For one thing, healing is still very easy iff your DPS and tank are competent. The days of saving bads standing in fire - over. Or at least put on hold, pending better gear and familiarity with the new palette of healing tools.
One thing I'm particularly interested in is the new relative ordering of stats for healing. Mastery in particular is quite tricky, as it depends on conditions outwith your gearing strategy to a large extent and this makes it hard to model.
As I'm messing around with Clojure at the moment, I decided to whip up a little 'healing simulator' that spam-heals a single target taking damage with some simple profiles, using a minimal priority setup for healing spell selection. Whilst this is in many, many ways incomplete and buggy, I thought the results were interesting enough to leave them here (and when I iron out a few bugs, I'll post the code somewhere too).
I used my shaman's base stats as of a day or two ago, in a mixture of pre-heroic blues and a smattering of heroic and rep items, and then simulated after adding a hefty chunk of one stat - 1000 ratings worth, although the ordering seems consistent for smaller increases and after changing the base stats somewhat.
First, bumpy damage. This attempts to model somewhat low, consistent damage, with periodic large spikes, which fits a number of fights if painted in sufficiently broad strokes.
| Bumpy | Intellect | Mastery | Crit | Haste |
|---|---|---|---|---|
| e-hps change (%) | 9.56 | 5.06 | 4.34 | 7.75 |
| mps change (%) | -1.61 | -0.56 | -2.34 | 6.62 |
Next, essentially random damage. This is some kind of nightmare scenario, where the tank's relative health is completely unpredictable from one GCD to the next. The simulator, unphased, simply makes its best guess as to an appropriate heal from one moment to the next.
| Random | Intellect | Mastery | Crit | Haste |
|---|---|---|---|---|
| e-hps change (%) | 8.09 | 5.67 | 4.24 | 6.92 |
| mps change (%) | -1.76 | -0.16 | -2.82 | 6.28 |
So, immediately it can be seen that in this very crude scenario, intellect is king for e-HPS. It also lowers MPS, by allowing efficient heals to do more of the heavy lifting and having some tiny effect on Improved Water Shield procs. Considering it'll improve spirit-based regen and replenishment, intellect is looking really quite tasty.
Next up, haste is a good throughput stat, but that throughput comes at considerably increased MPS. So far, so unremarkable, although it's interesting that intellect appears to provide a stronger bonus to effective healing under the new healing model.
Mastery is surprisingly not-bad. It's not quite as good as haste, but it's not far off, and doesn't have a downside. Sure, it does nothing for you when you're healing people at full health, but in that situation so does every other stat - they just generate extra overhealing. Considering the variance I'm seeing in the results, it has negligible impact on MPS, although a smarter sim could probably save more mana here.
Crit is, to a surprising degree, not a great throughput stat. It does provide a decent chunk of mana saved, and the bonus of Ancestral Healing can't be underestimated, but it still feels underwhelming.
The result of all this: it's probably irrelevant. I'll wait for some actual raiding experience to inform my iteration of the simulator, but healing really isn't a problem suited to this kind of thing. Still, the braindead results do at least convince me that 'haste-stacking uber alles' is out of the window for the foreseeable future, Mastery is probably not that bad (I think I'll stop reforging it, although the JP stuff has a ton), and intellect is currently looking delicious.
AH uptime, pretty realtime graphs, and some priority-queue optimiser are things I'd like to add to the sim. I'll see how that goes; for now I need Earthen Ring rep!
No comments:
Post a Comment