Enhance Shaman: The Collected Works of Theorycraft, Vol I - Page 89

Tornhoof · 09/06/07, 7:01 AM

What I noticed in Yo's source is that he multiplies the WF AP with the UR factor.
I'm not completely sure that this is correct, as far as I know, the WF AP is just a static value added to the windfury swing, always 475 AP. Back when I wrote my alpha windfury plugin I found no evidence that wf ap is increased by UR, but I never tested explicitely for this, 47.5 AP (UR gain) are obviously not that simple to verify.

Nemaa · 09/06/07, 7:59 AM

Can we get a support for [Totem of the Astral Winds] as well?

Neithan · 09/06/07, 9:33 AM

Hello guys,

Just one question. Are BT LW haste mail pieces (bracer and shoulder) still worth getting after the coeficient change?

Edit: Just woke up, edited for clarity

Nemaa · 09/06/07, 9:54 AM

Originally Posted by Neithan

Hello guys,

Just one question. Are BT LW haste mail pieces (bracer and shoulder) still worth getting after the coeficient change?

Edit: Just woke up, edited for clarity

Shoulders of the Hidden Predator or Mantle of Darkness (and I think even the T6 shoulders) will be better after the haste nerf. Same goes to Insidious Bands, Deadly Cuffs and Wraps of Precise Flight for the wrist slot. I don't think I'll ever use haste items after 2.2

Nemaa · 09/06/07, 10:04 AM

Yo: Your AEP calculator is really good, this is what I was thinking about in one of my previous posts. You said "It is assumed that BoK is on" so when I enter stats I should sum all buffs/potions/food/etc except unleashed rage and blessing of kings, am I right?

Wizkid · 09/06/07, 10:24 AM

1) In my humble opinion the weapon section in the first post is still a bit lacking. There really arent that many epic options for us (less than trinkets and those are all listed) and you are only listing the 100+ DPS weapons for MTH/BT. Most players will (want to) start with a MH weapon from Kara or the BS option and try to get one of the MH upgrades from TK or even another Arena2 for the MH.

2) I recently switched from Decapitator (Mongoose) + Arena2 (Crusader) to Arena2 (Crusader) + Netherbane (Mongoose) and am wondering about two things:

First, with two different proc enchants it should not matter if they are on the MH or the OH weapon, or does it ? Maybe add the answer to the Enchanting section.

Second, with the upcoming buff to SSC/TK weapons it seems rather pointless to replace Netherbane or Talon of the Phoenix with a second Arena2 weapon ? The Lootzor link already lists the Talon as the best MH even above the 100 DPS drops from MTH/BT, which frankly irritates me a bit (yes, I noticed the proc on the Syphon). Should that be fixed or is it true ?

3) Which weapon does an Enh Shaman pick up on Kael ? The dagger although it is rather fast or the 2h axe because it has the speed proc ? Or just your regular weapon set, because it is slow/slow aka optimal ?

Stigmata · 09/06/07, 10:30 AM

Originally Posted by Wizkid

1) In my humble opinion the weapon section in the first post is still a bit lacking. There really arent that many epic options for us (less than trinkets and those are all listed) and you are only listing the 100+ DPS weapons for MTH/BT. Most players will (want to) start with a MH weapon from Kara or the BS option and try to get one of the MH upgrades from TK or even another Arena2 for the MH.

Do people really need a complete list of every possible item they could ever get, the original post is a guide, not a complete instruction manual.

Originally Posted by Wizkid

3) Which weapon does an Enh Shaman pick up on Kael ? The dagger although it is rather fast or the 2h axe because it has the speed proc ? Or just your regular weapon set, because it is slow/slow aka optimal ?

None.

Yo! · 09/06/07, 10:32 AM

Originally Posted by Disquette

Ok, going to pull out interesting lines from the code as I see them...

Why a 32% chance to proc windfury on an attack instead of the tested and empirically determined 36% chance?

How is MissMH determined, and why does it have a different rate than the OH miss rate? I see you did something similar with crit. Is this for Orcs or other people who have +skill ratings who use non-similar weapon types between hands?

*** editing out my WF dmg thing - this is referred to later on, but I can see where I made a mistake in analysis ***

the following stuff I still don't get, however:

I assume you mean to have WFap/14 in the "hit" statement for the MHWf hit case.

Also, now that I think of it, is Windfury Actually hitting twice? Again, I'm pretty good at being dumb, so I hope this isn't s stupid question, but how do you have the code proc'ing 2 windfury attacks? I'm only seeing one. I'm not so familiar with the array coding, however, so maybe you take care of it somewhere in there.

I assume that when you say that adding Hit wasn't capped, the result in the sim is that you end up pushing dodge off of the table? I'm not sure how you determine MissOH and MissMH, so I can't be sure. I still don't get why having 60% Hit was better than 50% hit, but as long as you say you're fixing it, I'm happy with that.

Also, I'm not sure what's happening here:

This part in particular:
if (R1>(0.05+Dodge))
{ // landed

is the 0.05 supposed to be the default miss rate? If so, then why don't you have +hit reducing that 0.05 result off the table? I know you mentioned a two-roll system for yellow attacks, but I don't see that anywhere in here - Is this somehow related to that?

Other than those things, I don't see any glaring mistakes (either on your part in the code, or mine in understanding it). Not to say that these mistakes might not explain the huge dps gaps seen - If you're not granting the WF AP bonus to MH WF hits, and if you only have WF hitting once instead of twice, that could easily explain the 200dps delta between your and tornhoof's models. But, if these are the only problems with it...

Then this, in fact, is by far the cleanest code I've seen, and makes the step by step versions (like mine

) look really ugly in comparison. Your method of finding the minimum value of the "next event" is so much more efficient - now I'm starting to understand why your sim runs so much quicker than mine ever could as I check every x milliseconds for an event.

I'm really impressed man :-) The only thing that I haven't mentioned so far that I'd like to see, that I haven't requested yet, is more detailed reporting - how many main hand hits were there, for how much damage. How many off hand glances, etc etc. That sort of thing really helps us compare the code results against empircal or theoretical examination.

You spotted alot, thank you! The only part not being a bug is second attack from WF - it is covered by this code:

if(WF){WF=false;}
else {Time[1]=inf;}

WF is boolean variable that shows if seond attack needed

Also Miss is different for MH and OH because of weapon skill.

MissMH = 0.24 + Miss(MHskill)-Hit

now I changed it to

MissMH = Math.max(0.24 + Miss(MHskill)-Hit,0)

What I noticed in Yo's source is that he multiplies the WF AP with the UR factor.
I'm not completely sure that this is correct, as far as I know, the WF AP is just a static value added to the windfury swing, always 475 AP. Back when I wrote my alpha windfury plugin I found no evidence that wf ap is increased by UR, but I never tested explicitely for this, 47.5 AP (UR gain) are obviously not that simple to verify.

Will fix it. Thanks.

Did you miss Weapon Mastery?, I can't find any 1.1x factor to the dps

Yes I did

Can we get a support for Totem of the Astral Winds as well?
Your AEP calculator is really good, this is what I was thinking about in one of my previous posts. You said "It is assumed that BoK is on" so when I enter stats I should sum all buffs/potions/food/etc except unleashed rage and blessing of kings, am I right?

1) Totem of the Astral Winds is on by default - yet another checkbox to be able to switch it on/off is needed.
2) Sum it all up with BoK but without UR. UR is Shaman's buff with random behavior that is modelled.

I'd recommend you not require people to hand-jam BoK onto their stats, adding it as an optional buff would be great. Ideally people want to see a sim match up closely with their real world DPS so they can see that its accurate, and for people in 5mans or KZ, they're not going to BoK anyways. Hell, I don't get Kings all the time in a raid even

Basically the fact that BoK is "on" affects only AEP values of Agi and Str by flat 10% increase.
Put your stat totals buffed just as how you are going to be buffed in raid (with or without BoK) and divide AEP values for Agi and Str that it finds by 1.1.

Malan · 09/06/07, 10:37 AM

Originally Posted by Wizkid

First, with two different proc enchants it should not matter if they are on the MH or the OH weapon, or does it ? Maybe add the answer to the Enchanting section.

Second, with the upcoming buff to SSC/TK weapons it seems rather pointless to replace Netherbane or Talon of the Phoenix with a second Arena2 weapon ? The Lootzor link already lists the Talon as the best MH even above the 100 DPS drops from MTH/BT, which frankly irritates me a bit (yes, I noticed the proc on the Syphon). Should that be fixed or is it true ?

3) Which weapon does an Enh Shaman pick up on Kael ? The dagger although it is rather fast or the 2h axe because it has the speed proc ? Or just your regular weapon set, because it is slow/slow aka optimal ?

1 - Doesn't matter which hand.

2 - The lootzor link for weapons is garbage, don't refer to that. I'm not sure about the Talon vs Gladiator, I ran Pater's sim using both with my gear and the Talon did *not* perform as highly as the Gladiator weapons, which was pretty disappointing to me. However, I'll probably try to pick one up next time it drops just to try the combo.

3 - Stigmata has elaborated many times in this thread that he grabs the 2H, uses the proc, and then uses his normal weapons.

@Yo - I'd recommend you not require people to hand-jam BoK onto their stats, adding it as an optional buff would be great. Ideally people want to see a sim match up closely with their real world DPS so they can see that its accurate, and for people in 5mans or KZ, they're not going to BoK anyways. Hell, I don't get Kings all the time in a raid even.

Xoya · 09/06/07, 10:49 AM

For what it's worth, on Kael this week I was toying around with using the dagger in the mainhand. It has approximately the same range as my arena season 2 weapon but much higher dps. The windfuries were slightly lower (1.8k to 1.9k compared to 2.1k to 2.2k) but I'm wondering if the extra dps from the fast weapon (combined with the fact you'll actually get more windfuries from the fast weapon on the maindhand) makes up for it.

I do remember the long discussion that was had back in June or so when Stigmata discussed all this, but I don't remember whether he ever tried it in the mainhand.

Tornhoof · 09/06/07, 11:01 AM

@Yo:
Did you miss Weapon Mastery?, I can't find any 1.1x factor to the dps

Malan · 09/06/07, 11:06 AM

I'm half wondering if it might be time to start a Volume II of this thread.

Stigmata · 09/06/07, 11:33 AM

Originally Posted by Xoya

I do remember the long discussion that was had back in June or so when Stigmata discussed all this, but I don't remember whether he ever tried it in the mainhand.

I can't remember the last time I entered TK, let alone fought Kael, i certainly never picked up the dagger, so there would be no chance of me using it MH.

Tornhoof · 09/06/07, 1:05 PM

Hmm I compared Yo's values to my most recent model and I'm a bit surprised, even after reducing some caps to his values (miss to be 24% etc.)

Stormstrike DPS is fine, even though we're using different methods for SS, I can reproduce his values (e.g. ~120 DPS for the default values)

WhiteDPS is a tad on the lower end for his model, even if I multiply the missing Weapon mastery (which I think is missing) to his value. His result is 517 DPS, mine is 628 DPS.
We're missing some 50 DPS there.
Now my white mh dps is calculated:
float mhwhitedps = 0.01f * (hitwhite + crit - glancing - dodge) * (wdps1 + ap / 14);
which is straightforward.
I'm using 25% glancing probability and 25% glancing reduction.
This whitedps is further multiplied with haste effects, common factor is 1.25 (flurry uptime around 70% here + 5% haste).
This formula is certainly correct, but 50 DPS can not be just explained by bad luck.

YellowDPS is obviously carrying those missing DPS from the whitedps over to yellow dps. I'm missing around 50 DPS there again.
Similar formula as above, again hasted:
float mhyellowdps = 0.01f * 0.01f * (hitwhite - dodge) * (hityellow + crit - dodge) * 2 * wfprocmh * 1.4f * (wdps1 + wfap / 14);
(we can only proc wf if we actually hit the target, so we need to hit with white first, than we can crit/hit with yellow

)
This value is again hasted, the wfprocmh probability is corrected to cover the 3.0s cooldown:

Example: two 2.6s weapons, 5% haste, we assume the 25% overall haste effect (incl. flurry)
Each weapon swings with 2.08s now (on average), so we swing every 1.04s.
With a wf proc chance of 36% we expect our wf to proc every 2.08/0.36 = 5.777s
So even hasted we are way slower than the wf cooldown of 3.0s.

Basically I'm missing 100 DPS, 50 DPS in white and 50 DPS in yellow, probably the missing yellow dps is already accounted for by Disquette's bugsearch. So I'm pretty much still missing a fair amount of normal white DPS.

The current model in c++ is here: Pastey.net - paste bin

Disquette · 09/06/07, 1:15 PM

Tornhoof, there's something that I don't like talking about because it reveals a fundamental lacking of my part, but perhaps it will help figure out the white dps difference. Your closed form value for # of flurried attacks is a lot closer to the game-tested values than the ones produced in a sim. I can't for the life of me figure out why. The sims use the logic "haste the next x attacks", and when you put 3 in for x, you get lower flurry uptime (in my experience) than using the closed form (1-crit)^3

Maybe it's a bug in coding the sims? Maybe it's a fundamental lack of understanding on my part? I don't know. Once the flurry uptime portion is operational on Yo!'s page, and assuming it's attack based instead of time based, we can look into it.

Yo! · 09/06/07, 1:47 PM

Originally Posted by Disquette

Tornhoof, there's something that I don't like talking about because it reveals a fundamental lacking of my part, but perhaps it will help figure out the white dps difference. Your closed form value for # of flurried attacks is a lot closer to the game-tested values than the ones produced in a sim. I can't for the life of me figure out why. The sims use the logic "haste the next x attacks", and when you put 3 in for x, you get lower flurry uptime (in my experience) than using the closed form (1-crit)^3

Maybe it's a bug in coding the sims? Maybe it's a fundamental lack of understanding on my part? I don't know. Once the flurry uptime portion is operational on Yo!'s page, and assuming it's attack based instead of time based, we can look into it.

Flurry is something that I can not code fully because I am not convinced on how it has to work... So far it is "haste the next x attacks".

New version of sim is released with basic FAQ and changelog.
Crazy Shaman's DPS & AEP calculator (c) Yo

Now about the difference with math expactation approach:
The main fundamental reason is that math expactation for function of some random values is not equal to function of math expactation of these values.
Example - on crit you have both UR and Flurry activated leading to all flurried attacks being made under effect of UR - something that math expactation approach is not counting as Flurry is assumed to be x uptime, UR y uptime so there is possibility when Flurried attacks are not affected by UR.
Sims potentially are more correct while formulas are very fast. For AEP calculation derivatives can be found out from formula while sims have to use brute-force approach: increase something by z, calculate dps and divide result by z afterwards and the accuracy/time depends on z choice greatly.

Nite_Moogle · 09/06/07, 1:47 PM

Originally Posted by Disquette

Tornhoof, there's something that I don't like talking about because it reveals a fundamental lacking of my part, but perhaps it will help figure out the white dps difference. Your closed form value for # of flurried attacks is a lot closer to the game-tested values than the ones produced in a sim. I can't for the life of me figure out why. The sims use the logic "haste the next x attacks", and when you put 3 in for x, you get lower flurry uptime (in my experience) than using the closed form (1-crit)^3

Maybe it's a bug in coding the sims? Maybe it's a fundamental lack of understanding on my part? I don't know. Once the flurry uptime portion is operational on Yo!'s page, and assuming it's attack based instead of time based, we can look into it.

Are you accounting for Windfury procs and Stormstrike generating flurry charges but not consuming them? I am guessing that you are but this is the easiest way to get way out of whack on uptime.

Tornhoof · 09/06/07, 2:03 PM

Originally Posted by Disquette

Tornhoof, there's something that I don't like talking about because it reveals a fundamental lacking of my part, but perhaps it will help figure out the white dps difference. Your closed form value for # of flurried attacks is a lot closer to the game-tested values than the ones produced in a sim. I can't for the life of me figure out why. The sims use the logic "haste the next x attacks", and when you put 3 in for x, you get lower flurry uptime (in my experience) than using the closed form (1-crit)^3

Maybe it's a bug in coding the sims? Maybe it's a fundamental lack of understanding on my part? I don't know. Once the flurry uptime portion is operational on Yo!'s page, and assuming it's attack based instead of time based, we can look into it.

Hmm I have no idea, as far as I've seen Yo's sim has a working and correct flurry calculation.

On a side not, I'm not even considering a full stormstrike cycle in my model (no haste effect for wf currently), so it's probably even missing some more dps

As for my program, It's on hold until atleast 2.2 and probably until I port it on c++, since I'm running into serious performance issues currently.

Shabadu · 09/06/07, 2:17 PM

Originally Posted by Malan

I'm half wondering if it might be time to start a Volume II of this thread.

Make the next one The Enhancement Shaman Omnibus: SLOW WEAPONS WF ON BOTH

Tornhoof · 09/06/07, 2:18 PM

As for Flurry works:
- Every melee attack can proc it (e.g. white, stormstrike, wf)
- only white hits consume a charge

So you'll need to change your model to reset your flurry counter to 3 for every crit in your switch case (since every melee attack can crit and proc flurry) and only decrease the counter for a successful white melee hit.

Yo! · 09/06/07, 2:26 PM

Originally Posted by Tornhoof

As for Flurry works:
- Every melee attack can proc it (e.g. white, stormstrike, wf)
- only white hits consume a charge

So you'll need to change your model to reset your flurry counter to 3 for every crit in your switch case (since every melee attack can crit and proc flurry) and only decrease the counter for a successful white melee hit.

It was done exactly this way from the start as can be seen in code above - whenever any attack crits Flurry = 3, and Flurry-=1 only on hasted white hits.

What I do no model is the effect of Flurry affecting more than 3 attacks sometimes for some unclear reason.

MrProcrastin8 · 09/06/07, 6:31 PM

I know, I know. Daggers make us all cringe. I understand how the calculations work of how we want to have the slowest weapons possible. But I find myself in a pickle as of last night.

First of all, I'm very much so undergeared. Like, pre-Kara undergeared. Time constraints, as well as helping the girlfriend level, has hindered my ability to run the dungeons to get the good gear. However, I found some time last night to run Karazhan with my guild. They have been running it for months now, so the group I was running with had a good bit of the gear that drops there already, so I ended up getting some easy loot:

Emerald Ripper
Blade of the Unrequited

They are fast items, yes, but, damage-wise, they are arguably better than my previous two:

Reflex Blades
Mag'hari Fury Brand

My question is probably one that gets asked often... will using the daggers give me an overall dps increase? When I happen to obtain slower, higher damage items, obviously I will make the switch, but I don't know when I will be able to obtain one, and it may be a long time coming. So, until then, what do I do?

Armory Link for if my crit/ap/whatever is needed to help me figure this out.

Thanks.

Malan · 09/06/07, 6:34 PM

Actually Strawberry (I think?) asked about this a few pages back with the Reflex blades, and after testing he found exactly what I told him - that the Reflex in MH and an AH green 2.6 was better than either of the epic daggers. I'd hang onto them for whenever those shaman talent changes come out just in case though.

Ekto · 09/06/07, 6:35 PM

Originally Posted by Malan

Here's a more updated version of the math that I linked on the first post from pre-TBC. I just took the equations used and substituted in values appropriate for level 70.

Dexterity Enchant = 12 Agi = 12 x 1.8 = 21.6 AEP = ~1.6 DPS (if 12.73 AP = 1 DPS because of Weapon Mastery)
Length of fight = 6min = 360 seconds
Shaman DPS = 800 (fairly achievable value)

Contribution of Dexterity enchant to damage in a 360 sec fight
360sec * 1.6dps = 576
The run speed enchant would need to make up 576 damage by allowing you to reach your target faster.

576 = 800*[x-x/1.08]
576/800 = x - x/1.08
576/800 = x(1-1/1.08)
Simplifies to x = 9.72

So at level 70, for a 6min fight where your average DPS is 800, a run speed enchant beats out an agility enchant if the fight forces you to move between targets for more than 9.7 seconds. Note that the math above doesn't even account for the 6 Agility on the Cat Swiftness enchant, which would further this as such:

If the DPS contribution of the Dexterity enchant is 576 damage over 360 seconds, than the Agility from the Cat Swiftness enchant (0.8 DPS) would be worth 288 damage in the same fight. Therefore you really only need to beat out 576 - 288 = 288 damage over the fight.
This would lead to the equations above simplifying to x = 4.85, so you really only need to have less than 5 seconds of movement on a 6 minute fight for the Cat Swiftness enchant to be come superior to the Dexterity enchant.

this is completely wrong. in your first equation, you set the wrong quantities equal to each other. you set the damage done by dexterity equal to the damage GAINED by the minor run speed, NOT the actual damage done. hence your incorrect conclusion. the following is the correct derivation...

A = dps offered by 1 agility
T = total dps time for a boss fight (this does not include times where melee must stop dps, like running out of RoF on azgalor, but it DOES include the time you spend outside of melee range when there is no fire)
t = total time spent outside of melee range, during which you COULD have been dpsing (see above example)

over the course of the fight, dexterity will be responsible for the following amount of damage...

12 A (T - t)

over the course of the fight, cats swiftness (6agi+minor speed) will be responsible for the following amount of damage...

6 A (T - t') = 6 A (T - t/1.08)

so lets suppose that cats swiftness is better than dexterity and see what conditions T and t must satisfy for that to be true...

6A(T-t/1.08) > 12A(T-t)

...now you cancel out the A (so it really doesn't matter how much dps 1agi gives, just the assumption that 1agi gives a static amount of dps), and do some algebra to obtain...

T < (2-1/1.08)t ===> T < 1.074 t

so cats swiftness is better than dexterity if and only if the total time of the fight is LESS THAN 1.074 times the time you would normally spend outside of melee range. clearly, this condition is never satisfied on a normal boss fight. for example, on a 5 minute fight, you would have had to spend 4 minutes and 39 seconds outside of melee range for cats swiftness to be worthwhile.

please correct this on the front page for those thinking that minor run speed is worth getting.

Ilmater · 09/06/07, 6:47 PM

Originally Posted by Boratz

This also lends itself to people in my guild not necessarily believing that Enhancement Shaman dps is viable, considering my spots on the meters at present. My armory link should work if anyone would like to have a look (Bloodlust brooch incoming within a week or two).

Leather, leather, leather. Period. The armor mail gives you is minimally better than leather, and not worth it. If you were resto or elemental, then your shield makes it worth sticking with mail, but as enhancement, you don't need the INT or MP5 that your gear gives you, so go with leather in almost EVERY situation.

And I agree with what others said... there is no one stat to increase, just get items with the most AEP.