I've redone some simulation, and yes sola5ide told the true, so I've changed the gear into simulation to mach the Best in Slot for 3.1 (without Ulduar gear). It was my bad that I didn't consider 10-men gear (mainly because I don't run them) so I didn't take into account those 2 ring, without those two ring Dawnwalkers are better than Vile Deceit (they are better on a general basis but with Dawnwalkers you will overcap hit if you take those two ring). As for Wristwraps of the Cutthroat vs. Sinner's Binding, it depends on your ArPen rating, regemming everything with ArPen (exception made for 1 8hit-12 stam gem to mach metasocket requirment) WoC becames better. So sola5ide is completly right. I've finished redoing and updating my simulator with the latest PTR push and I'm now running new sims, I hope I'll update the main post in about 1 hour.

The pre-Ulduar BiS gear (with 3.1 rules) is:

Slot	Item
Head	Valorous Dreamwalker Headguard
Neck	Favor of the Dragon Queen
Shoulders	Valorous Dreamwalker Shoulderpads
Chest	Chestguard of the Recluse
Waist	Stalk-Skin Belt
Legs	Leggings of the Honored
Feet	Footwraps of Vile Deceit
Wrist	Wristwraps of the Cutthroat
Hands	Frosted Adroit Handguards
Finger1	Surge Needle Ring
Finger2	Greatring of Collision
Trinket1	Grim Toll
Trinket2	Darkmoon Card: Greatness (agility)
Back	Drape of the Deadly Foe
MainHand	Journey's End + berserking
Idol	Idol of the Ravenous Beast
ExtraWaistSocket	Fractured Scarlet Ruby

Everything socketed with 16 Arpen + one 8 hit/12 stam and 21 agi +3% crit meta.

Would using a +6all stats pearl not be the better choice for meta requirements?
Probably into the legs as it has the best bonus for a blue slot (6 agi on legs vs 8ap on neck)

I find this very surprising. I've been searching around and brainstorming on why this would suddenly be beating out the Idol of Worship. Primal Gore would only improve increasing the baseline of Rip. Unless the armor-pen change impacts Shred more than Rip somehow. I just don't see how.

You answered your own question basically. If you go back to Nightcrowler's simulation and read the preference of stats, Armor Penetration was exceptionally close to Agility, and that was before the removal of Rake from Primal Gore. That one change makes it so a bit more of our total DPS is from direct attacks and a bit less from DoTs which would boost the relative power of Armor Penetration even more. Ravenous Beast made Armor Penetration preferential from the start and now I see no reason to stress over the Rip Idol any longer.

I was hoping that I could prepare a bit better for 3.1 by already having Agility in most of my slots since it isn't the worst stat for Live but would be the best choice going into Ulduar. It appears I'll be fully regemming every piece of gear I have for tanking and DPS now, with very few exceptions.

Ah, thank you Fasc. I assumed that armor pen impacted our bleeds as well, and clearly was very wrong.

At least for cat the main page is updated. Just for a note, I'm using the new rules for armor reduction and armor penetration (you can find the link to the post in the main page) but it seems to be bugged. At the moment with the bugged version is our best stat, if the "remove the bug" it will be a "no brain stats" going from about 2.8 AEP to 3.5 AEP (I'm looking at SimCraft last PTR report and both dps and stats value are "converging").

It was fixed in 3.0.8.

I was wondering how the armor ignore mechanics worked...

Sunders + FFF will negate 45% of a bosses armor and with the Grim Toll proc being 51% would that not generate a lot of wasted stats from gemming ArP when the proc + raid buffs makes you ignore close to 100% of armor anyway?

With my current gear + Raid buffs I ignore 102% of boss armor without gemming with the Grim proc.

Sunders+FFF are 24% (4% for each Sunder stack additive =20% plus 5% moltiplicative for FFF = 24%). Also armor/arpen formula works in a strange way atm on ptr (bugged?) basically if you need about 102.5% arpen from items plus sunder and FFF to completly negate boss Armor.

Quoting from main post page:

In my simulation I'm using best in slot gear and I have gemmed everything with ArPen,given that, at least without Ulduar gear Rake seems still better than shred from a DPE point of view. Yes rake will no longer be "pressing" but it's still better than shred.

If you suppose that you have enough ArPen to completly nullify boss armor, with pre-ulduar BiS gear you will have:

Shred: 10780 avg damage
Rake: 8335 avg damage

Given that Shred energy cost is 42 and rake energy cost is 35 you will have:

Shred: 256.7 avg DPE
Rake: 238.1 avg DPE

For shred to be better DPE than rake you need: 1-8335*(42/35)/10780 = 7.2% reduction from armor or (given the actual "bugged" PTR formula) 995 ArPen Rating +5 Sunder + FFF.

You should also remember that:
- Rake cost less Energy than shred, that means that also if it's a little less energy efficient you will gain a quicker cycle so less downtime.
- Rake is a bleed and you need bleed for shred max damage (it shouldn't be a problem for 25 men).
- 2T8 proc clearcasting with dots.

Now your questions:

1) You should follow this path to gear up: 2T7 (head and shoulders) -> 2T7 (Gloves+Legs) +2T8 (Head+Shoulders) -> 4T8 (all tier exception made for the chest)
2) Using rake and gemming for agility are 2 different things, you should gem for ArPen ALSO if you use rake.
3) As you can see before rake is likely to stay in our rotation also at high ArPen value.

EDIT: I've runned some simulation with BiS Ulduar gear (avaiable in PTR), you can see the results on main page. From my simulation you will continue to use rake also in full Ulduar Gear.

P.S. About scaling.

If you look at the dps upgrade from best Naxx and best Ulduar gear it's about 7.3% it seems a bit low and I start thinking that our too high dps (just look at simcraft if you want) is something that it's there to keep us where we should be in full Ulduar gear (basically our old scaling problem).

Could you link the this gear list, like you did earlier with pre ulduar gear. thanks


-fart

X Esari: you can find the Ulduar gear list on the main page, but I'll post it here for simplicity:

Slot	Item
Head	Conqueror's Nightsong Headguard
Neck	FBroach of the Wailing Night
Shoulders	Conqueror's Nightsong Shoulderpads
Chest	Tunic of the Limber Stalker
Waist	Relic Hunter's Cord
Legs	Conqueror's Nightsong Legguards
Feet	Runed Ironhide Boots
Wrist	Mechanist's Bindings
Hands	Conqueror's Nightsong Handgrips
Finger1	Loop of the Agile
Finger2	Metallic Loop of the Sufferer
Trinket1	Grim Toll
Trinket2	Darkmoon Card: Greatness (agility)
Back	Drape of Icy Intent
MainHand	Lotrafen, Spear of the Damned + berserking
Idol	Idol of the Ravenous Beast
ExtraWaistSocket	Fractured Scarlet Ruby

All socketed with +16 ArPen + one 8 exp/12 stam + two 8 hit / 8 agi.


X Allev: obviosly the rake/shred threashold depends also on crit, basically it dependes on:

- Armor Penetration + debuffs (FFF + Sunders) (scales with shred)
- Crit chance (scales with shred)
- AP (scales better with shred if you consider crit/arpen interactions)
- 2T8 bonus (obviosly scales with Rake).

Obviosly, given the "very near" value of rake/shred threasold you can build a set without tier bonus stacking crit and ArPen, probably with that set you'll be better to shred instead of rake, but overall you'll probably lose dps to do a "niche" rotation.
The problem will be completly different in the next tier, but we don't know what kind of changes Blizzard will do for that time.

Rake's initial damage portion can crit, so it does somewhat scale with it.

Your remark about scaling may be a bit premature, given that for example Lotrafen's ilevel is 232 while the highest weapons are 239. Also, Ghostcrawler said recently that there are over 800 items in Ulduar. We haven't seen half of them.

I was wondering, did your search of BiS include just Ulduar drops or does it include badge and crafted items?

I'm curious how you model Berserking enchant (and procs in general). Do you model procs in your simulator or average the values out?

I'm using the formulas here and assuming 1.2 PPM with no internal cooldown. With BiS gear I get around 33% haste (with buffs), which leads to about 1.68 hits per second (with specials).

400*(1-(1-1.2/(60/weaponSpeed))^(15*hitspersecond)) = 127 AP

Now that Predatory Strikes affects weapon enchant, Massacre is worth 110*1.2 = 132 AP.

I recall some tests that were made back in the BC days about haste not affecting PPM (in an old rogue thread). Meaning that during heroism, you would actually have less chance of procing Bezerker with each hit because you actually attack more often.

I know rogues used to tests this with Mongoose procs during Blade Flurry.

I think it was in Beta testing actually that Ghostcrawler specifically stated that Haste would in fact be a small boost to our Omen of Clarity procs and other PPM's. That's the last time though that I remember any major discussion about it and unless something has changed since then or someone has done extensive testing that I'm unaware of, that should still be the case.

To clarify, are you valuing berserking at 200 AP or are you actually modeling it as a proc?

You're showing more specials than my spreadsheet estimates. I'm seeing 7.6 specials for a 22 second cycle, or one every 2.876s. With BiS I'm estimating 33% haste buffed giving a weaponspeed is around 0.752 per sec. I'm not counting berserk or bloodlust in either of these. I can see the yellow hits getting as high as yours but not the white hits.

Regardless, given your numbers:
0.66 weapon speed
116 / 60 = 1.933 hits per second

400*(1-(1-1.2/(60/weaponSpeed))^(15*hitspersecond)) = 127.9 AP (~32% uptime)

Are you saying this formula from the think tank is not correct? I have not see anywhere showing Berserking uses different mechanics. If the PPM is not modified by haste the uptime would go up to ~44% by this formula, or ~177 AP.


PPM mechanics are still affected by haste as far as I know. Blues posted that OoC specifically would not follow PPM rules, but changed to a flat chance per hit normalized to 3.5 PPM. I never saw a blue stating haste would now affect all PPM mechanics.

I was sure that Blizz get rid of ppm mechanic completly, I should test it.


I've always assumed (since Wotlk) that all proccing mechanics work in a simple way:

ppm/60 = chance to prox x attack
So the average number of proc is: (# attacks) * (proc chance)

And yes then there is an overlap chance.

With my bis gear I have a 0.72 weapon speed + bloodlust (in an average fight you have at least a 10-20% uptime) reducing effective weapon speed further to 0.67-0.69 (depending on bloodlust uptime, yes 0.66 was too much)

As for yellow attacks, this is a simulation report:
11 yellow attacks over 27 seconds = 1 attack every 2.445 seconds


Using Proc Mechanics

Uptime = 1- (1-p)^(D*v)

where p = chance to proc for each attack
v = attack /second
D = Buff duration

so for us:
p = ppm/60 = 1.2/60 = 0.02 = 2%
D=15
v= 1/2.445 + 1/0.68 = 1.88

Uptime = 1-(1-0.02)^(15*1.88) = 0.434 = 43.4%

Average ap = 400*Uptime = 174 AP.

Yes 200 was overestimated (not that 26AP really changes everything, but we should be accurate)

P.S.
Yes I'm averaging trinkents and weapons atm, I'm stll trying to find a smart method to avoid a lot of extra coding.

Okay, I have seen a few confirmations of PPM not being affected by haste anymore. However, they also indicate Mongoose and Berserking procrate has been reduced to 1.0 PPM.

Btw... 27 second cycle? Is that using the 4t8 set you posted or current 2t7? I guess you could get 0.68 weapon speed with current BiS. I was looking at Ulduar BiS which has less haste and shorter cycle.

Edit: Hopefully you can add procs to actual simulation. I'd be really interested to see if even the average value is the same, are procs worth more or less than static increase. My gut is to prefer the static increase, but I'm not sure what kind of interaction procs can have with each other or cycles in general. This goes back to similar analysis in BC where there were more "On Use" trinkets. Even if the average values were equal, "On Use" trinkets were more valuable than proc trinkets due to burst control. For example, Massacre will always be active when Rip is applied but Beserking only has a chance. Over time these might average out, but I prefer the more reliable static value. Again, this goes back to the same mitigation argument of armor vs avoidance.

27 second cycle is using BiS pre-Uduar gear (with 2T7). The values of yellow attack and average weapon speed (both count berserk and heroism uptime) are for a 5 minute fight.

Surerly adding on use trinkets and procs will be the next step, I have a few ideas about proc but on use trinket will be difficult to simulate due to extra conditional coding (probably different for each trinket, for istance you should want to use ArPen before FB? But AP before RIP?).
My guess is On Use > Proc > static

For the On Use the answer is pretty simple. Suppose that you have a 600 AP for 10 seconds with 1 minute CD trinket and another trinket with 100 static AP. The average AP is the same (100) but if the fight doesn't last a "multiple" of 1 minute the first trinket will always give more average AP. For istance if you use it at fight start and then the fight last 2 minutes and 10 seconds you have used it 3 times in 130 seconds giving an average AP of 138.
Also you can use it before specials to optimize the cycle.

As for proccing mechanics I'm under the impression that "a lot for a small time" is better than an averaged value, for example we know that ArPen becames better the more you have so having a lot of ArPen for a small time is better than having a few ArPen all the time.

I'll test it soon.
Actually I'm testing PPM mechanics (15k+ hits on target dummies and counting...)

When I have some reliable data I'll post it here.