory already take account of that and you can still just directly add it to dodge+parry?