Originally Posted by Malleus
Exemplar: why do you suggest refreshing Inquisition at 1 or 2 HP to get more DivPurp procs when in your post immediately prior you'd said that we'll be even closer to GCD-locked in Mists? Every Inq-1 or Inq-2 we did would be in place of a TV and delay the next TV if it didn't proc DivPurp. There's no way that wouldn't be a colossal DPS loss.
|
CS and HotR share CD - lots of free space, 1 or 2 HP Inq are possible.
CS and HotR do NOT share CD - GCD lockage. 3 HP finishers only, please.
I did not use ZotC, either. I went with 3.5 speed weapon, since I did Haste calcs and it would gain less benefit. Erring towards a sort of worst case scenario for the talent.
Can we agree Haste will only improve the DivPurp side of the equation? Faster autoattacks, thus more Exo procs. Shorter CS/Judge CD, thus more HP generation, while Finisher GCD does not reduce.
So, let's use a 3.6 speed weapon, no HotR usage (it's exclusive to CS), and no Haste at all.
So HP generation is CS, Judge, and the Exo procs. DivPurp is the 17.65% chance (the baked in recursion).
Repeating sequence of CS, Judge, Filler, CS, Filler, Filler. (Note this slights Judge, since we could slot it in the last filler with a longer sequence before repeat, but this is dirt simple for napkin math).
Every 9 seconds you get 2 CS, 1 Judge, and (9/3.6 = 2.5 melee. Take 2.5 melee * 20%) 0.5 Exo. Total of 3.5 HP.
60 seconds / 9 seconds = 6.67 cycles/min.
6.67 cycles/min * 3.5 HP/cycle = 23.33 HP/min.
23.33 HP/Min * 1 Finisher/3 HP = 7.78 Finisher/Min.
7.78 Finisher/Min * 17.65% DivPurp/Finisher = 1.37 DivPurp/Min
So, zero Haste, no ZotC we get an average of 1.37 DivPurp (full finishers) per minute.
Malleus, I do not understand your returns. One usage of GoAK grants 7 extra finishers. GoAK has a 5 min CD.
From my above numbers we get 7.78 Finisher/Min, which can be flopped to 7.71 seconds per finisher. Hit Inq, immediately start hammering finishers. Anything up to 307 seconds is your first CD. That calculated, let's ignore the 7 seconds so we have a nice clean 300 seconds (5 min) for napkin math.
300 seconds (or 5 min) grants 7 finishers at times 0, 1.5, 3, 4.5, 6, 7.5, 9.
7 finishers / 5 min = 1.4 Finisher/min
This drops until the 10min mark when you can re-use.
| Minutes | Total HA Finishers | CS no HotR Total DivPurp Finishers |
|---|
| 1 | 7 | 1.37 |
| 2 | 7 | 2.74 |
| 3 | 7 | 4.11 |
| 4 | 7 | 5.48 |
| 5 | 7 | 6.85 |
| 6 | 14 | 8.22 |
| 7 | 14 | 9.59 |
| 8 | 14 | 10.96 |
| 9 | 14 | 12.33 |
| 10 | 14 | 13.7 |
Holy Avenger wins every timeframe. You get a tie at about 5:06 if you delay GoAK by those 6 seconds. If we'd delayed that first 7.71 seconds, then there's about a 1-2 second window where DivPurp wins! Start throwing in Haste effects and both the tie-time and window for DivPurp to win will grow larger. This is the situation where cheesing some 1 HP Inq would get you some gain. I won't math that - based on the calculated HP generation, you'd probably clip Inq to perform it and could as easily lose more than you gain.
Now, let's go to an environment where you could use both CS and HotR. New rotation of: CS, Judge, HotR, Finisher. Depending on damage we either replace HotR with Exo when it procs, or replace Judge with Exo, then Judge replaces HotR - end result is none, we still get 3 HP and a finisher every cycle.
Every 6 seconds you get three HP, whether CS/Judge/HotR, CS/Judge/Exo, CS/Exo/Judge, or Exo/CS/Judge.
60 seconds / 6 seconds = 10 cycles/min.
10 cycles/min * 3 HP/cycle = 30 HP/min
30 HP/min * 1 Finisher/3 HP = 10 Finisher/Min.
10 Finisher/Min * 17.65% = 1.76 DivPurp/Min
On Napkin we pull ahead at 4 minutes, but that's not accounting that actually using the DivPurp pushes back your cycle - there were no free GCD! Not one! So every DivPurp finisher eats 1 HP by using a HP-generating attack's GCD. So let's apply recursion a few times.
1 DivPurp = -1 Finisher
10 Finisher/Min - 1.76 Finisher/Min = 8.24 Finisher/Min
8.24 Finisher/Min * 17.65% = 1.45 DivPurp/Min
Recalc:
10 Finisher/Min - 1.45 DivPurp/Min = 8.55 Finisher/Min
8.55 Finisher/Min * 17.65% = 1.51 DivPurp/Min
Recalc:
10 Finisher/Min - 1.51 DivPurp/Min = 8.49 Finisher/Min
8.55 Finisher/Min * 17.65% = 1.50 DivPurp/Min
Looks like further recursion will just narrow decimals. We'll call it 1.5 DivPurp/Min.
New table time.
| Minutes | Total HA Finishers | CS+HotR Total DivPurp Finishers |
|---|
| 1 | 7 | 1.5 |
| 2 | 7 | 3 |
| 3 | 7 | 4.5 |
| 4 | 7 | 6 |
| 5 | 7 | 7.5 |
| 6 | 14 | 9 |
| 7 | 14 | 10.5 |
| 8 | 14 | 12 |
| 9 | 14 | 13.5 |
| 10 | 14 | 15 |
DivPurp pulls ahead at about 4:40. So your window is 20 seconds (your new HA was at 5 min and 0.01 seconds). Delay your GoAK and the window expands a bit, but not much. Wait those first 6 seconds to put up a full Inq and we've got a 26 second window.
26 seconds / 300 seconds = 8.67% chance that DivPurp would pull ahead. Just going on averages (like the rest of the napkin math), that's pretty lousy.
Let's go so far as to maximize. 50% Haste drops our CS, HotR, and Judge GCD to 1 second, and their CD to 3, 3, and 4 respectively. Entire cycle becomes 4.5 seconds. This would jump us to 13.33 Finisher/Min. Playing with recursion we'll find it winds up more 11.33 Finisher/Min and 2 DivPurp/Min.
So, in this insane world where we had a finisher every 4.5 seconds, DivPurp could be ahead around the 4:00-5:00 window (it tied at 3:30). Even with a larger window, I think the drawback is the "winning" windows for DivPurp are the least likely timing for a fight to finish.
End results: Holy Avenger seems a clear winner. Additionally it has other benefits. If you need a specific burn phase, being able to hold GoAK and use at that time for the burst would hands down beat the utility your averagely distributed DivPurp. HO is not at the mercy of the RNG - statistically DivPurp will underperform the above numbers 50% of the time.