Forgive me for necroing this, but I wanted to thank you for putting together this information. I also wonder how much this holds true in 5.2, with the various balance adjustments and the better gear now available.
It still holds true. See below.
Originally Posted by dkdk
The OP did touch on the fact that as we gear up and have more haste it will actually make it more likely to overwrite a KM proc if we don't use it right away. Meaning in current gear waiting will be more likely hurt you and on top of that have an even smaller dps boost if you were able to do it perfectly (very unlikely).
It's possible that I overestimated haste's true impact on the scaling here, but more haste is certainly going to cause problems by shortening your window of opportunity. However, this only shows itself at higher haste values than those being simmed, but the sim is also bypassing latency and has a stellar reaction time.
Stalling in 5.2:
It's still not worth it. The gain is even smaller.
The gear being used was pulled from the EJ 2H BiS profile (minus the legendary gem) with reforging done by myself. You can see the total stat values (from gear+consumables) being used on the last sheet in the spreadsheet.
As a side note, if you've noticed that the pre-execute phase DPS is higher than the execute phase DPS, you're not insane. This is simply because Army of the Dead is being attributed to the pre-execute phase DPS.
I investigated two possibilities with 5.2.
1. Proper stalling as described above in the OP.
2. Proper stalling as described above in the OP, but including stalling for more KM-SR's via DPS T15-4pc.
Looking at the first, it's clear that the gain has diminished since i509. Proper stalling only amounts to a 0.83% gain. The important factor here is the ratio of damage dealt by FS to that by OB--with more mastery (which you acquire naturally as you gear up), FS becomes stronger relative to OB. Thus, the gain from a KM-OB is reduced. Other than that, there's not really much else to say that hasn't been covered already. I anticipate this gain shrinking further into the expansion pack, as you will continually gain more mastery (provided that FS/OB damage coefficients stay the same).
For the second test, I'm sure people have been wondering given how much damage per rune a single SR does. For the execute phase being simmed, it is possible to fit in 23 SR's. The normal priority and first stalling priority are able to use 20-21. The KM-SR stalling priority only uses 17, and 10 of these are KM-SR's compared to 6 in the other priorities. So, it is effective at stalling, however, if you consider gaining 4 critical SR's, but losing 4 normal SR's, where do you stand?
You simply break even: 17 + 4*2 - 4 = 21, and the data supports this. For every second you delay using SR to try for a KM-SR, you are wasting time that could be going toward cooling another SR (i.e., a cumulative wait time of 6s causes you to lose 1 SR because it has been effectively pushed out of the fight window). Thus, similar to KM-OB, although it seems appealing at first glance, the opportunity cost is mounting up over time to eliminate any gain.
Therefore, even though proper stalling for KM-OB still leads to a miniscule gain (+0.83%), for the 7 reasons mentioned in the first post, I still recommend that you do not stall for KM. Without significant changes, you will likely continue to ignore it for the rest of the expansion.
Does "ignore KM procs" (as they relate to FS and OB in a vacuum and ignoring rp capping) basically imply that OB>FS in your rotation at all times, in this scenario?
The act of ignoring KM does not imply this, but the overall result is the same. The sim was already prioritizing N-OB > N-FS even when KM was being checked (and obviously KM-OB > N-FS), so removing the KM check doesn't change the fact that N-OB (and therefore just OB) > N-FS was already prioritized.