Every spec I look at keeps talking about the value between WE and DD, stop ignoring me! Keep WE 2/2 Keep DD 2/2 remove one point of Sword Spec to make it 4% instead of 5%, that is definitely the smallest loss/gain factor between the 3 talents.

As for the lower level mobs making Hemo look better, your wrong imo. The lower level mob means less gain from serrated blades, not to mention they still don't have 0 armor or close to it which is another DPS loss compare to a raid where the -armor debuffs would be put up (sunder, FF, and CoR) making Serrated Blades scale even more. You'd actually gain more in a raid with SB than on those dumb mobs. Its a huge difference in overall dps too considering the large effect SB has on both white and yellow damage.

Anything on those Blasted Land mobs is shear speculation and opinion. Hemo vs Combat is so inaccurately tested on those mobs its not even worth taking the trip down there. Do it in a Raid where you get the debuffs and have raid buffs and post a WWS or just wait till the patch comes live to post your test results.

Maybe I am stupid, but I having trouble understanding what this statement means.

Are you implying that because enhance shaman won't be able to twist in the future, that a resto shaman will be closer to an enhance? or is there something I am missing?

No not at all.

I am implying that one of our Resto shamans, who was originally enhance/elemental and made resto due to losing a shaman, is going to spec back to enhance. Come 2.3 he will do this and considering his own DPS boost he will remain that spec instead of going back and forth to elemental. The part I said about totem twisting was that because he would be in our group at all times and be enhance, he would have very little trouble throwing down the totem twisting.

Edit: I just realized my typo. I said "come 2.3 when Enhance gets nerfed" I meant when Elemental gets nerfed. I however didn't say Resto was going to be better than enhance for a group buff, I don't know where you got that comparison from. I also have no idea where the idea of Totem Twisting getting nerfed as Blizzard hasn't done anything about it for a long time while knowing it exists. I was simply stating how I couldn't wait to get my enhance shaman back over the resto shaman I currently have because it's a large inc of damage. Again sorry for the typo =P

Two things here:

1) Yes, because of how the AC curve works, serated blades is naturally better against lower AC targets. In most cases lower level mobs will also have a lower base AC.

2) The armor mitigation % formula makes serated blades less valuable against lower level foes (assuming they actually have the same armor value as a level 70 which is highly unlikely).

mit(armor,level) = armor value / (armor value + 467.5 * level of attacker - 22168)

Since you're higher level, that means each point of armor is worth less to the target than it would be against a foe of the same level. Serated blades is a flat amount of AC reduction, so its value based on the formula above, is lower.


It's better to test this against a real level 73 raid boss. Glancing blows, dodge rates, parries from RSTS attacks (Morogrim turns to grave people, etc.), AC modifiers, and the mix of white/yellow damage also has an impact on which spec does more overall damage.

Yea as I said the whole buisness with the SB and the fact it scales so differently depending on the target and raid buffs/debuffs you really can't model an accurate comparison out side of raids. Same thing with CP and combat spec by comparison.

OK, as promised, I've done a less handwavy comparison of deep subtlety builds in damage terms.

I took the current best version of the DPS spreadsheet
http://rogue.raidcal.com/RogueDPS_2_3_0_3.xls

From this, I put in my own gear and buffs (combat maces, T4/T5 borderline, Dragonstrike and S2 arena offhand), and then looked at the calculated DPS for a range of different builds. Finally, I repeated the exercise using the gear and buff set that came with the sheet (high end combat swords gear). For this second run, I looked at the same set of specs, but with sword specialisation in place of mace specialisation where appropriate.

My armory is here if you want to check my own gear level.
The World of Warcraft Armory


I'm giving three numbers for each of the Hemo builds. In order, these represent
1) DPS ignoring all Hemo effects
2) DPS with "Hemo" ticked as a self buff - i.e. the number WWS will see as "personal" DPS
3) DPS with "Hemo" ticked on the talents sheet but not as a sefl buff - i.t. taking into account the full estimated damage added by the debuff

For the Shadowstep builds, "Hemo" is ticked on the talents sheet, thus measuring the full debuff effect. The other option is whether to use shadowstep on instants or finishers. Using it on finishers gives marginally higher DPS.

Results for my gear

Talent Calculator - World of Warcraft
20/41/0: 1376.06

Talent Calculator - World of Warcraft
11/27/23 (A): 1327.19 / 1388.61 / 1431.69

Talent Calculator - World of Warcraft
11/28/22: 1327.31 / 1388.90 / 1428.56

Talent Calculator - World of Warcraft
11/27/23 (B): 1336.26 / 1398.06 / 1440.57

Talent Calculator - World of Warcraft
11/21/29: 1353.13 / 1414.53 / 1457.73

Talent Calculator - World of Warcraft
11/20/30: 1359.15 / 1417.95 / 1461.40

Talent Calculator - World of Warcraft
11/9/41: 1373.98 / 1385.57

Talent Calculator - World of Warcraft
0/20/41: 1329.57 / 1337.59

Conclusion: Tri-spec beats out combat maces by a large margin once you factor in the debuff. It's not as overpowering as some have suggested, and if you ignore the debuff (for example if debuff slots are tight), the trispec has lower DPS. Hemo is thus a useful build, but only if you don't run heavy on affliction locks / shadowpriests / other debuffers. You could save the Hemo debuff slot by using IP instead of DP on the offhand: I haven't checked how this affects the calculations.


Results using the default gear from the sheet

20/41/0: 1700.48

11/27/23 (A): 1634.06 / 1706.22 / 1746.08

11/28/22: 1636.76 / 1709.13 / 1745.40

11/27/23 (B): 1636.15 / 1708.48 / 1748.49

11/21/29: 1620.25 / 1688.58 / 1731.86

11/20/30: 1628.29 / 1695.95 / 1742.27

11/9/41: 1657.29 / 1668.11

0/20/41: 1621.40 / 1632.38

Conclusion: Tri-spec beats out high end combat swords by a slightly smaller margin. It only pulls ahead once you factor in the whole debuff, rather than just the charges you consume yourself. This is because of the power of sword spec (note that the best mace tri-spec doesn't have mace spec, while the best sword tri-spec has 4/5 sword spec). Once again, Hemo is a useful build if you have the debuff slot available


Now, what can we say about Shadowstep? Well, in static fights where you can't make use of the increased mobility, it doesn't exactly shine. Nevertheless, if you use Shadowstep on your finishers, the Shadowstep build still fractionally edges out the combat maces build. For combat swords it's a clear loser. I haven't looked at fists because I don't know enough about fist itemisation, but the consensus seems to be that fist spec and mace spec are now about equal, with sword spec being the superior talent. Shadowstep thus is at least as viable as combat for mace rogues and probably also fist rogues

If you can use the mobility gains from Shadowstep, the results are harder to predict. You won't be able to time it to always use it on finishers, thus the right comparison to make is the number for using shadowstep on instants.

Comparing this to the optimal tri-spec, we find:

Mid-geared maces: 1461.40 / 1373.98 = 106.36%
Conclusion: Shadowstep needs to give an extra 6.36% DPS time (3.8 seconds per minute) in order to draw level with the optimal tri-spec. Note that any improvement from the mobility factor will cause shadowstep to pull further ahead of combat mace spec.

High end swords: 1748.49 / 1657.29 = 105.50%
Conclusion: Shadowstep needs to give an extra 5.5% DPS time (3.3 seconds per minute) in order to draw level with the optimal tri-spec. Note that you only need about two thirds as much improvement in order to pull ahead of combat sword spec.

Note also that Shadowstep seems to scale better than tri-spec with gear, probably due to the multiplicative interaction of Deadliness and Sinister Calling. One to watch when Sunwell comes out, perhaps?

Nope. I have no experience with that build, no idea how to work the itemisation, and I'm absolutely convinced that it's unmodellable by spreadsheets in any case, since it's so proc-dependent. Just go and try it out: empirical data trumps theory any day. Be sure to note the difference between proper empirical data and mere anecdote though :-)

Real world use of this spec is very frustrating. Chain sf procs forces miss timed eviscerates or wasted cp while lack of procs causes gaps in snd/rupture. Actual dps was far lower than it calculated at.

I must admit I'm having difficulty tracking this thread's discussion, not because of any technical terms, mathematics, or statistical data; I'm just unclear on what everyone is trying to figure out in all this. What the objective of this thread is, I mean.

Justify speccing hemo rather than combat for raiding.

That's the thread.

I must agree, you cant maintain decent cycle with that spec in raid situation.

I lead raids for our guild and Im most likely going to have myself and the two other rogues in the guild spec hemo, the reason being is we run a very melee dps heavy raid. Most raids we bring 2 ferals, 2-3 rogues, 2 hunters, 2-3 warriors, 2 enhance shamans, this along with pets is alot of melee attacks per second. I am going to track the uptime of hemo with demon(pdebufflist2) and see exactly how fast its getting used with 3 rogued refreshing it, if there is 100% uptime (highly doubt) then I might have 1 rogue spec combat, but if the hemo debuff is getting used to fast then having one out of 3 rogues combat is less raid dps. Also, take into account that not all 3 same rogues are on every raid, id like to have 2 hemo rogues always to benefit the other classes, so having all 3 raiding rogues spec hemo seems logical.

This is all under the assumptions that the spreadsheets are correct in modeling hemo at comparable dps with swords PRIOR to the hemo debuff factored in.

did some more testing today on the trispec. this is a fairly small sample. i would be up for testing a large sample.. say 1hour of testing if i could get a healer on the pvp ptr to toss me heals.

New testing conditions: 1703AP, 227hit, 25.70%crit, 15expertise(-3.75%chance to be dodged/parried). Lvl 57 servant in blasted lands
I took off my 2 normal weapons and used these:
Broadsword: 2.3speed @2.25 28-53damage(mainhand)
Cutlass: 2.2speed @2.15 10-20damage(offhand)
the increased attack speed is from my shoulders.
I also took off any items that had proc chances(trinkets/enchanted weapons). The only thing affected is the sword spec and the inert haste from my shoulders.

Goal: using a modified trispec conclude if 5/5 sword spec with 1/2 Dirty deeds does more or less damage than 4/5 sword spec with 2/2 dirty deeds

Here is the build i used: Talent Calculator - World of Warcraft

i left out talents which would skew the numbers further. i picked up 2/2 imp Sinister and 3/3 imp gouge instead of 5/5 lightning reflexes. i did not pick up set up or opportunity, as setup would proc more combo points, making the data less accurate. This is because i wouldn't normally be attacking from the front and therefore would not be receiving combo points for dodging attacks. I also did not pick up any points in dirty deeds. This is because we can use the math to calculate the damage it would add easier than we can take the damage it caused out.

The numbers: i did 2 tests and merged the data to one.

test time: 566.31seconds
total damage: 327,790
dps: 578.82 dps

melee: 221,596 damage 734melee swing, 533hit(235average damage)+ 201crit(507average damage)=27.38%crit
hemo: 68,438 damage 154swings, 114hit(335average damage)+40crit(691 average damage)=25.97%crit
rupture: 37,362 damage 160ticks(8ticks per rupture= 20ruptures) 234 damage per tick
eviscerate: 2,430 (i did 1 evis to end each of the 2 seperate tests.)

Dirty deeds increases yellow damage by 10/20% below 35%. By not having this in the build, it is like the mob was a normal mob that i was attacking that had 327,790 health.

35% of 327,790health =114,726 health
35% of 566.31 seconds =198.21 seconds

assuming damage was approx. even for the duration of the test:

melee damage caused below 35% = 77,559 damage
hemo damage caused below 35% = 23,953 damage
rupture damage caused below 35% = 13,104(7ruptures=56 ticks at 234 damage per tick)
assume 2 eviscerates below 35%= 2,430 damage

1/2 dirty deeds(+10%) would then add:
0 damage to melee damage
2,395 damage to hemo
1,301 damage to rupture
243 damage to eviscerate
3,939 total yellow damage from 1/2 dirty deeds.

2/2 dirty deeds(+20%) would then add
0 damage to melee damage
4,790 damage to hemo
2,602 damage to rupture
486 damage to eviscerate
7,878 total yellow damage from 2/2 dirty deeds.


Sword spec: i have a 4% chance to proc an additional attack due to the spec i used with testing. we will assume i recieved exactly 4% additional attacks

4% of 734 melee swings= 29 additional swings @27.38%crit= 21 hits and 8 crits
(21x245=5,145) + (8x507=4,056)=9,201 damage

4% of 154 hemo swings= 6 additional swings @ 27.38%crit= 4 hits and 2 crits [rounded up]
(4x245=980) + (2x507=1,014)=1,994damage


9,201+1,994=11,195 damage from sword spec at 4% proc rate

327,790total damage - 11,195sword spec damage= 316,595
316,595/327,790 = 96.58%
100% - 96.58% = 3.42% of damage was from sword spec.
3.42/4 = .855% of total damage per 1point in sword spec.

assuming this to hold true for the 5th point.
327,790x.855% = 2,803 additional damage for 1 more point in sword spec


final numbers:
11/28/22 trispec (5/5sword spec with 1/2DD)
2,803 damage(swordspec) + 3,939 damage (1/2 DD)=6,742 additional damage

11/27/23 trispec (4/5 sword spec with 2/2DD)
0 additional damage from sword spec and 7,878 damage (from 2/2 DD) = 7,878 additional damage.

Conclusion: 11/27/23 is slightly better than 11/28/22 with the testing/gear used. However, with an increase to white damage through haste multipliers that inturn were to increase sword spec procs or if 4% is lower than the actual %received 11/28/22 would edge back to the front.
The other advantage 11/28/22 has in these calculations is that this assumes the time to cause X damage is the same sub 20% as it does any other time. If all raid members are up then the time the boss is sub 20% would decrease due to heroism/execute spam. The shorter the time the boss is alive sub 35% the more the advantage goes to 11/28/22

EDIT: final numbers portion and calculations leading to them are incorrect. Values sword spec damage are based off of average damage of BOTH weapons instead of just the MH. This means that, under these testing conditions, the 11/28/22 build is more damage than the 11/27/23

Huge nerf to the Hemo attack in the latest PTR build. It now increases any Physical damage dealt to the target by up to 11. Lasts 10 charges or 15 seconds.

*** Disregard ***

There was no change to Hemo. It was my error from reading the tooltip on the talents page instead of from the spellbook.

Go and buy the higher ranks.

The shadowstep scaling seems rather promising.
Because, figuring in the added PvP and utility power from speccing for Shadowstep makes a ~5% loss quite ok overall.

Wouldn't have thought it faires so well - thanks for the testing

The scaling is a bit of a red herring. An increase of ~300 dps between the two gear levels changed the deficit from 6.36% to 5.5%

Assuming linearity (which is a particularly dodgy call), Shadowstep doesn't catch up with tri-spec until your DPS reaches about 3666. Not gonna happen, even with Sunwell gear.

The viability (if there is any) of Shadowstep in raids depends on the mobility factor. To make Shadowstep come in ahead of tri-spec, you need to save four seconds (28 yards) per minute. Shadowstep range is 40 yards, melee range is 5 yards, so a max range shadowstep saves you 35 yards of running.

So, one full-range shadowstep per minute (or two shadowsteps from ~20 yards) puts you ahead of the game. Anything less than that and you'd be better off as tri-spec. And if you're the third rogue in the raid, you should probably be combat, as both the tri-spec and the shadowstep spec depend on having all Hemo charges used up on every hit.

OK, look in your spell book - what ranks are shown? What is the debuff strength for Rank 1, Rank 2, Rank 3 and Rank 4?

Looks like I made the same mistake as AnalogKid. I logged on briefly to check something on my talents page and that's where I saw the 11 reported. I did not check the tooltip from the spellbook.

Sorry for the misinformation.

Nice testing
but this is wrong, the real huge side of sword spec is that OH procs make MH hits, and MH hits hard. To calculate the real benefit of sword spec you have to have the MH and OH hits separate and calculate MH damage from them. This have to be done with 1.5 OH, not 2.2 too

blah.. you're completely right, don't know why i overlooked this. That would make the 11/28/22 build jump ahead of the 11/27/23 build under those test conditions. I'm still willing to do some longer(uninterrupted) testing if anyone would like to heal for me on the PTR.
As far as a 1.5 speed sword, i could find one on the ptr. 0 on the AH and i checked 7 weapon vendors... all 2.3 and 2.2 for swords. Maybe i'll buy some off the AH on live and recopy my character over.

Any one else doing some sample testing and have some raw or calculated data?

Thanks for all the testing Songster. It's encouraging to finally see Subt have some value!

I'm curious about the talents you chose, specifically in the Subt tree. I noticed that for several of them, you picked up Ghostly Strike. Was that used in your DPS cycle, or did you stick to a Hemo, SnD, and Rupture cycle only?

Thanks in advance man!