 |
09/17/12, 12:08 PM
|
#466
|
|
Soda Popinski
|
They don't chain cast, so you lose casts that way. They're only getting 6 casts each time, fading during the 7th cast (tested with full Haste and naked, same result). Doing some tests now:
Raid buffed at the Patchwerk dummies, no meta gem crit bonus. Level 90 Worgen, boss level dummy. 5% magic dabuff present.
SP 19527, Haste 5103 (17.61%), Crit 826 (14.07%), Hit 4286 (12.61%). 5 Force of Nature casts.
Total Wrath casts: 90. 78 hits (min: 5183, max: 5317, avg: 5251), 9 crits (min: 10369, max: 10594, avg: 10453), 3 misses.
SP 1681, Haste 0 (5%), Crit 0 (8.46%), Hit 0 (0%). 5 FoN casts.
Total Wrath casts: 90. 71 hits (min: 1437, max: 1565, avg: 1502), 5 crits (min: 2940, max: 3025, avg: 2980), 14 misses.
Can't spend any longer on it now, but if you let me know some specifics you want looked at I'll check those later tonight.
|
|
|
|
09/17/12, 12:11 PM
|
#467
|
|
<Druid Trainer>
|
--NS affects DoC.
--Basic FoN model. Assumes the treants inherit your total haste, total crit, and total hit. e: assumes 6 casts per treant. Looks really weak.
--Basic Windsong model. Assumes 1PPM works as intended (each proc procs once per minute * haste). This seems to make it quite strong, but will have to work out assumptions carefully.
--Basic Elemental Force model. Assumes it's buffed by blanket spell damage bonuses, but still weak (but does it have a spellpower coefficient like DMC:V?).
Tecton: right now when you select Windsong it makes mastery and crit numbers look ugly since it just throws the computed values in with all the rest. What did we do in the past with trinkets with mastery/crit/sp procs? In any case, maybe best to make a table in basiccalcs that sums averaged effects of non-haste procs and adds them in after the front pages. That the way the front pages don't look weird, and also, should math people's character sheets.
e: Fixed the below, and FoN actually casting 6 casts instead of 3.
Last edited by Hamlet : 09/17/12 at 12:34 PM.
|
|
|
|
09/17/12, 12:24 PM
|
#468
|
|
Great Tiger
Night Elf Druid
Echo Isles
|
Very small error/fix: Rotation!D49: Incarnation is a shapeshift and has a full 1.5s GCD (no haste benefit).
|
|
|
|
|
09/17/12, 12:36 PM
|
#469
|
|
<Druid Trainer>
|
Btw, what's left for me to do, aside from trying to figure out Windsong better. Are there any weird trinkets I should know about? Still haven't looked at items really.
|
|
|
|
09/17/12, 1:31 PM
|
#470
|
|
Don Flamenco
|
Originally Posted by Hamlet
Btw, what's left for me to do, aside from trying to figure out Windsong better. Are there any weird trinkets I should know about? Still haven't looked at items really.
|
Haven't seen or heard tell of anything terribly unusual so far. I think Windsong is probably the biggest one out there.
|
|
|
|
|
09/17/12, 2:33 PM
|
#471
|
|
Von Kaiser
|
Hey Hamlet,
I'm trying to account for why Wrathcalcs rates Heart of the Wild ahead of Nature's Vigil. I think I've found the answer, but I want to make sure.
For the DPS calculations in WrathCalcs you are assuming one 3minute segment where cooldowns like Incarnation, CA, & NV are only used once, correct?
If so, your valuation of these talents would represent their minimum value, because it minimizes their up time. Therefore, Nature's Vigil might be the better option in fights that last 4 - 5 minutes because that would allow for the Cooldowns to be used twice and have a greater overall uptime.
|
|
|
|
09/17/12, 3:31 PM
|
#472
|
|
<Druid Trainer>
|
That's true, and that's always been how WC works--it models an infinite steady-state cycle. For a full picture, there's no way around having to combine the information from something like WC and something like Simcraft (the advantage of each would be a much longer post). e.g. even if common simcraft results show a high value for NV, it's important for readers to know that that relies on a particular window of fight length, so they understand fully where it's coming from.
The strength of WC is attaching specific numbers (that show the correct scaling patterns) to everything. And if you dig into it and make use of the various information other than the final DPS you can often answer more complex questions.
For example, if you wanted to estimate NV for a 4.5 minute Patchwerk fight, you could assume NV was giving you 4/3 the amount of DPS shown on WC and compare that to HotW (4/3 because you're getting 2 uses of NV in 3/4 the amount of time that usually takes).
|
|
|
|
09/17/12, 3:34 PM
|
#473
|
|
Soda Popinski
|
Ok, so to expand on the FoN scaling after installing a castbar addon that works on beta, they definitely do scale with Haste.
0 Haste: cast time 2, 6 casts.
5% Haste: cast time 1.91, 6 casts.
7.21% Haste: cast time 1.87, 6 casts.
12.01% Haste: cast time 1.79, 6 casts.
17.01% Haste: cast time 1.70, 6 casts.
21.23% Haste (NG): cast time 1.65, 6 casts.
23.29% Haste (NG): cast time 1.62, 6 casts.
25.34% Haste (NG): cast time 1.60, 6 casts.
Treant casts started getting weird here, with some casts being back to back and some having a delay between them. They also got slightly out of sync (2 would cast together, then the third just after them, meaning you can lose a cast if you're at one of the breakpoints). I tested the following results 3-4 times each, so some of them have multiple results.
25.64% Haste (NG): cast time 1.59, 6 casts.
26.34% Haste (NG): cast time 1.58, 6 casts.
26.77% Haste (NG): cast time 1.58, 6/7 casts.
26.87% Haste (NG): cast time 1.58, 6/7 casts.
27.31% Haste (NG): cast time 1.57, 7 casts.
27.34% Haste (NG): cast time 1.57, 7/8 casts.
27.78% Haste (NG): cast time 1.56, 7/8/9 casts.
28.30% Haste (NG): cast time 1.56, 8/9 casts.
28.81% Haste (NG): cast time 1.55, 9 casts.
35.25% Haste (NG): cast time 1.48, 9 casts.
Well, that's annoying. Basically, weird pet behaviour means when you are close to gaining another cast you'll sometimes get it, sometimes not and sometimes lose a cast. Not sure how best to handle this for the sheet? It's possibly one to write up on the class balance thread on the US forums, if someone has access.
As an aside, they retain your haste from the beginning of the cast (i.e. if you cast them, then procced NG after a couple of their casts, they'd continue casting at the same rate). They do not appear to gain any benefit from Eclipse. I would like to take this opportunity to thank whatever designer put in Astral Communion, made testing those a whole lot less annoying!
Trinkets are all pretty straightforward, I think putting an expected time to proc for Light of the Cosmos (45 second ICD, 15% chance on DOT tick to proc) would be useful, though.
And yeah, that'd be a good way to handle the procs/etc. Let me know when I'm ok to make some edits Hamlet and I'll knock something together.
|
|
|
|
09/17/12, 3:41 PM
|
#474
|
|
<Druid Trainer>
|
You can do it now--I don't think I'm going to do anything with that FoN at the moment (maybe change 6 to 7 since that's probably the common window assuming you always cast with NG). But since they're clearly inferior anyway, the details aren't a priority (will try to give feedback about it though).
Expected time to proc for trinkets will be something to glance at, will do it next time--usually not very hard. Don't have to be too detailed with long-ICD trinkets since the scaling with haste is negligible. But good to have some delay in there so they're not globally overvalued.
|
|
|
|
09/17/12, 3:57 PM
|
#475
|
|
Von Kaiser
|
That's all true. I wasn't implying that there was something wrong with WrathCalcs. I'm just in the process of writing my guide and wanted to be able to explain why the results from Wrathcalcs are different then what is generated in Simulation Craft.
|
|
|
|
09/17/12, 4:09 PM
|
#476
|
|
<Druid Trainer>
|
Yup. And WC is far enough along now that it's probably useful to keep comparing results more.
|
|
|
|
09/17/12, 6:27 PM
|
#477
|
|
Soda Popinski
|
Done
- Table of procs (with space for extras to be added below) and a total for each procced stat listed. I've moved the benefit for Synapse Springs, Lightweave, Jade Spirit, Windsong and all trinkets over here. Totals added into the stats on the calcs page.
- Double benefit from Synapse Springs fixed.
- Belt buckle renamed.
- Fixed trinkets highlighted as having new values earlier in the thread.
- Alchemy bonus added under Misc benefits to avoid double counting if in Target Dummy mode.
Regarding FoN, I was getting 9 casts in the crappy PvP gear. With the profile loaded on the sheet, we're getting up to 45.39% Haste under NG, so it will likely have more casts than that (possibly 11 or 12?). It's within a few hundred if you set it to 9 casts, which should be fairly easily achievable in PvE gear. Depending on how it scales beyond that point, it could become more powerful (on paper) than the other options. I think it could work out pretty strong.
Dream of Cenarius is now showing as pretty solidly ahead of the other level 90 talents too (2-3k). Time to dust off the HT button, I guess.
|
|
|
|
09/17/12, 7:44 PM
|
#478
|
|
<Druid Trainer>
|
Oh yeah, didn't look too closely at that actual haste values you used for testing, higher values will be accessible pretty quickly.
That test is still strange though. At a 1.55 s cast, you're getting 9 consistently, which is what's expected for a 15s lifetime (so at least, you're within rounding error of what's expected). But at very close value, 1.59, you're only getting 6 (while 9 would still be expected). So I have no idea what to expect from haste beyond that.
|
|
|
|
09/17/12, 8:27 PM
|
#479
|
|
Soda Popinski
|
Yeah, I was kind of surprised by how quickly it ramped up too, although I think it was more to do with the level of haste correcting for the poor pet behaviour than anything else. They seemed to chain their casts properly once I got to that haste level, whereas at lower haste levels they stutter and stop randomly.
I'm not sure how the pets and client/server stuff works, but me testing from EU to US servers might be exaggerating the problems at lower haste levels?
|
|
|
|
09/18/12, 12:25 AM
|
#480
|
|
Glass Joe
|
Originally Posted by Tecton
Dream of Cenarius is now showing as pretty solidly ahead of the other level 90 talents too (2-3k). Time to dust off the HT button, I guess.
|
That's interesting because SimC is showing them really close together for a T14 H profile. About 1k apart, with Nature's Vigil actually being ahead.
Now, I'm definitely a novice when it comes to SimC and I'm just using the T14 H pre-loaded profile. I plan on spending some time tomorrow looking through the profile and trying to figure it out. But I'm wondering if the fact that WC doesn't have a combat time length is throwing things off a bit for DoC versus NV.
Just for comparison purposes, I'm seeing 99.8k for DoC and 100.7k for NV - those are both rather different from what WC is showing.
edit: rereading above a little bit, I know this is what Hamlet and Graylo were discussing. I guess what I'm trying to point out is, while I expect WC and SimC to give slightly different numbers, they are very different. And I'd expect a talent like NV to change based on fight time, but DoC should remain fairly constant. Yet we're seeing a 7k DPS difference between the two tools.
Last edited by Eluial : 09/18/12 at 12:32 AM.
|
|
|
|
|
|