[Ret 5.0 Pre-MoP] - Panda-ring to all players - Page 19

Exemplar · 08/27/12, 1:48 PM

I keep forgetting HoW has Mastery - that's captured on a separate line of my spreadsheet and I just don't glance down at it. With Mastery HoW does indeed do more than Exo. I'm seeing HoW recommended before Exo for most setups after correcting some bugs (example - HoW wasn't being used during AW, which devalued it a lot).

Fixed a lot of small bugs (that add up) reviewing my spreadsheet. Stomped a few more comparing my data to SimCraft (thanks for the raw data, Balhale!). Using 3-4HP TV in fillers is indeed showing a slight gain. You overall wind up with very few fully open GCD, but you have a reasonably large number of fractional GCD. That means there's a reasonable value of forgiveness for the average human being (being a fraction of a second late a few times shouldn't cause huge ripples in all subsequent attacks).

My spreadsheet is indeed using a reasonable number of 1 or 2 HP Inq under all scenarios, rather than let the buff fall off. I'll explore an option to forcibly only use 3 (beyond the initial) and see what happens (other than less uptime on the buff).

I'm using the precise gear list you sent in your sims - and I've reviewed each of those items on mop.wowhead to verify stats hadn't changed (they matched what I had with 100% accuracy). I think I may have found the issue - I may have been double-multiplying Censure (5 * 5 = insane!). That took ~4k to ~20k (there was also a glitch in buffs, so it's really closer to 5.7k. I, too, am now seeing less DPS than live.

Live at 47033, MoP using DP Talent at 42455. Drop of roughly 10%.

I plan to release a Beta version of my spreadsheet sometime within the next 8 hours. I greatly encourage anyone interested to dive into formulas and PM me any bugs you find.

cky71321 · 08/27/12, 3:59 PM

I'll take a look into it once you've got it ready. In the meantime, do you have stat weights that differ from Balhale's? It would be interesting to compare those numbers.

Balhale · 08/27/12, 4:07 PM

That matches my results much much better. I'm actually getting 1-2k more than that, probably from things like not averaging potions and GoAK and including the GoAK's melee attacks.

I'm messing around with getting the sim not to delay HoW so much during wings, but getting the conditionals to work properly is tricky going.

Edit: What stat priority are you seeing? SimC is showing haste > mastery > crit, with haste actually showing breakpoint like behavior where for some increases in haste it actually gets stronger.

Exemplar · 08/27/12, 4:51 PM

Beta spreadsheet available: here. Beta means the spreadsheet is in Beta, not that MoP is.

Balhale - I was missing the Gur tentacle damage in my totals - that's around 2k DPS. So our margin is even more narrow and easily explained with our modeling differences. I'm also getting free time in the 6-8% range, which seems to match you.

I've added some features, such as the ability to use Sacred Shield in empty GCD. You can even choose to prioritize it over 3HP TV - you lose about 200 DPS or under half a percent total, but your SS uptime goes from ~25% to ~50%. Tier 6 talent is probably going to GCD lock us and SS would only be usable in phases where you can't hit anything (or doing anything to survive trumps damage). That's somewhat sad.

I have not yet done a stat comparison. The method Excel is currently modeling is by creating a simulation where each Row of a sheet is 0.1 seconds. Lots of cells referencing lots of other cells (non-recursively, thankfully). Every change causes all the cells to recalc. An individual run takes less than 10 seconds, so isn't painful, but trying to calc stats would take 100 runs multiplied by # of stats. I just debugged it and did a test - 1% took around a minute or so on this lame laptop. Estimate between 30 min and 2 hrs on a decent system.

The spreadsheet has that functionality open for others, if they desire to try such a run. Warning - I did a prelim debug and it appeared to function, but I have not run fully. As such it's entirely possible the routine could fail or generate spurious data.

I do expect Haste to have some plateau effects, where it's really great to pass a threshold, then mediocre, then really great. It's also possible that more Haste could produce negative results at certain points because it causes some conflict that gets resolved in an unpleasant fashion.

Cloudio · 08/27/12, 8:16 PM

Thx Hamsda/Exemplar for correcting my mistake.

Balhale · 08/27/12, 8:52 PM

I got the HoW issue fixed - here are stat weights for SW post that fix, reforging and gemming for haste (meeting socket bonuses)

And SW: (reforged but not gemmed for haste)

And DP: (reforged but not gemmed for haste)

And HA: (reforged but not gemmed for haste)

averykey · 08/28/12, 7:50 AM

Inq > TV (5) > HoW > Exo > Crusader > Judgment > TV(3)

Where does dp fit into this?

Do we just use it whenever it's up, does it give us a 5hp templar? or should we wait until we have 5hp and then templar twice?

I think we should wait until we build up 5hp and then templar twice, but I have no math to back it up.

What do you guys/girls think?

Voldemort · 08/28/12, 8:10 AM

TV (DP), I think, we need use like a filler for empty GCD, but not forget about a duration of DP buff

P.S. It's my first post here and English isn't my native language.

averykey · 08/28/12, 8:18 AM

Originally Posted by Voldemort

TV (DP), I think, we need use like a filler for empty GCD, but not forget about a duration of DP buff

P.S. It's my first post here and English isn't my native language.

I think that might be the best way to, Only use the proc when everything else is on cooldown, but be wary of its duration.

Belatar · 08/28/12, 8:40 AM

Always use before other HP generators, because a DP powered ability might proc another DP. Often had up to 4 DP TVs in a row.

Zephur · 08/28/12, 8:45 AM

Either SimC or the beta spreadsheet (or both) are calculating Seal of Truth procs (the 12% damage, not Censure) incorrectly. The spreadsheet lists it as 20% of overall dps, which seems way too high, and SimC is calculating that the number of Censure ticks = the number of seal procs, placing it at about 5% of overall dps.

Both of these situations seem wrong to me.

Edit: This also (at least partially) explains why haste was ridiculously overvalued in the spreadsheet.

Fordragon · 08/28/12, 8:46 AM

Originally Posted by Belatar

Always use before other HP generators, because a DP powered ability might proc another DP. Often had up to 4 DP TVs in a row.

DP isnt proccing from generators but from finishers since like the 1st mop beta build,what are you talking about now.

averykey · 08/28/12, 8:52 AM

Originally Posted by Belatar

Always use before other HP generators, because a DP powered ability might proc another DP. Often had up to 4 DP TVs in a row.

We are talking about 5.0, are you?
If we use a divine purpose proc while under 5 holy power with a hp generating ability available, I would think of that as a dps loss, because if you do that 3 times during a fight, you lose 1 Templar but in a 5 minute fight you could get 30 procs, that's a loss of 10 templars or 500k+ damage.

Exemplar · 08/28/12, 9:01 AM

Originally Posted by averykey

Inq > TV (5) > HoW > Exo > Crusader > Judgment > TV(3)

Where does dp fit into this

You continue to follow the priority. Simply ignore it until one of four scenarios. First, you need to refresh Inq - poof, instant full duration Inq. Second, you hit 5HP and must dump, at which point you would use TV twice because the first eats DP and the second drops you from 5 to 2. Thirdly, everything else is on CD and you have a free GCD, in which case you're following the 3HP TV rule. Fourthly, the DP buff (8 second duration) drops low and is at risk of running out (my recommendation is 2 seconds or below) - treat that like a 5 HP.

Like pre-5.0.4 you want to minimize the pushback to other abilities (in order to maximize HP generation overall). That said, chaining 4 TV back-to-back is still great because TV is our best ability - if that happens (0.4% probability, 1.5% if one of those TV used actual HP), enjoy it and use them.

Unrelated:
Someone ran the graph in my spreadsheet (thanks Zephur!). I'm linking the image they PMed me.

Spreadsheet DP Stat Weights
As you can see, Haste drops below the chart, then suddenly spikes through the roof. Clearly some plateau was reached where 1 more CS/TV/something was possible, which raised damage by 30+ DPS.

Overall it says:
Haste - 22.62 DPS (false result)
Str - 3.38 DPS
Crit - 2.04 DPS
Mastery - 1.19 DPS

If the weights SimCraft and spreadsheet have generated remain true at 90 (most likely), then mixed gems (Str + secondary) will be the way to go for red slots, and possibly full yellow or blue gems in orange/purple slots. For example: in the above Crit is 60% of Str and you get twice the rating on a gem (160 Str Ruby vs 320 Crit Smooth Yellow) - the secondary stat got you 120% total value. SimCraft shows haste at 100%+ of Str, so in a Yellow slot you would be better with the 320 Haste Yellow than 80 Str and 160 Haste Orange.

Rough guesstimate based on eyeballing these rankings, SimCraft's rankings and the buffed stats on the gearset we're using - go for rough equality. That gearset has 16.1% Crit, 12.3% Haste, and 15.1% Mastery. Aim to move these 3 stats as close to parity as possible. This gearset has higher Crit and lower Haste, so try to adjust one to the other.

P.S. For now definitely use SimCraft's Haste values above mine - since Excel does one rather static simulation, it's going to get a fixed result and one more or less attack throws havoc with results. Whereas SimCraft does many, many simulations and one more/less attack gets averaged out across the sims, providing far more accuracy for something like Haste. If/when I can get my probability system functional that helps smooth such rough edges, giving portions of every attack.

Edit:

Originally Posted by Zephur

Either SimC or the beta spreadsheet (or both) are calculating Seal of Truth procs (the 12% damage, not Censure) incorrectly. The spreadsheet lists it as 20% of overall dps, which seems way too high, and SimC is calculating that the number of Censure ticks = the number of seal procs, placing it at about 5% of overall dps.

Both of these situations seem wrong to me.

Looks like I was taking 12% base weapon + full bonus of AP. Should be (base weapon + AP bonus) * 12%. This is about an 80% drop in total SoT damage (it's now 4.34% total damage, but a lot more procs than Censure ticks). Total DP DPS drops to 37.6k, which seems low compared to SimCraft. Hoping there's an undervalue somewhere to pick up the slack, because that's a huge DPS drop from 4.x to 5.x.

Fixed SoR in same fashion. Since we'd found SoJ wasn't using AP it didn't change. Amusingly, at least with a Heroic Gur, SoJ with no AP scaling beats SoR (on one target) with AP scaling.

Belatar · 08/28/12, 12:49 PM

@Fordragon, averykey: You are right in that I still had the "old" DP in mind. HP generators cannot proc DP anymore. My point was to not let DP run out unused as TV is our best ability and you get an extra one without needing to generate HP. Which Exemplar explained far better and in more detail than I could