FeralbyNight (Cat & Bear simulation tool) - Page 14

animagi · 04/24/09, 3:11 PM

just throwing the data out there, i spent 2x 4 minute sessions on a target dummy with the same gear. first time i used the FeralByNight addon and was at 3700 dps, the second time i used the FaceMauler addon and was at 4100 dps. these were completely unbuffed (not even mark). i haven't tried on a dummy with full raid buffs yet.

stats:
7147 AP
261 Hit
228 ArPen
51.96% crit
23 expertise
0.93 Attack Speed

Shred idol
2pc T7
Berserking
Relentless metaGlyph of Rip
Glyph of Shred
Glyph of Savage Roar

bn3 · 04/24/09, 3:23 PM

i dont think 2x 4 min sessions is enough to eliminate the rng factor maybe tray a few more sessions would defenatly be intresting if there was a big diffrance

mesh7 · 04/24/09, 4:10 PM

Originally Posted by syntax53

Rawr 2.2.1.0 was released today. While "- Updated to the new ArPen model for 3.1" is in the patch notes, it still isn't valuing ArP anywhere near Agility. It's showing me as having 355 ArP (with grim toll avg) and with only sunder armor and faerie fire checked off on buffs it's weighting agility at 0.94 and ArP at 0.76 (below crit and strength). Fully raid buffed agility is at 1.36 and ArP is 1.20 (only below strength at 1.22).

This is an interesting note, because Toskk's feral dps site also puts the new ArPen way below Agi. Toskk's site does this regardless of what stats you put in. I know ArPen scales the more you have, but after using ArPen values of 500, 600, etc., Toskk still has Arpen at or alittle less then Agi.

Who's correct Toskk/Rawr or Nightcrowler's method?

Mijae · 04/24/09, 4:29 PM

Originally Posted by mesh7

Who's correct Toskk/Rawr or Nightcrowler's method?

This is theorycrafting. If that were an easy answer, everyone would already be agreeing. The main issue is that calculators try to guess what your cycles will look like averaged over time. However, they will only be accurate if you are performing your cycles exactly like the calculator describes. So, it depends on your cycle and I can guarantee that none of them match exactly what occurs in game.

If the FBN addon follows the exact same rules that his simulator does AND you are following the addon with zero errors, his results should be valid. Otherwise, they won't be a perfect match and your real stat values will vary.

Nightcrawler - I really think you should post the explanation of stat importance you gave here on the main post. It could really help to clear up the questions people might have as to why one stat might be better than another. I've always tried to explain the difference in value between Str and Agi in the past (when they were nearly equal), and it seems that basically ArP has just replaced Str now.

mekell · 04/24/09, 5:10 PM

Night and others, other than just providing data via brute force simulation, can someone explain to me why its generally not desired to gem for +hit if you are below the hit cap for feral cats? I seem to recall reading about the two-roll system and it having something to do with it since a large portion of cat dps is yellow special damage. I have tried searching EJ and google and everything and just cannot find a good explanation.

Cryect · 04/24/09, 8:15 PM

Originally Posted by mesh7

This is an interesting note, because Toskk's feral dps site also puts the new ArPen way below Agi. Toskk's site does this regardless of what stats you put in. I know ArPen scales the more you have, but after using ArPen values of 500, 600, etc., Toskk still has Arpen at or alittle less then Agi.

Who's correct Toskk/Rawr or Nightcrowler's method?

The value of ArPen varies with your gear and the numbers in the opening post are for BiS before 3.1. So if you weren't close to BiS they will be quite off.

Example in the newest version of Rawr it shows my ArPen being valued at 1.56, Str at 1.44 and Agi at 1.16 for fully buffed without a mangle bot. With a mangle bot ArPen at 1.53, Str 1.41, and Agi at 1.50. Completely unbuffed though it drops to Agi 0.88, Str 0.84, and ArPen 0.81. Minor things have pretty drastic results on the effects of stats for your gear.

samboy29 · 04/24/09, 11:03 PM

Couple of quick questions, first:

What compiler did you use? When I downloaded the cpp file and tried to compile it with DevC++ it returns a heap of error messages.

Also, what is the var "wep_dmg" because I went through the file and couldn't find any values assigned to it (other than 0). Basically I'm trying to figure out what the difference between aDamage and bDamage is. Yes I did read the comments, just I'm trying to figure out what the numbers behind it all.

double base_paw=54.857+wep_dmg
weapon=base_paw

wep_dmg = ?? (This is where I couldn't find the value assigned to wep_dmg (other than wep_dmg = 0)

paw.bdamage=(1-red)*(weapon)*(1+naturalist_m);

Interesting that the Shred Idol is better than the Rip one, considering the Rip one does ((26 / 2) * 104) damage per Rip (presuming you have 2p t7, shred glyph, rip glyph). I presume it is only better if you have someone applying Mangle or Trauma for you? otherwise I wouldn't have thought you'd be able to get enough (7) Shred's off for it to be viable.

P.S. Does your addon consider if Rake and Mangle is about to wear off? I'm sure I've seen a couple of instances where both wear off within a second of each other, and it is telling me to use Shred when I have just over 40 energy with barely any time left on Mangle and Rake (didn't see the exact time left). Just figure that higher uptime on Mangle and Rake is preferable to Shredding and having downtime.

Great work ^^

BioNiT · 04/25/09, 4:35 AM

Try intergated facemauler with checked "suggest FFF", and then with unchecked.
See the difference on predictions ? ^^

Wickedgirl · 04/25/09, 10:48 AM

Originally Posted by samboy29

P.S. Does your addon consider if Rake and Mangle is about to wear off? I'm sure I've seen a couple of instances where both wear off within a second of each other, and it is telling me to use Shred when I have just over 40 energy with barely any time left on Mangle and Rake (didn't see the exact time left). Just figure that higher uptime on Mangle and Rake is preferable to Shredding and having downtime.

In my experience this only happened if I had a Clearcasting proc, although it will often choose Mangle on Clearcasting too. Also, from what I have seen, Mangle will always have priority over Rake, if they are both down.

My question would be to Nightcrowler: does the addon have an "internal" counter of the number of Shreds you used, in regards to the Glyph of Shred, does it in anyway change the suggestions order and could it be included as a visible option somehow?

Thank you for the upgrades and work on the addon

Murna · 04/25/09, 2:12 PM

Originally Posted by Nioreh

Rawr assumes 100% rip uptime and doesnt use FB in its cycle.

This is just not true!

I've got pretty good pre 3.1 gear. Not completely BiS, but there's not too much missing.

Checked basically every raid buff / debuff including Flask and armorpen food, only missing the draenei hit bonus.

No Mangle Bot!

AP 10.374
Agi 1597
Str 369
Crit 667
Hit 263
Exp 40
Haste 280
ArPen 361 (this is with Grim toll already, which is averaged to 124 ArPen Rating)

I am using Rip Idol.

Melee 26,6%
Mangle 6%
Shred 18%
Rake 14%
Rip 29%
FB 6%
DPS 6.7k

Stat Values:

Agi 1.35
Str 1.2
ArPen 1.15
Crit 1.11

Not only does RAWR rates ArPen much lower than FBN, it also shows that Rip Idol is nearly double as good as Shred idol.

After using FBN Addon, my impression is, that FBN leads to some little mangle downtime and quite a few Rip downtimes in favor of more FB's.

However, this is not really my playstyle. I guess Toskk and RAWR aim for more Rip uptime, which should in return lead to higher value of Agility and Rip Idol.

Deathwing · 04/25/09, 4:38 PM

Murna, try putting the armor pen gem and gear in manually. Or, better yet, select some of the best gear from naxx and ulduar, and then run the optimizer.

Armor pen builds on itself. Rawr and fbn just disagree on the inflection point as to when ArP passes agi.

nightcrowler · 04/25/09, 8:33 PM

@animagi: 3700 dps on dummy with your stats is pretty low, the stats are pretty similar to mine and you should do around 4k on a dummy. For a single 4minute fight you can see a dps range from 3500 to 4500. So your data doesn't mean a lot.

@Mijae: FBN simulator and FBN addon follow exactly the same rules, as for arpen explanation i'm writing something and doing tests, then I'll update the main page as you suggest.

@Murna: Without a mangle-bot rip-idol is better than shred-idol also for me (at least at normal level of ArPen). Do you have the latest version of Rawr?

@Wicked: I don't understand why sp many people are asking for a shred counter, I'm thinking about adding it but it will require some work. Actually the addon simply read rip timer to decide the next spell. When you shred, rip timer = rip timer +2, unless you have shreeded 3 times.
Usually I can get 3 shreds pretty soon, the only things that can change is if you don't FB and the rip timer increase and you think "I could have used FB". But things like that are really unlikely to happen. I must try it.

Murna · 04/26/09, 10:38 AM

@nightcrowler: Of course I am using the latest RAWR version. With Mangle-bot Rip Idol is rated at 181 Points and Shred Idol at 122. Without Manglebot its again 181 for Rip Idol and only 95 for Shred Idol.

@deathwing: Sorry, I don't really understand what you mean by "putting the armor pen gem and gear in manually". I downloaded my char from the armory and just added all raid buffs / debuffs. I also added a gemming template, completely focused on ArPen.

I am using RAWR now for over a year I guess and never had any problems. The DPS it shows for me is always very, very accurate. For example it suggests 4410 DPS completely unbuffed (except my imp MotW) and I manage to do 4300 - 4500 on the target dummy every time...
I used to completely trust RAWR when I choose my gear. But this is the first time I am getting sceptical. RAWR rates ArPen quite low, which is in extreme contrast to FBN.

The Point, where RAWR shows ArPen equal to Agility as best stat, is 750 ! (using Rip Idol). Even then Rip Idol is rated with 181 Points against Shred Idol with 108 Points.

I am just curious about the cause of these differences. Is it because RAWR and FBN use other DPS cycles? If so, everyone of us would have to choose, which of the two theoretical models better match the very own playstyle.

Monfalaris · 04/26/09, 11:46 AM

No, RAWR assumes a maximum bleed uptime, which results in more damage coming from bleeds than it is the case for FBN, where the cycle has gone through quite a number of modifications now in order to achieve the best results for a static combat situation, reducing the overall bleed damage percentage. If this is also the best for 'real world' (still talking about a computer game here) encounters is not yet known, but most probably it is not. In which way it needs to be modified to implement an optimized cycle considering movement, we do not know yet. Be patient, we will be coming closer to finding an answer

EDIT: Until then, if you are really working hard on keeping bleeds up and don't bother with weaving in FBs, then you might want to prefer agility. If you use FbN to do the work for you, you should support this accordingly and rather stack ArP, in case you have seperate cat dps gear. It really comes down to playstyle right now.

@Murna: Sorry, I got something wrong the first time I read your post. The first word should be 'Yes' not 'No'.

Cryect · 04/26/09, 7:09 PM

Originally Posted by Murna

I am using RAWR now for over a year I guess and never had any problems. The DPS it shows for me is always very, very accurate. For example it suggests 4410 DPS completely unbuffed (except my imp MotW) and I manage to do 4300 - 4500 on the target dummy every time...

Quite strange unless your crit chance is 0 since it assumes absolutely 0 wasted combo points (my results were it was normally a couple hundred DPS short due to crits at 4 combo points etc and rip not being possible to keep up 100% as a result like rawr assumes should occur). Also your ArPen is quite low so wouldn't surprise me it rates ArPen less than Agi.

Murna · 04/26/09, 8:15 PM

Originally Posted by Cryect

Also your ArPen is quite low so wouldn't surprise me it rates ArPen less than Agi.

Are you actually reading my post?

I would need 750 ArPen Rating for ArPen to become as good as Agi. This is not really low in my opinion.

I also would be VERY surprised, if RAWR really assumes no CP waste...

Boevis · 04/26/09, 9:49 PM

Rawr shows ArP better than Agi far earlier for me, I can actually get it to cross over at 362 ArP. The thing you're forgetting is that while ArP becomes better as you get more, there's other factors to consider like your Crit, Hit, Expertise, and AP as well as buffs, talents, and glyphs. With that 362 ArP, I'd have 53.8% Crit, 10605 AP, and 2.78% Avoided, and while switching 1 Agi gem for ArP gives me more DPS, switching 2 makes me lose DPS because of the crit/ap loss from agi.

I'm using v2.2.1 for these results.

Pheratia · 04/26/09, 9:56 PM

i have the same results as both murna AND boevis, although to get both results, I have to switch from 2.2.0.9 (low ArP threshold ~350 same as Boevis) and 2.2.1 (same as Murna) I'd like to know what's going on, but if agi really turns out to be better than ArP i will be very glad as that merges cat and bear gear sets alot better.

Nadir_Eonar · 04/26/09, 11:42 PM

Something which I don't see mentioned much in these ArP vs Agi discussions is that the models assume 100% uptime on the boss, i.e a patchwerk type scenario. It's worth bearing in mind that in any situation where you lose melee contact with a boss the rip-favouring (agi) builds will be negatively impacted less severely than the ArP-favouring builds.

To assess the actual extent to which this is a factor someone would need to make calculations of average uptime/downtime on the various boss encounters in ulduar I suppose, which would be both time consuming and highly subjective, dependant as it would be on adopted strategies etc. However I feel it's worth bearing in mind this 'hidden' advantage of agility when deciding on gear to use in a real-world situation.

Cryect · 04/27/09, 3:56 AM

Originally Posted by Murna

Are you actually reading my post?

I would need 750 ArPen Rating for ArPen to become as good as Agi. This is not really low in my opinion.

I also would be VERY surprised, if RAWR really assumes no CP waste...

As well, I've stated that Rawr has shown that ArPen is my best stat with both the rip & shred idols. ArPen won't be very good if your other stats though aren't very good since depends on having already heavy hitting atacks since ArPen is a % damage increasing stat where as stats like Str & Agi are flat DPS boosting stats.

And Rawr doesn't account for CP waste if you look at the code.

Rawr

Just looking at the code used for the combo point generator

float cpPerCPG = (chanceHit + chanceCrit * (1f + stats.BonusCPOnCrit)) / chanceNonAvoided;

Used with the below

CatRotationCalculator rotationCalculator = new CatRotationCalculator(stats, calcOpts.Duration, cpPerCPG,
				maintainMangle, (character.DruidTalents.GlyphOfMangle ? 18f : 12f) + timeToReapplyDebuffs, 
				12f + stats.BonusRipDuration + timeToReapplyDebuffs, 9f + timeToReapplyDebuffs, stats.BonusSavageRoarDuration,
				character.DruidTalents.Berserk > 0 ? (character.DruidTalents.GlyphOfBerserk ? 20f : 15f) : 0f, attackSpeed, 
				character.DruidTalents.OmenOfClarity > 0, character.DruidTalents.GlyphOfShred, chanceAvoided, chanceCrit * stats.BonusCPOnCrit, 
				cpgEnergyCostMultiplier, stats.ClearcastOnBleedChance, meleeDamageAverage, mangleDamageAverage, shredDamageAverage, 
				rakeDamageAverage, ripDamageAverage, biteDamageAverage, mangleEnergyAverage, shredEnergyAverage, 
				rakeEnergyAverage, ripEnergyAverage, biteEnergyAverage, roarEnergyAverage);

Which goes to be used in the file Rawr which you check every place where CPPerCPG is used that there is no factor for taking into account combo point waste.

Murna · 04/27/09, 5:40 AM

@Cryect:

I am not checking the code, but it seems like you know what you're talking about. I am just gonna trust you on this.
Actually, this is quite a big flaw in RAWR math - not taking combo point wastes into account is significant!

I do understand, that ArPen's efficiency is greatly influenced by all the other stats. But for example, I've got nearly the same stats as boevis, look at this:

I checked every raid buff/debuff except Draenei Hit and Trauma.

I am in cat gear and talent build on the armory. Perhaps you could load my char into your RAWR and test it yourself. I'm even a bit scared, that I'm doing something terribly wrong, but I just can't find it...

Duilliath · 04/27/09, 5:54 AM

Originally Posted by Nadir_Eonar

Something which I don't see mentioned much in these ArP vs Agi discussions is that the models assume 100% uptime on the boss, i.e a patchwerk type scenario. It's worth bearing in mind that in any situation where you lose melee contact with a boss the rip-favouring (agi) builds will be negatively impacted less severely than the ArP-favouring builds.

Nightcrowler has it modeled in this exact same thread. Read on from here.

Vaccine · 04/27/09, 9:41 AM

I get similar values as Murna, it isn't until the extremely high levels of Armour Penetration that it seems to push ahead, around 550 with Shred Idol and 650 with Rip Idol.

On another note I recompiled the BiS list now with the newly discovered items from yesterday. Rawr suggests using both Runestone and Grim Toll but I don't know how the similar procs interact with each other. If it is negatively then the next best trinket is Blood of the Old God.

Slot	Item
Head	Conqueror's Nightsong Headguard
Neck	Pendulum of Infinity
Shoulders	Conqueror's Nightsong Shoulderpads
Chest	Embrace of the Gladiator
Waist	Soul-Devouring Cinch
Legs	Conqueror's Nightsong Legguards
Feet	Flamestalker Boots
Wrist	Fluxing Energy Coils
Hands	Conqueror's Nightsong Handgrips
Finger1	Loop of the Agile
Finger2	Cindershard Ring
Trinket1	Grim Toll
Trinket2	Mjolnir Runestone
Back	Drape of the Faceless General
MainHand	Dark Edge of Depravity + berserking
ExtraWaistSocket	Fractured Scarlet Ruby

For slots I used all APR and JC gems to activate the meta.
For Idol well I guess it is personal choice. Rawr refuses to ever give me a result without Rip idol as number 1, but the difference is small in the above setup, about 20 dps. The final DPS result is 7908. It is a little below hit/exp cap which I'd have to fix with gems as they are too important to be below cap.

Cryect · 04/27/09, 10:37 AM

Originally Posted by Murna

I am in cat gear and talent build on the armory. Perhaps you could load my char into your RAWR and test it yourself. I'm even a bit scared, that I'm doing something terribly wrong, but I just can't find it...

Hmmm just seems your stats are a fair bit lower. Setting up the same buffs and idol as you I get a very different looking relative stats for my gear. I do get the exact same low ArPen value using your character (edit as you showed in your picture. The main differences I see is I have extra 200 AP, lot more expertise, lot less haste, slightly more Agi and a lot more ArPen).

Allev · 04/27/09, 12:01 PM

Obviously something non-apparent is happening in Rawr. I'd specifically look for things like Rawr changing which cycle is optimal. With the "average stat values" graphic, we may even see a case where adding a certain number of stat points changes the ideal cycle.

I haven't seen ANY specifics about how Rawr's cat cycles are implemented lately, whether in patch notes or on the site. For all I know, they've got multiple different implementations of cycles, and crossing stat barriers makes you jump from one cycle to the other. However, if you're playing with a priority list, calculating a static cycle is wrong-- sometimes you'll be using a 1-point SR just to get it active after an FB, sometimes you'll be using a 5-point SR in the middle of a berserk.

Essentially: either 1) Rawr is broken, or 2) Rawr isn't clear about what's changing between Murna and Cryect (specifically with Cryect, as the rating of his crit doesn't agree with any other current theorycraft). Either way, I wouldn't trust Rawr for more than getting calculator values.