Rogue DPS Simulation Spreadsheet - Page 8

saedo · 06/20/09, 4:10 PM

Originally Posted by Septhaka

[Footpads of Silence] appear to be slightly inferior or at best equivalent to [Flamestalker Boots] for most builds.

It depends if you have the JC prismatic gems available to you for that slot.

Mavanas · 06/20/09, 4:15 PM

Originally Posted by Chyan

About rupture damage calculation, there is one difference between your sheet and Aldriana's. The difference is on the rupture multifier calculation when considering the following three factors: (1) T7 2pc bonus, (2) Blood Spatter, and (3) Serrated Blades.

Aldriana's sheet is like:
(1 + "T7 bonus") x (1 + "Blood Spatter bonus") x ( 1 + "Serrated Blades bonus")

Your sheet (and Vulajin's sheet, too):
(1 + "T7 bonus" + "Blood Spatter bonus" + "Serrated Blades bonus")

I remeber I tested this about a month ago and my results show that you and Vulajin are right. Therefore, Aldriana's sheet will slightly overestimate the dps of 51/7/13 (when Blood Spatter and Serrated Blades are both active).

I did not independently test it and actually originally, I had them stack multiplicatively. However in 3.0.8 I changed it to match Vulajin's formulas. Multiplicative formula results in about 1.3% higher rupture damage.

chalon · 06/20/09, 4:21 PM

Oh yeah, I forgot of course the crit cap thing wouldn't affect the actual calculations! Wasn't thinking properly.

Ok, so most likely the discrepancy is 100% related to the Rupture bonus modifiers. If it's multiplicative, 51/7/13 is very slightly ahead, and if it's additive, 51/13/7 is very slightly ahead. Going to go run over and try this out on a training dummy.

EDIT: Ok, so I did a little bit of testing with my current spec 51/7/13...with 3820 AP and no Rupture glyph. I calculated the multiplicative damage/tick as 609 and the additive as 576. On the target dummy I was seeing 599/tick, though. I'm a little confused why it doesn't line up with neither of the two, unless maybe the (217 + 0.0375 * AP) damage/tick formula on Wowhead is wrong. Any ideas?

razorblade · 06/20/09, 5:03 PM

I've been looking at the current crit formula for a bit now and there are a few things that just don't make sense to me. Please forgive me if I'm being dense.

($K$10+612*$C$165)/83.333*1%+(AvDamage!$V$12+300*(1+Talents!$K$23*3%)*(1+Inputs!$H$20)*(1+Talents!$L$19+Talents!$H$27)*(1+Inputs!$H$12)*$C$157)/45.90598*1%-(0.295/100+8*0.006-Inputs!$H$16-Inputs!$H$5-Talents!$C$4-IF(OR(AvDamage!$Q$2="Fist",AvDamage!$Q$2="Dagger"),Talents!$H$10,0))

($K$10+612*$C$165)/83.333*1%
This is grabbing total crit rating and dividing it by the agility per crit conversion number.

(AvDamage!$V$12+300*(1+Talents!$K$23*3%)*(1+Inputs!$H$20)*(1+Talents!$L$19+Talents!$H $27)*(1+Inputs!$H$12)*$C$157)/45.90598*1%
This is getting total agility and dividing it by the crit rating conversion number. Also, it seems to be modifying the darkmoon card's 300 agility proc by the talent Savage Combat, and a 10% attack power buff on Inputs!H12.

I'm also wondering if the crit cap formula should be starting out at 104.8% due to bosses having 4.8% crit reduction.

The final thing I noticed was it seems that the base agility of 186 (for Orcs) calculates to 1.94% crit in-game and not 2.23% like it should. I adjusted my current crit calculations to adjust for this and it matches my character screen now. I apologize if this is a known UI bug.

Caffeine · 06/20/09, 5:20 PM

Originally Posted by chalon

EDIT: Ok, so I did a little bit of testing with my current spec 51/7/13...with 3820 AP and no Rupture glyph. I calculated the multiplicative damage/tick as 609 and the additive as 576. On the target dummy I was seeing 599/tick, though. I'm a little confused why it doesn't line up with neither of the two, unless maybe the (217 + 0.0375 * AP) damage/tick formula on Wowhead is wrong. Any ideas?

My guess would be you forgot Find Weakness, the additive formula with another additive 0.06 for Find Weakness gives 599 (598.015 actually) for 3820 AP.

• Aldriana · 06/20/09, 5:25 PM

Originally Posted by chalon

EDIT: Ok, so I did a little bit of testing with my current spec 51/7/13...with 3820 AP and no Rupture glyph. I calculated the multiplicative damage/tick as 609 and the additive as 576. On the target dummy I was seeing 599/tick, though. I'm a little confused why it doesn't line up with neither of the two, unless maybe the (217 + 0.0375 * AP) damage/tick formula on Wowhead is wrong. Any ideas?

Find Weakness. It's additive. Which, honestly, I thought I was already doing - I did fairly thorough testing of this a while back and came to the same conclusion, but apparently I messed it up at some point.

However, this probably isn't the sole difference - it's certainly a relevant one, but it may or may not be the entire story. I rather suspect there are some differences in our cycle definitions such that one or the other of us is predicting higher rupture uptime, which will also cause differences in terms of figuring out which is superior. Which set of assumptions is more accurate is, of course, a matter that can be debated, but I'd be surprised if our cycle definitions were matching up exactly.

Kytrarewn · 06/20/09, 5:34 PM

Originally Posted by Caffeine

My guess would be you forgot Find Weakness, the additive formula with another additive 0.06 for Find Weakness gives 599 (598.015 actually) for 3820 AP.

Nice catch... still, Chalon's an experienced enough TCer (he drafted the very first Rogue DPS spreadsheet, and in fact I think the very first DPS spreadsheet for any class, back in the day) that he would have noticed if almost all of his Rupture ticks were for 598 with the extremely-occasional tick for 599, which is, I believe, how all periodic effects have worked in the past when between two discrete data points?

Just for a sanity check, we'll all finding the initial predicted tick of rupture, with AP and no modifiers, to be 360.25, then we're multiplying it by 1.66 (the additive formula for serrated blades + find weakness + blood spatter?

With 360.25, we'd have 3 ticks for 360 and 1 for 361.

360 * 1.66 is 597.6

361 * 1.66 comes out to 599.3.

I guess you'd see more 599 ticks if the "damage per individual tick leading to total damage across the course of the effect" was determined before modifiers, but that'd still only be 1 tick in 4.

Mavanas · 06/20/09, 6:29 PM

Razorblade, that crit cap calculator on the gear page seems to be messed up, I have to fix a couple bugs in it, in particular the conversion coefficients and AP multipliers for agility. That's all incorrect. As for 4.8% reduction, it's there, depending on the version it's either with a plus sign on the left, or a minus sign on the right. In either case, the crit cap is extended by 4.8%.

Also did testing of envenom buff and it does go up even if an envenom is dodged. Finally, I did testing of whether murder affects the crit multiplier, and I can confirm that it no longer enters the formula. As you can see on the screenshot, my gouge hit for 461 and crit for 1097, after in-game rounding. I have RED, murder talent and full lethality. If murder entered the formula, crit damage would round to 1146, without murder, it rounds to 1096, which is within rounding error from the observed 1097.

So I am making a few adjustments to the sheet, and going to release a new version soon.

chalon · 06/20/09, 9:20 PM

Ok, that makes sense about Find Weakness. And yeah, I only tested it very briefly (like 2 minutes prior to the raid), but I was seeing more 599s than 598s. If it's more likely to be 598 instead of 599 as the calculations would suggest, maybe I was lucky. I'm not really sure though, and I don't care enough to do like 1,000 Rupture ticks in order to see whether or not 598 or 599 should be more likely :P.

Septhaka · 06/20/09, 10:23 PM

I am trying to determine how the spreadsheet models Turn the Tables. My initial quesiton was going to be how does the simulation estimate how often a raid member blocks, dodges, parries, etc. to produce uptime for TtT. However, I notice each point of TtT adds 2% to the multiplier amount in c26 on the Talents tab but that cell appears to have no dependents.

Chyan · 06/21/09, 12:06 AM

Originally Posted by Mavanas

As you can see on the screenshot, my gouge hit for 461 and crit for 1097, after in-game rounding. I have RED, murder talent and full lethality. If murder entered the formula, crit damage would round to 1146, without murder, it rounds to 1096, which is within rounding error from the observed 1097.

Maybe you should test it on "Murder"able mobs, not on the dummy. Though I guess the conclusion will be the same (Murder no longer affects crit multifier).

• Aldriana · 06/21/09, 12:07 AM

So, the usual assumption is that it has 100% uptime, which, of course, isn't wholly accurate... but it isn't ridiculously inaccurate either. Any time there's multiple things being tanked, or any boss that hits at a reasonable speed, uptime will be well over 90%, and 100% is probably as good an estimate as any.

The place where you get into trouble in this assumption is 1) exceptionally slow-hitting bosses, and 2) mobs that don't melee at all. For instance, in Mimiron phase 2, this assumption clearly falls apart.

So, in reality, it might be better to estimate it as 90 or 95% uptime, but from a practical perspective that's not going to change too much in the modeling of these abilities. It's a weak talent, and it's worth taking 1-2 points in if you don't need 3/3 MP, and past that it's not clear that it's worth worrying about the details.

Edit: To Chyan's point: the training dummies aren't murderable. As I recall the easiest way to test the murder multiplier involves taking off all proc AP gear and gouging rats in the Deeprun Tram.

Mavanas · 06/21/09, 1:13 AM

You are right about murderable, I guess I am going to IF tram.

As for TtT, 100% uptime is assumed. It's assumption #7 in How To tab, but I don't think many people read it anyway. The lack of dependencies is an Excel bug. I've seen it before, it says it does not have any dependencies but in fact it is used in a whole column. Just so you can rest assured, C26 on the talents page that corresponds to TtT goes directly into crit rate calculation in the combat tab (column CM) and it only applies to combo moves (which have an id 5 in the abilities column). I've tested that it's only a bug with the trace dependencies tool and the link does exist.

TtT is very close to MP in terms of dps, the differences are so small that they usually lie within RNG variance. In my simulation MP is usually slightly ahead, although some people reported that switching some points in TtT gives them a slight increase in dps. For myself I just decided to put 3/3 MP purely from raid buff considerations in case we are missing a ret pally or an elemental shaman.

EDIT: I just tested murder on rats. No RED gem this time. Regular hits were 282, crits were 648-649. Given lethality, with murder entering the crit multiplier formula, one would expect crits to be 678. Without it, one would expect the crits to be 649 on average. So it has been changed indeed, murder has been ninja fixed/nerfed sometime around 3.1. Here is a screenshot of Recap, I did not bother with a combatlog.

The latest version 3.1L has the murder talent and other issues brought up today fixed.

Feldro · 06/21/09, 4:45 PM

Mavanas,

In the current spreadsheet, Perilous Bite has the wrong stats.

48 Stam
54 Ap
47 Agi
31 Hit
20 ArP
171.1 DPS
No Socket

Live stats are

53 Stam
62 Ap
51 Agi
34 Hit
24 ArP
178.9 DPS
1 Red Socket

Mavanas · 06/21/09, 5:16 PM

Wow it becomes almost equal to Fang of Oblivion. I am going to make the change for the next version, Please update your own versions manually.

Septhaka · 06/21/09, 6:20 PM

With the BiS Mutilate setup (the gear/gem setup the spreadsheet comes pre-filled with), the 6 points that can be distributed between MP, CQC and TtT yield virtually equivalent results no matter what configuration they are in. Is it clear that where TtT put specials over the crit cap that its taken into account in the simulation?

I noticed the spreadsheet's assumptions state the boss target is murderable but the default mutilate build puts the 2 points you'd normally expect to be in Murder in Quick Recovery. Of course, 2/2 Murder yields a more favorable result that 2/2 Quick Recovery on a murderable mob.

Tosa · 06/21/09, 6:27 PM

Combatant Bootblade is also different live than in the current sheet.

Live stats are as follows:
200-301 damage, 1.4 speed (178.9 dps)
44 Agi
40 Sta
34 Haste
74 AP
32 Crit
1 red socket, 4 Haste bonus

Feldro · 06/21/09, 6:53 PM

I think what has happened here is some of the 10 man hardmode weapons had there stats increased. The sheet has not been updated to reflect some of these changes.

Mavanas · 06/21/09, 8:20 PM

Originally Posted by Septhaka

With the BiS Mutilate setup (the gear/gem setup the spreadsheet comes pre-filled with), the 6 points that can be distributed between MP, CQC and TtT yield virtually equivalent results no matter what configuration they are in. Is it clear that where TtT put specials over the crit cap that its taken into account in the simulation?

I noticed the spreadsheet's assumptions state the boss target is murderable but the default mutilate build puts the 2 points you'd normally expect to be in Murder in Quick Recovery. Of course, 2/2 Murder yields a more favorable result that 2/2 Quick Recovery on a murderable mob.

There is no crit cap for specials, it only applies to white attacks. So TtT is always effective, but it only applies to combo point generating moves.

If you want to see what damage you do on a murderable target, select 2 in Murder. Simple as that. Otherwise, put the points where you want them to go. I selected Quick Recovery as default, feel free to put points elsewhere, just keep in mind that dps impact of some talents, such as Fleet Footed, is not modeled.

razorblade · 06/21/09, 8:24 PM

Twirling blade's stats are incorrect also. It is also mispelled in cell BX215 which causes it to not show up in the socket bonuses list when you have it equipped.

Sarlunas · 06/21/09, 8:47 PM

How would I go about adding an ability like EA to the sheet, or would this be prohibitively difficult? I'm trying to figure out if the EA glyph is at all viable for 10-man raiding without a Warrior. It would be a nice boost for windup and target switching but how would it perform over an entire fight? Would I use something like 3s2e5r? My napkinmath is not up to this.

I guess it's a longshot considering the value of the KSpree glyph but it really sucks to start off with a 6 sec EA, especially in 10-mans where the shorter fights/phases make the windup more significant.

krauz · 06/21/09, 9:08 PM

I see dps/points in the gear sheet, and also in the dps output. Both are different, do they change according to my gear?
Which ones should i use to compare gear?

Mavanas · 06/21/09, 9:22 PM

Yeah I think I had ptr stats for a few items. I found a couple more among weapons.

EA is not trivial to add, it's another finisher that has to be modeled separately.

Gear sheet has most recent dps/point estimates. DPS output does not have updated mutilate dps/point estimates. Keep in mind dps/point estimates have about 0.05 standard error in that despite 10000 iterations I used to estimate them.

Ticia · 06/22/09, 1:42 AM

I can also confirm the murder is not applying to the crit multiplier anymore. I just tested a few times on the Unbound Seers directly below Dalaran. I had Murder and Lethality but no RED or AP procs of any kind.

First against a Level 78 one:

6/21 21:23:06.875  SPELL_CAST_SUCCESS,0x02000000009A2833,"Ticia",0x511,0xF13000828E5D0E7D,"Unbound Seer",0x10a48,1776,"Gouge",0x1
6/21 21:23:06.937  SPELL_DAMAGE,0x02000000009A2833,"Ticia",0x511,0xF13000828E5D0E7D,"Unbound Seer",0x10a48,1776,"Gouge",0x1,1135,0,1,0,0,0,1,nil,nil
...
6/21 21:23:17.187  SPELL_CAST_SUCCESS,0x02000000009A2833,"Ticia",0x511,0xF13000828E5D0E7D,"Unbound Seer",0x10a48,1776,"Gouge",0x1
6/21 21:23:17.312  SPELL_DAMAGE,0x02000000009A2833,"Ticia",0x511,0xF13000828E5D0E7D,"Unbound Seer",0x10a48,1776,"Gouge",0x1,493,0,1,0,0,0,nil,nil,nil
...
6/21 21:23:27.375  SPELL_CAST_SUCCESS,0x02000000009A2833,"Ticia",0x511,0xF13000828E5D0E7D,"Unbound Seer",0x10a48,1776,"Gouge",0x1
6/21 21:23:27.703  SPELL_DAMAGE,0x02000000009A2833,"Ticia",0x511,0xF13000828E5D0E7D,"Unbound Seer",0x10a48,1776,"Gouge",0x1,493,0,1,0,0,0,nil,nil,nil
...
6/21 21:23:37.593  SPELL_CAST_SUCCESS,0x02000000009A2833,"Ticia",0x511,0xF13000828E5D0E7D,"Unbound Seer",0x10a48,1776,"Gouge",0x1
6/21 21:23:37.859  SPELL_DAMAGE,0x02000000009A2833,"Ticia",0x511,0xF13000828E5D0E7D,"Unbound Seer",0x10a48,1776,"Gouge",0x1,493,0,1,0,0,0,nil,nil,nil
...
6/21 21:23:47.921  SPELL_CAST_SUCCESS,0x02000000009A2833,"Ticia",0x511,0xF13000828E5D0E7D,"Unbound Seer",0x10a48,1776,"Gouge",0x1
6/21 21:23:48.093  SPELL_DAMAGE,0x02000000009A2833,"Ticia",0x511,0xF13000828E5D0E7D,"Unbound Seer",0x10a48,1776,"Gouge",0x1,1135,0,1,0,0,0,1,nil,nil
...
6/21 21:23:58.171  SPELL_CAST_SUCCESS,0x02000000009A2833,"Ticia",0x511,0xF13000828E5D0E7D,"Unbound Seer",0x10a48,1776,"Gouge",0x1
6/21 21:23:58.312  SPELL_DAMAGE,0x02000000009A2833,"Ticia",0x511,0xF13000828E5D0E7D,"Unbound Seer",0x10a48,1776,"Gouge",0x1,1135,0,1,0,0,0,1,nil,nil
...
6/21 21:24:08.421  SPELL_CAST_SUCCESS,0x02000000009A2833,"Ticia",0x511,0xF13000828E5D0E7D,"Unbound Seer",0x10a48,1776,"Gouge",0x1
6/21 21:24:08.765  SPELL_DAMAGE,0x02000000009A2833,"Ticia",0x511,0xF13000828E5D0E7D,"Unbound Seer",0x10a48,1776,"Gouge",0x1,494,0,1,0,0,0,nil,nil,nil

So a non-crit value of somewhere between 493 and 494, with a crit value of 1135 = 2.3*(493.478)

Then against a level 79 version of the same mob:

6/21 21:24:58.953  SPELL_CAST_SUCCESS,0x02000000009A2833,"Ticia",0x511,0xF53000828E533397,"Unbound Seer",0x10a48,1776,"Gouge",0x1
6/21 21:24:59.218  SPELL_DAMAGE,0x02000000009A2833,"Ticia",0x511,0xF53000828E533397,"Unbound Seer",0x10a48,1776,"Gouge",0x1,1124,0,1,0,0,0,1,nil,nil

6/21 21:25:09.328  SPELL_CAST_SUCCESS,0x02000000009A2833,"Ticia",0x511,0xF53000828E533397,"Unbound Seer",0x10a48,1776,"Gouge",0x1
6/21 21:25:09.328  SPELL_DAMAGE,0x02000000009A2833,"Ticia",0x511,0xF53000828E533397,"Unbound Seer",0x10a48,1776,"Gouge",0x1,489,0,1,0,0,0,nil,nil,nil

6/21 21:25:19.781  SPELL_CAST_SUCCESS,0x02000000009A2833,"Ticia",0x511,0xF53000828E533397,"Unbound Seer",0x10a48,1776,"Gouge",0x1
6/21 21:25:19.781  SPELL_DAMAGE,0x02000000009A2833,"Ticia",0x511,0xF53000828E533397,"Unbound Seer",0x10a48,1776,"Gouge",0x1,1124,0,1,0,0,0,1,nil,nil

6/21 21:25:31.000  SPELL_CAST_SUCCESS,0x02000000009A2833,"Ticia",0x511,0xF53000828E533397,"Unbound Seer",0x10a48,1776,"Gouge",0x1
6/21 21:25:31.406  SPELL_DAMAGE,0x02000000009A2833,"Ticia",0x511,0xF53000828E533397,"Unbound Seer",0x10a48,1776,"Gouge",0x1,1124,0,1,0,0,0,1,nil,nil

6/21 21:25:41.328  SPELL_CAST_SUCCESS,0x02000000009A2833,"Ticia",0x511,0xF53000828E533397,"Unbound Seer",0x10a48,1776,"Gouge",0x1
6/21 21:25:41.328  SPELL_DAMAGE,0x02000000009A2833,"Ticia",0x511,0xF53000828E533397,"Unbound Seer",0x10a48,1776,"Gouge",0x1,489,0,1,0,0,0,nil,nil,nil

6/21 21:25:51.562  SPELL_CAST_SUCCESS,0x02000000009A2833,"Ticia",0x511,0xF53000828E533397,"Unbound Seer",0x10a48,1776,"Gouge",0x1
6/21 21:25:51.562  SPELL_DAMAGE,0x02000000009A2833,"Ticia",0x511,0xF53000828E533397,"Unbound Seer",0x10a48,1776,"Gouge",0x1,1124,0,1,0,0,0,1,nil,nil

So a non-crit value of 489 with a crit value of 1124 = 2.3*(488.696)

Mavanas · 06/23/09, 6:14 PM

After making the changes to crit multipliers and murder the dps table has changed a bit:

Tree		Spec	DPS	Special Notes
vHaT (1 c/s)	8/20/43	7810	ArPen gems, AG food, MR+DMC:G
Combat		7/51/13	7741	ArPen gems and food, CT+MR
Combat		15/51/5	7753	ArPen gems and food, CT+MR
Combat		18/51/2	7762	ArPen gems and food, eviscerate finisher, CT+MR
Mutilate	51/7/13	7698	Exp capped (from gear), hit cap, AG gems up to crit cap, then AP, DM+CT
Mutilate	51/13/7	7719	Exp and hit capped (from gear), AG gems up to crit cap, then AP, DM+CT
Mutilate	51/13/7	8064	Murder

Combat in BiS gear is slightly ahead of mutilate.

In addition to BiS gear, I did some analysis of BbH (Best before Hard modes) gear, which is more applicable to the majority of rogue population. The dps table for BbH gear looks like this:

Tree		Spec	DPS	Special Notes
vHaT (1 c/s)	8/20/43	7208	AG gems and food, DMC:G+GT
Combat		7/51/13	7027	ArPen gems and food, DMC:G+GT
Combat		15/51/5	7023	ArPen gems and food, DMC:G+GT
Combat		18/51/2	7033	ArPen gems and food, DMC:G+GT
Mutilate	51/7/13	7210	Hit cap from gear, AG gems and food, DMC:G+BotOG
Mutilate	51/13/7	7183	Hit cap from gear, AG gems and food, DMC:G+BotOG
Mutilate	51/7/13	7487	Murder

At that level, mutilate pulls ahead by a sizeable margin. vHaT has higher dps with agility gems. Gemming for agility and ArPen gives almost equal dps output for combat. Also interestingly, 51/7/13 outdpses 51/13/7. The difference is significant at 10% confidence level.

The BbH gear list for Combat and Mutilate obtained using the simulation is given below:

Combat					Mutilate
Insanity's Grip				Daschal's Bite
Golem-Shard Sticker			Golem-Shard Sticker
Giant's Bane				Siren's Cry
Conqueror's Terrorblade Helmet		Conqueror's Terrorblade Helmet
Favor of the Dragon Queen		Favor of the Dragon Queen
Conqueror's Terrorblade Pauldrons	Conqueror's Terrorblade Pauldrons
Drape of the Deadly Foe			Drape of the Deadly Foe
Winter's Icy Embrace			Conqueror's Terrorblade Breastplate
Mechanist's Bindings			Mechanist's Bindings
Conqueror's Terrorblade Gauntlets	Conqueror's Terrorblade Gauntlets
Relic Hunter's Cord			Death-warmed Belt
Conqueror's Terrorblade Legplates	Conqueror's Terrorblade Legplates
Footpads of Silence			Footpads of Silence
Loop of the Agile			Loop of the Agile
Cindershard Ring			Godbane Signet
Darkmoon Card: Greatness		Darkmoon Card: Greatness
Grim Toll				Blood of the Old God