[Fury of the Five Flights] is the single best trinket in the game for a DW Deathknight. It grants no less than 320 static AP for essentially the entirety of most fights.

I saw one of the newly released Thundering Skyflare Diamond on the AH and decided to test it out since wowhead has no data on it and haste is quite good for DW. I just ran a normal rotation on a lvl 80 target dummy, watching the Blizzard stopwatch and recording times between procs. Over 25 procs I found an average of 56 seconds between procs (range: 44-80 seconds). I'm guessing a 45 sec ICD, and a rather disappointing proc chance. Since its a 6 second 480 haste rating proc (14.64%), with an average up-time of 6 seconds of every 56, thats roughly equivalent to a constant 56 haste rating (1.6% haste).
I'm not going to try to do the math for comparison, but I'm sticking with Chaotic.

I've mostly been 2h builds for raids, but occasionally all the buffs get covered. That being the case, I try different DW specs. So far, I've not been impressed with either 0/44/27 nor 0/32/39. I was using this rotation:

IT-PS-BS-BS-HB-dump
IT-PS-IT-IT-HB-dump

I did the little things as well, using gargoyle when all procs + trinket are up, and got heroism benefit. Recount said I only did 4700 dps (including AotD), and I'm pretty sure WWS says I was lower at ~4500.

http://wowwebstats.com/ey5ppm5rmis15

I got really frustrated after Patchwerk because I did 5300 the week before as 51/0/20. Because we had the 10% AP buff, I went 21/50/0 to try it out, so ignore Thaddius and later for DW.

The World of Warcraft Armory

Note - I had the spiked titansteel helm and the 203 IT damage sigil.

Anyone got thoughts on what could be improved upon?

Hey guys, I just picked up [Mark of Norgannon] from 25 man Malygos today. I wonder if it is worth using over mirror.

The expertise itself gives me about 2% hit rating. Assume my melee damage is about 30%, so that's 0.6% increase. Seems inferior to 2% crit rating. But consider it increases KM proc rate and reduces miss rate on PS(aka desecration), does that balance out the difference?

Also about the use on the trinket itself, this one allows me to use when both greatness card and FC are up, making it 3 buffs for the gargoyle. On the other hand the mirror has about twice as much uptime, so does this balance out the difference on the procs?

I think the biggest area where [Mark of Norgannon] suffers is the inherently low value of expertise to DW DK's. It's hardly a bad trinket but would be much more valuable to a class that doesn't rely so heavily on spells and melee in tandem. In the matter of this v. Mirror, I would probably opt for Mirror simply because of the much stronger static stat value.

Agree with Orion. The static benefit from mirror does benefit you more over Mark of Norgannon. Also worth considering is the chance of having mirror proc twice in two minutes time compared to Marks 2min cooldown.

Statistically speaking the uptime for mirror and mark are very close. Mirror has a ICD of 45 seconds and a duration of 10 seconds. So that's about 22% uptime.
However, consider the chance to proc to be around 15%, one will need about 6.6 hits to proc. Assume I'm using fast/fast, and I attack about twice every second.
That requires about 3 to 4 seconds to proc mirror, pushing the cooldown to about 48 to 49 seconds, so that makes the uptime to about 20%.

On the other hand mark has a constant 20 seconds duration over 120 seconds uptime, that is about 16% uptime.

The question is, whether haste outweight AP. If each point of haste is 2.5 points and AP 1 point. Mirror gives about 1000*1*20% = 200 static points. Mark gives 491 haste, so 491*2.5*16% = 196 static points.

For those that asked me to post Angry Dread OH numbers with Last Laugh main hand -

This is self buffed with HoW and bone shield only - using pet on boss dummy in ebon blade while it is above 35%. Tests are approximately 3 minutes duration. I am 32/39 spec with only 1 filler point in desecration - I am using no trinkets / racial / gargoyle. Just rotation. I can add more if anyone else wants any test.

Blood – LL / LL – 2690
Unholy – LL / LL – 2560

Blood – LL / AD - 2625
Unholy – LL / AD – 2511


So LL edges it out slightly which means even in OH the superior DPS of LL outweighs the stat boost of AD. (I wasn't sure today when i was theorycrafting it, because weapon dps in OH would have less of an impact then MH due to miss and innate damage penalty)

There is also the added advantage - albeit slight - of pet scaling with last laugh due to it's str component.

How do you figure that 1 haste is 2.5 ap? That flies in the face of all the information posted on these boards.

I never said it does. I'm saying IF 1 haste is 2.5 AP then the proc ends up to be the same value. And about the information that you are talking about, there are people that value it at 0.5 point, 1.5 points and even 2.5 points. There is still no conclusive estimation on haste yet, although it's unlikely to be 2.5 points.

Also that's just an assumption on the proc alone. You need to consider other factors such as the ability to control the proc/use to maximize its usefulness.

the value of expertise went up witht he changes to KM.

Most DW DK's have gone back to slow/fast from what i've seen.

Using BS instead of BB in my rotation lately and have noticed a very large amount of dodges resulting in blood runes not going to death runes, screwing with my rotation more than before. If you want to use BS in your rotation i'd suggest capping expertise, if you are using blood boil for your death runes, then obviously expertise loses it's value.

IMO picking up mark of norgannon is worth it but that's jsut me.

Try running your tests again with your ghoul. It is part of your dps and impacted by your stats so I'm not quite certain why you excluded it.

The tests run by thinkrevolutionx suggest that LL/LL is the best combination. Try repeating the process he did and see if you get similar results.

I think he did it just right considering he wanted to just find raw weapon damage as opposed to raw damage plus a ghoul variable.

what was your rotation?

mine was
IT-PS-HB-BS-BS-RP
IT-PS-HB-IT-IT-RP

I use a priority rotation that isn't really set.
It's pretty much:

PS (if no BP) IT (if no FF) - - > HB - - > IT - - > BS (for death) - - > Death coil - - > BS - - > PS

I try to spam IT every time its up as long as KM is on CD and sometimes i'll wait .5-1 second if I know it wont lose much but will allow me to use a "dead" unholy rune on a KM with a refreshing frost or death. I'm going to run again on the boss dummy with angry MH LL OH to see how that goes. I tested it before but it wasn't boss dummy.

To bring a bit of sanity to this thread:

3 minute tests on a test dummy without pets or raid buffs are in no way indicative of real raid DPS values. Trying to figure out which weapon combo is the best in this kind of situation is completely worthless, especially when the values that result from these tests are within such a close margin anyway. The fact that raid buffs affect many different aspects of a death knights DPS due to spell and melee buffs working in combination on some skills and not on others, along with being able to provide some of those personally depending on talent subsets, can and will swing results so that they are not valid when compared to each other. The reason that people use spreadsheets and simulators is that generally the results from them will be accurate in a perfect world. Doing tests where you miss as much as one rune due to a dodge, and don't in another rotation can make all the difference to make you conclude from short testing that one weapon set is better than another because it scores 20 dps higher when it's actually not.

What needs to start happening in this thread is testing of specific talents to show individual gain per point in raid buffed situations in order to be able to compare those values to the expected in order to make sure that spreadsheets/simulators generated from expected values are correct. Throwing random DPS values from 3 minute tests in a completely unbuffed situation is going to get this thread no-where (which is basically where it's been going for a while).

As useful as Patchwerk DPS parses may seem, they aren't of much use without specific information about the setting that they were generated in. Were you experiencing any lag or lag spikes? What rotation / priority system were you using? What weapons were you using? What spec (exact link to wowhead/similar, not just x/x/x please)? What enchants/runes? Did you forget to use Empowered Rune Weapon? Did you forget to use Blood Tap? This information is all valid and relevant to being able to progress this thread in a meaningful manner towards the best possible DPS output. Arguing about whether dual wield is better or worse than 2h specs is not the point, the point is to look at dual wield DPS and see how to maximise it.

This is also not the thread to be asking single questions about stat weights, whether one presence is better than another, or any similar questions. The simple questions thread is there for that purpose.

I also think a lot of people need to read the announcements about forum rules. Every 2 posts someone who is apparently a "long time reader, first time poster" makes 4 different rule violations and gets smacked down for it. As useful as information and discussion is about the subject, if it's not easily readable then people are likely to ignore it.

so true.


Also to the above poster saying the dummy tests are worthless all I can say to that is that I completely disagree. Perhaps the duration could be longer - but I suppose it would depend on what your trying to test. I'm testing weapon damage, and since the two weapons i'm testing have different stats in terms of offensive / defensive I need to do my rotation as well in order to see it's impact on that.

You can pretty safely assume that if you give me 2000 more AP it's only going to scale my dummy results. So I guess it depends on what exactly your looking for. I never said this was a "point for point spec" test - its a "what weapon setup is best for DW" test and completely viable imo.

It is not viable if by that you mean indicative of the damage you'd actually do in a raid. Weapon stats impact pet damage, and raid buffs do as well. If you don't include pet damage or raid buffs in your testing, you put too much value on weapon dps and not enough on stats.

No. You're basing your assumption on what weapons provide best DPS based on a rotation that has many factors of weapon DPS. Without analysing the actual combat you just produced and seeing if the stats match up at all reasonably (crit rate, hit rate, damage range, proc numbers, space between procs, etc etc) you can't draw any conclusions. I could go and test slow/fast and come out with a DPS value of 2500, try fast/fast and get 2000 DPS, yet if my miss rate is 20% in the second test and 5% in the first the test is completely invalidated because they are not similar. The longer the test, the less these discrepancies should affect the test, but even with a large sample size you can easily get big differences in something such as crit rate, especially with an effect like Killing Machine that can back-to-back proc and proc directly before an ability you didn't want to eat the charge.

Also consider that different things scale at different rates. Spell damage will scale differently from white damage because of different things affecting them. If your melee raid buffs give you +20% melee damage, but your spell raid buffs give you +40% damage, then your proportion of spell damage is being undervalued in a dummy test. Your weapon stats may therefore be more important than the DPS of them, depending on how the different factors scale. This is especially important in regards to pets, as they gain some contribution of some stats but not of others. If you have two melee weapons, one that is 100 DPS and has 50 strength (~100 AP = 7 DPS) and another of 110 DPS and 0 strength, then you may overvalue the one with pure DPS if you do a test without raid buffs and without pets. The other one may actually pull ahead because of its contribution to pet DPS and spells in a fully raid buffed test with pets.

Also look at the difference in DPS values being posted in raid buffed situations and those on dummies. The average values I've seen from dummy DPS vary between 2k and 3k, while the raid buffed situations vary between 5k and 7k. This variance due to short term effects, lucky crit rates and other thing are exactly the kind of things that can completely throw off dummy DPS tests.


Disclaimer: This is an example. Although the methodology of thinking about what may affect your dummy DPS compared to raid DPS is correct, the actual conclusion (slower is better because of frost strike) is not a real conclusion, merely an example.

Take, for example, the spell portion of your damage (IT/HB/DC as the major parts). Spells are affected by +3% hit, +13% damage, +AP contribution, and 8% haste. A build including Frost Strike may significantly underestimate the portion of damage they're doing with FS on a dummy because of misses and lack of haste giving an extra GCD per cycle which means more runic power dump time is available. By doing dummy tests you may come to the conclusion that a fast weapon in the main hand is just as profitable as a slow weapon because frost strike damage is not a signficant portion of your damage, while in an actual raid buffed situation you have a lot more opportunities to runic power dump, therefore you get more damage from frost strike and a slow weapon does edge out a fast weapon.


You should be thinking about these ancilliary effects whenever you do a test and trying to minimise them as much as possible. If you cannot minimise them, think about their impact and whether it is negligible or not. If not, then estimate what kind of effect you think it may have, and include it in your results.

I mostly do heroic dungeons and 10 man raids. I'm the only DK in my guild. Many AoE packs don't survive past 2 or 3 HBs, but Epidemic ensures that if it does go to 3 HBs, every mob is diseased without me having to use a blood rune on Pest.

Plus, I don't like the perma-ghoul, so shifting those point to Ravenous Dead is a waste to me. Epidemic affords me flexibility.

But, perhaps you're right and I'm gimping my DPS by not taking the ghoul. If I got the ghoul, I'd flip those points to RD.

Perma-Ghoul with NoTD is simply amazing - it should be a staple for this spec. You will always be stuck at 32/38, with one floater point. I went with CorpEx because it will eventually get a buff, so I'm feeling myself into the mechanics of it now (plus it's fun making bosses disappear), but you could throw that point in Epidemic for an extra 3 seconds of diseases if you wanted - I assumed you were a 25 man raider.

I would like to second the sentiment of this post, Dukes has the right of this. 3 minute tests, or even 9 minute tests are absolutely not the correct way to make definitive statements. About 15-20 pages back we were discussing updating the DPS compendium page regarding Dual Wield specs, I for one think it is time to put our heads together and come up with a new "Basic Info" post regarding what we can PROVE about current DW topics.

Last page we had an excellent break down of Fast vs. Slow weapons, I think the only small error in it was the assumption that misses don't proc KM. (They do)

Yet on this page here we are again, discussing 3 minute tests on ebon hold dummies regarding weapon speed.

Would anyone be interested in focusing our efforts on creating a new "OP" for this thread? perhaps a "3.0.8 Dual wield specs" post can be created to replace this thread? Sure, it may take a moment but honestly this thread has become a loop. I feel sorry for the moderators, and I bet they just hate this thread by now.