As far as army goes keep in mind that first, as Otou mentioned, while army lasts 40 seconds, it's only active for about 35, as the stupid ghouls take a few seconds before they do anything, which means it's closer to ~42% increase in duration then a 37.5% increase. This is offset by the fact that you'd normally have bloodlust/heroism up for the duration of army, in addition to a pot, trinket or two, and any other cds. These are all likely over by the end of the normal army duration, leaving the ~42% increased duration to be a significantly lower damage increase.

How low? A quick simcraft unholy BiS with army taken out shows us that unholy bursts at 150k, and averages 104k, and dips to about 95k at the 40 second point in a fight. Assuming one armies on the pull, the dmg of army would drop to about ~70% of what it is during the first 40 seconds with a lust, pot, and cds/trinket procs. Another sim with army shows that it averages 81k dps for unholy, and 70% of that would be 56.7k for an extra 15 seconds, or 850k every 10 minutes.
On a fight exactly 10 minutes long, that is a 01.70% increase, on a fight 5 minutes long, it's a 03.31% increase, and on a fight with a 100 second duration, it's a 08.57% increase.


As far as potting before army, it's a bad idea because army ghouls change dynamically with your stats, meaning that potting after casting army still grants the ghouls the strength boost.

EDIT: The Zandalari does physical damage and doesn't benefit from Unholy mastery.

Ghostcrawler edited his post, the summoned Zandalari does not scale with unholy.

GC added some more information:

Death Knight
- We are changing the 4pc Frost / Unholy set bonus from Army of the Dead to Killing Machine and Sudden Doom proc more often. We don't have numbers yet and it's possible they won't work out. In addition to all the concerns mentioned, it was a really valid point that Army of the Dead is hard to use in Raid Finder, where most of our players experience raids. You rarely get countdowns before a pull and players in general are used to yelling at DKs who use Army because they had a bad experience at some point with the ghoul taunt.


In short, it seems we can kiss any set bonus modifying AotD goodbye since they've realized not everyone thinks it gets used well.

Changing it to work on KM and SD has some interesting effects. Since this is the Unholy thread, I'll go over that.

Increasing SD procs grants even more resources through the L75 talent, will perhaps increase Super Timmy's uptime, and also fill in more GCDs. Chances are it'll actually boost haste since more autoattacks mean more procs and all that fun stuff, if you can spend it all.

The 2p being physical damage means mastery might see a decline relative to other stats. Since it's already on the bottom for level 90 though, it doesn't change much.

It's more likely to lower the relative value of haste, not raise it. Haste provides four benefits to a death knight. First, it increases your swing speed, giving you more white damage (auto attacks). Second, those extra white attacks give more opportunities for passive abilities to proc. Third, it provides the same benefit to minions. Fourth, it speeds up rune recharging.

The contribution of haste to the first and third of these damage pieces will be the same. But with more free death coils, these will make up slightly less of your total damage. So this lowers the value of haste, but not by much.

The contribution of haste to the second piece will go up slightly. More auto attacks means more chances for sudden doom procs, and a higher proc chance makes this more important.

The fourth piece is where things get interesting. More haste means more runes. Spending more runes means more runic power. More runic power means more death coils. It all adds up to less time waiting to use a damaging ability. Once you become gcd-capped, more haste doesn't help you this way at all. And since you don't have resources available at a steady rate, you're probably already occasionally gcd-capped. More free resources will amplify this effect.

Patch 5.2 goes Live today, so I'll be updating the Wowhead profiles to reflect Heroic Thunderforged gear as soon as possible. I'll also be updating the meta-gem section to reflect the inclusion of the Capacitive Primal Diamond.
Keep in mind that losing the 3% increased Critical effect (since the Capacitive Diamond doesn't use it) will mean a slight devaluation of the value of the stat, but likely not enough to actually have its place in the stat priority disrupted.

Apologies if I'm repeating someone, but I haven't seen this in my quick scan of the comments and the front of the post hasn't been updated to reflect this change:

1) Plague Strike now applies both Frost Fever and Blood Plague to a target.
2) Icy Touch now procs Reaping.

This means that when we enter an AoE situation, we no longer want to use Festering Strike to convert our frost runes into death runes. Instead, we'll want to just go ahead and apply diseases with either Outbreak or Plague Strike, then use BB/IT twice in conjunction with Death and Decay before continuing into Blood Boil spam. So the rotation or priority should look kind of as follows:

Diseases (Outbreak or PS), spread with BB (assuming Roiling Blood talent), Death and Decay
IT, BB [if FFBB]
IT, BB [if DFDB, at least one D on cooldown]
DC as needed when reaching RP cap
unholy rune spell as needed (SS/SR/Dark Transformation/DnD)
BB spam [if UUDDDD, one U on cooldown]

I hate to ask, but in your Best in Slot lists you have Uroe and Zerat as the heroic best in slots, wouldn't the Greatsword of Frozen Hells be a better itemized sword for us?

Eh, I don't know... Jackwar @ The Maelstrom - Community - World of Warcraft
I don't think it's as rare as you think.

With the changes to RPPM trinkets and the very real existence of the Festerblight playstyle (covered in Mendenbarr's "Advanced DK tactics" post), it looks like Spark of Zandalar has been displaced by Fabled Feather of Ji-Kun as our BiS trinket. Regardless of whether or not someone is actively pursuing the Festerblight playstyle, Feather's potent proc puts it considerably ahead of any other alternative this tier.

Unholy Gargoyle mechanics:

30 second duration, 3 minute cooldown, casts gargoyle strike with a base cast time of 2.000 seconds. This scales with the players spell haste linearly, eg. if you have 20% spell haste, the cast time becomes (2/1.2) or 1.666 cast time. However, despite the expected 15 shots with a cast time of 2 and a 30 second duration, gary only manages to get off 11 casts with 0 haste.

Testing at first indicated that, no matter his cast time, he would always get off 11 casts. But then...

Spell haste	Cast Time	Casts
00.00%	2.000	11/10/9
19.08%	1.679	11/11/11
19.18%	1.678	11
21.03%	1.652	11/11
21.41%	1.647	11
21.43%	1.647	11/12
23.31%	1.622	11
23.60%	1.618	11/11/12
24.18%	1.611	12
24.29%	1.609	(11) * 8/(12)*4
24.48%	1.607	12
24.88%	1.602	13/14
25.01%	1.600	15/16
25.04%	1.599	12/12/13/14
25.26%	1.597	16
25.39%	1.595	16
27.27%	1.571	11
27.34%	1.570	15
27.36%	1.570	16
27.67%	1.567	14
28.71%	1.553	16/16/16/17
29.64%	1.543	(16)*5/15
30.51%	1.532	15/15/16
31.28%	1.523	(16)*6
42.77%	1.401	13 (using zerking)
52.00%	1.315	17
62.95%	1.227	16
65.57%	1.208	16
69.66%	1.179	17/17
111.83%	0.944	16 (lv 80)

Unholy seems to have a haste breakthrough! At this value of spell haste, which SEEMS to be between 27.27 and 27.34, which is 9015 and 9044 haste rating raid buffed, gary starts to scale with haste. This is a major breakthrough, because all of that haste starts to apply to gary at once, and napkin math puts it at as much as a 2-4k dps. This needs more research to find the exact point and confirm the findings. I'll be compiling data, but in-game it's difficult to test every single level of haste.

So here is what we are going to do. Spend 5 minutes to respec to unholy (unless you already are), go to a training dummy, and cast Gargoyle with recount up. Please PRIVATE MESSAGE me the following:
Haste rating.
Spell haste in spellbook.
Cast time of gary, (less digits are ok, easy for me to confirm this by looking at spell haste)
The number of casts gary gets off.

Tips!
The most important numbers are between 9k and 9.1k haste.
Much much higher numbers are also important, to see if there are multiple breakthroughs.
5% spell haste does effect gary.
If you want to test a higher number, try getting a shaman to cast bloodlust on you.
DO NOT use darkmist vortex, it will only be up for part of gary's duration, and screw with the number of casts.
Bloodlust only works if it's up for the entire duration of gary.
The more numbers we have, the better.


Credit for discovery of this little quirk goes to Devodante.
Thanks for the numbers from:
Frozendekay
lilucky136
Nangz
Vyoh
Minxyqt
Zaiku
markkali
php
Discodelya
Jujope



Update:
Still collecting more numbers, but this set of data indicates there is a haste breakpoint, and it's after 27.34%, but I'm still not sure where it starts. Gary's AI works in "ticks", every X ms, he decides what to do. If he finishes a cast, and there isn't a tick for another 300 ms, he'll just derp around and do nothing for that 300 ms. The "breakpoint" appears to be where he finishes a cast right before a tick happens.

Also seeing evidence from multiple places that his first cast always has a time of 2 seconds, regardless of haste. I'm not sure what this indicates, perhaps that it takes him a certain amount of time to scale to his master's stats.




Update 2:
https://twitter.com/Ghostcrawler/sta...91610425409536
I'm done with this for the time being.

Wouldn't it be a better choice to switch in the Tier15 Chest instead of [Carapace of the Core] and replace the Tier15 Legs with the heroic thunderforged version of [Legplates of the Lightning Throne]? Can't find a heroic thunderforged version of it, yet, as wowhead seems to be somekind of slow regarding their respective versions. I may be wrong, but I doubt it at this point, as I ran this already in chardev and I'd, personally, prefer a haste/expertise setpiece above a crit/mastery one, if the offsetpieces are almost on par (except the fact that you'd change two red for two yellow sockets).