[Resto] TreeCalcs - Page 9

lolofora · 10/05/12, 8:42 AM

I implemented "Soth" model with last version of Hamlet: filehosting.org - download page for TreeCalcs_121005_ModelSotH.xls
I updated too spirit/mp5 trinkets.

The result is surprising, Incarnation >> Soth

Parameters:

- 4T12
- glyph WG
- Sync SM+WG to 12 sec

• Hamlet · 10/06/12, 12:44 PM

Efflo is no longer based on the underlying Swiftmend. It doesn't double-dip mastery or benefit from the Swiftmend's crit, but the ticks can crit now. Probably doesn't double-dip Tree of Life, although I forgot to check that.

Wild Mushroom is a direct heal for Harmony purposes.

Efflo is doing a tiny fraction higher than it should--don't know why, just remembering for spreadsheet.

If you're testing in town, remember that PvP power will increase your heals (I think by half the listed value on the charsheet).

♦ Earen · 10/06/12, 1:44 PM

Something else to check, I think someone was reporting that each tick of efflo was affected individually - so if you popped a trinket halfway through efflorescence, for example, it updated the value of the heal at the point your spell power increased. I have not looked at this at all yet, so not sure. Just passing it along.

Just so I know how you determined the double dipping, what did you do for the testing? I stood in org one night for about an hour and played around with Efflo - but I must not fully understand what to look for in that regard. So if you are game to explain, I'd be curious to know!

• Hamlet · 10/06/12, 2:05 PM

Yup, that's true. Easy to test by just casting Swiftmend and watching Lightweave proc partway through.

So you have to know how to compute what you expect a Swiftmend or Efflo size to be based on your current spellpower and mastery. TreeCalcs can do this if you just manually drop your current spellpower and mastery into the right spots in Basic Calcs (don't save after you do). Or, just go to wowhead: Swiftmend - Spell - World of Warcraft

So you know that a Swiftmend will be (13966 + 1.29*spellpower)*(1 + Harmony). Read spellpower and harmony straight from your char sheet (take off proc effects and PvP power first). If your in-game SM amount matches that number, you know you're all calibrated. Then you can multiply that by 0.12 to find your expected Efflo tick, and it should match what you have in-game (or a few points higher for some reason, like I mentioned).

Unlike in Cata, you no longer have to multiply the Efflo tick by Harmony again to get the right outcome, and also, the Efflo size in unchanged if the Swiftmend crits.

• Hamlet · 10/06/12, 8:18 PM

Yeah, I see what you're doing with the SotF stuff. It's a bit complicated, but for right now we're just using it for our TC/guide purposes anyway and lots of things aren't user friendly. It does seem to show that it's hard to make SotF really good.

Notes to self:
Remove mastery from Efflo
Add mastery full-time to WM
Don't double-count mastery in NS-RG

• Hamlet · 10/09/12, 6:26 PM

Slightly rough but usable I think--simple rework of filler spells. Now instead of "Spend X% of leftover time on Nourish, spend rest on filler spell A," it's:

1) Spend X% of leftover time on filler spell A,
2) Spend Y% of leftover time on filler spell B,
3) Spend what's left on nothing (it's up to you to make X+Y less than 100).

Only quirk right now: mana/regen gains are evaluated by translating all excess mana into the first filler spell.

I think this lets people more easily examine something that roughly resembles current spell usage, which should make stat weights a little more informative.

e2: forgot to make "Nourish" an option. It's in there, just forgot to add to the dropdown menu (anyone who knows data validation ranges can change that and try it out)--will fix next time.

lolofora · 10/12/12, 6:17 AM

Windsong seems underestimated in TC.

In fact, every proc is between 25 or 30% uptime, an average gain of 375 or 450 for proc mastery and proc critical, instead of 127 in TC.

In addition, current uptime "Proc Haste" is 8.5% in Haste tab.

• Hamlet · 10/12/12, 9:02 AM

Yup--that was changed a day or two ago (in-game I mean). I saw you posted some links on the other thread but they didn't seem to work for me, did you do any test of the proc rates/ICD's?

lolofora · 10/12/12, 10:37 AM

There are 0 ICD
Buff can refresh

Expression Editor, Wol:
spellID = 104510 AND sourceName = "PlayerX"

Result :
[21:00:34.803] PlayerX gains Windsong from PlayerX
[21:00:46.904] PlayerX's Windsong fades from PlayerX
[21:01:18.323] PlayerX gains Windsong from PlayerX
[21:01:30.211] PlayerX's Windsong fades from PlayerX
[21:01:57.648] PlayerX gains Windsong from PlayerX
[21:01:58.459] PlayerX's Windsong is refreshed by PlayerX
[21:02:10.479] PlayerX's Windsong fades from PlayerX
[21:03:20.828] PlayerX gains Windsong from PlayerX
[21:03:32.864] PlayerX's Windsong fades from PlayerX
[21:04:11.069] PlayerX gains Windsong from PlayerX
[21:04:11.098] PlayerX's Windsong is refreshed by PlayerX --> less than a sec gap
[21:04:23.139] PlayerX's Windsong fades from PlayerX
[21:04:35.606] PlayerX gains Windsong from PlayerX
[21:04:47.630] PlayerX's Windsong fades from PlayerX
[21:04:50.534] PlayerX gains Windsong from PlayerX
[21:05:02.636] PlayerX's Windsong fades from PlayerX
[21:05:12.985] PlayerX gains Windsong from PlayerX
[21:05:24.984] PlayerX's Windsong fades from PlayerX
[21:06:23.319] PlayerX gains Windsong from PlayerX
[21:06:35.334] PlayerX's Windsong fades from PlayerX
[21:07:49.352] PlayerX gains Windsong from PlayerX
[21:07:58.596] PlayerX's Windsong is refreshed by PlayerX
[21:08:07.016] PlayerX's Windsong is refreshed by PlayerX
[21:08:11.046] PlayerX's Windsong is refreshed by PlayerX
[21:08:23.083] PlayerX's Windsong fades from PlayerX --> Multiple refresh, buff active ~34 sec
[21:08:23.906] PlayerX gains Windsong from PlayerX
[21:08:35.961] PlayerX's Windsong fades from PlayerX
[21:08:36.812] PlayerX gains Windsong from PlayerX
[21:08:48.836] PlayerX's Windsong fades from PlayerX
[21:09:12.206] PlayerX gains Windsong from PlayerX
[21:09:24.222] PlayerX's Windsong fades from PlayerX
[21:09:29.504] PlayerX gains Windsong from PlayerX
[21:09:41.654] PlayerX's Windsong fades from PlayerX

• Hamlet · 10/12/12, 10:46 AM

Ah, nice. I'd only tested a few minutes and never saw a refresh, only an overlap of different stats. And man, 18 procs in 9 min of that one buff alone.

This was as Resto with multiple targets? This thing is going to be quite good right now.

lolofora · 10/12/12, 10:55 AM

yes it is a real fight (multiple targets) with a resto druid (World of Logs - Real Time Raid Analysis)

• Hamlet · 10/12/12, 5:24 PM

Windsong updated to match Moonkin sheet. Probably a significant underestimate here due to all the HoTs we have going (the one log I've seen so far would confirm this).

Also made Ember meta no longer add 120 base MP5. I guess I should verify in-game, but Armory doesn't show that it does.

Arentios · 10/15/12, 1:30 PM

Somewhat comprehensive post on Windsong and Elemental force. They're 2/10 PPM respectively scaled by haste and using a new system that adjusts individual tic proc chance by time since the last proc. Does not sound the most pleasant thing to model beyond the straight 2 PPM * haste approximation.

Windsong and Elemental Force Information - Forums - World of Warcraft

• Hamlet · 10/15/12, 1:46 PM

Okay, read it. The key is that, according to the description, it's not time since the last proc, it's time since the last chance to proc. So it's basically trying to enforce a particular PPM rate (before haste scaling). In other words, what "PPM" means now is that each passing unit of time gives a little bit of proc chance, which is all aggregated into the next tick.

So, 2 PPM * haste should actually be quite accurate if the system works as described. (Note that what the sheet has right now is 3 PPM * haste--would be a simple adjustment).

In fact, the distribution of procs in time should be given by a Poisson distribution - Wikipedia, the free encyclopedia . In practice modeling that level of detail is probably unnecessary (ignoring overlaps and just taking uptime as PPM*duration*haste/60 should be fine), but it wouldn't even be that complex.

• Hamlet · 10/16/12, 4:25 PM

Windsong. Also I just linked a post on the WC thread where I elaborated on the math in the above post, for people who like the math.

Note that if we take the description in the blue post straight, we're expecting around a 15-16% uptime at high haste levels. If people are seeing substantially higher than that, may have to reexamine what's going on.