[Feral-Cat] Cataclysm Release - Page 6

tangedyn · 12/30/10, 9:37 PM

Originally Posted by righthorn

Things seem very much up in the air, especially when the various models are concerned. Depending on whether you use Toskk's formulation or the Mew simulations, you'll get differing answers as to which glyph (TF, SR, Bers) is the 'best'. Not only that, but stat weights will also vary depending on which method you use. Toskk's seems to show Crit as being better than Haste, while the simulation (with my current stats) show Haste has better than Crit.

Not really the clear cut answer you may be looking for; but at this point the best way is to burn gold with dust & at the reforger and try out various options yourself.

The Simulation will generally able to model the fine details of the rotation more accurately than the Formulation, so if they both recommend different things, I would go with the recommendations of the Simulation. That said, Toskk has done A LOT of improvements to the Formulation to model the stuff that Yawning and myself have previously thought was not feasibly possible to model with Formulation, and they should be fairly close now.

If they are recommending different things, most probably the values are very close and you may not really see a significant difference in actual gameplay. However, if you find that the values are actually very far apart, please report it as a bug and we will try to look into it.

righthorn · 12/31/10, 12:55 AM

@tangedyn:

The simulation always puts Haste ahead of Crit by ~0.03 (Crit @ 0.86, Haste @ 0.89), whereas the formulation puts Crit ahead of Haste by a much larger margin of 0.9 (Crit @ 0.85, Haste @ 0.76)

Perhaps not a very large difference, but it is a 10% difference and that seems a non-trivial discrepancy between the two methods. Also conceptually it seems odd that Haste would at any point be better than Crit. Crit affects almost 100% of our energy expenditure, whereas Haste only affects the number of FB and SR uptime. Maybe the simulation is accounting for some sort of 'step function' behavior with Haste where hitting a certain level allows enough energy generation to line up better with trinkets / TF?

If this seems like a significant enough discrepancy to warrant further investigation, please let me know via pm and I can start a new issue on the mew site and attach my .mew file.

EDIT:

Played around further with the settings by moving 100 from crit to haste, and now the simulation is agreeing with the formulation on stat weights. I'm guessing my earlier suspicions were right, and that the simulation is determining that I currently don't have sufficient haste to reliably maintain near-full uptime on Rip/Rake/SR (I have almost entirely crit/mastery gear, and reforged all hasted into mastery) - which is reflected in practice as I generally run SR uptime in the 80's.

What's interesting is that there seems to be a lower cap of sorts on haste, and that it may be more beneficial to get sufficient haste in order to maintain a higher SR uptime even at the expense of crit - this isn't something that's obvious given that SR buffs 1/4 of our damage, while crit affects everything. It also may be an artifact of the simulation that does not translate well into actual practice; this lower cap could also be caused by the default simulation script not properly prioritizing Rip/Rake over SR and 'bleeding' energy away from Rip/Rake when SR uptime suffers.

However, given that much of the t11 and equivalent raid drops have haste, I suspect that this is only an issue for heroic 5m geared druids, and will not be a factor going forward.

The Grog · 12/31/10, 4:30 PM

Originally Posted by RareBeast

The parsing of the trinket procs in Rawr are still being worked on. Refreshing from Wowhead actually removes the procs from a lot of the trinkets. You can easily edit the items of interest and add the correct procs to the items. Details of most of our useful trinket can be found in the rogue Cata mechanics thread here at EJ. Once the correct procs are in, the results have seemed correct to me.

The next Rawr release should have most of the trinket issues sorted out apparently.

The listings I posted were after adjusting most of those trinkets for correct ICD/proc numbers. That refreshing the item causes a clearing of the proc data does make sense with what I saw, as most items fell in value after doing so.

For the moment I'll stick with the Eye and Tia's Grace. Grace of the Herald procs with a terrible ICD, the Key should have a worse uptime than the Eye so long as you have > 30% crit, and while Skardyn's seems stronger if you can sync it with TF & a dot refresh I'm probably too out of practice to be doing that. I forgot how to make custom events in Dotimer and I need help watching those cycles.

weswtp · 12/31/10, 6:36 PM

Originally Posted by RareBeast

The parsing of the trinket procs in Rawr are still being worked on. Refreshing from Wowhead actually removes the procs from a lot of the trinkets. You can easily edit the items of interest and add the correct procs to the items. Details of most of our useful trinket can be found in the rogue Cata mechanics thread here at EJ. Once the correct procs are in, the results have seemed correct to me.

The next Rawr release should have most of the trinket issues sorted out apparently.

Although the parsing for trinkets is still being worked on, RAWR is throwing A LOT of weight behind STR proc trinkets. Even more so than many AGI proc based. I was wondering if anyone else was noticing this as well and how true to form OR why this would be the case. As it stands, [Heart of Solace](heroic) and [Heart of Rage] are basically BiS.

Alarron · 01/01/11, 3:00 AM

Rawr is (still) not correctly parsing a few trinket CD's, though most are fixed now. Since we seem to keep getting this question, here's a list of trinkets and their value (roughly sorted into tiers, feel free to disagree)

Tier 1:
Essence of the Cyclone (H)
Prestor’s Talisman of Machination (H)
Fluid Death

Tier 2:
Tia's Grace (H)

Tier 3:
Essence of the Cyclone
Prestor's Talisman of Machination
(Heart of Rage (H))
Left Eye of Rajh (H)
Key to the Endless Chamber (H)

Tier 4:
Unsolvable Riddle (best on-use trinket, excellent for burn phases)
Tia's Grace
(Crushing Weight (H))
(Heart of Rage)
(License to Slay)

Yes, the high agi from Tia's/FD (especially the mastery on Tia's) makes them better than their ilvl. Proc trinkets > Use trinkets, in general, due to greater potential uptime, but this balances better at lower gear levels or with less buffs (dropping eye/key back down). Unheeded Warning is pretty terrible. DMC: Hurricane is pretty bad as well, since it doesn't scale at all with our gear (unless they buff the PPM). More discussion on my blog at The Fluid Druid � Blog Archive � Trinket Ratings for Heroics/T11 Content.

slippyjz · 01/01/11, 11:51 AM

Originally Posted by Alarron

DMC: Hurricane is pretty bad as well, since it doesn't scale at all with our gear (unless they buff the PPM).

I'm curious about this, I have the card and was wondering if I should keep it or give it to a hunter in my guild. I can't find a lot of hard facts about it but my understanding is that it DOES scale with haste, or so i've read. You seem to have it listed very very low on the trinket list because you don't know what the PPM actually is. Does anyone know for sure? Also, you're listing the VP trinket Fluid Death as one of the top scoring trinkets, I've thought about buying it but I'm unsure. I've not seen any of the top druids (that i've looked at so far) using/buying it because of our low value on hit. Anyways, thanks for any advise/replies.

Edit: spelling >.<

Kokomala · 01/01/11, 12:07 PM

Originally Posted by Alarron

DMC: Hurricane is pretty bad as well, since it doesn't scale at all with our gear (unless they buff the PPM).

I see that on your blog you estimate 2 PPM, but from what I've been reading on Wowhead, I've seen tests up to 4 PPM - different class though. What would the rank be with 3 or 4 PPM in your trinket comparison?

Perhaps one our European members could test this out in a few hours and give some real results.

righthorn · 01/01/11, 12:47 PM

I would've expected Use trinkets to be generally better than proc trinkets. For one thing, despite Proc trinkets having 20% more maximum uptime compared to Use trinkets, in actual practice the gap is less, since trinkets do not immediately proc once the ICD is up.

But more importantly, because half our damage comes from Rake & Rip, more so than any other class our damage windows are very 'bursty'; instead of a constant stream of damaging attacks we have snapshots which disproportionately affect our DPS. Proper usage of 20s Use trinkets guarantees that you buff Rip & Rake twice every 2 minutes, which is something that you cannot reliably do with Proc trinkets beyond the opening attacks unless you sacrifice bleed uptime waiting for procs.

righthorn · 01/01/11, 1:06 PM

Originally Posted by slippyjz

Also, you're listing the VP trinket Fluid Death as one of the top scoring trinkets, I've thought about buying it but I'm unsure. I've not seen any of the top druids (that i've looked at so far) using/buying it because of our low value on hit.

Hit may be bad, but it's not like it's significantly worse than a better stat like Mastery (YMMV, but I'd be extremely surprised to find a stat setup where hit ends up being worth only half of mastery). However, Fluid Death is essentially 380 Agility, which is phenomenal. I suspect the reason you see nobody using it is because Tia's Grace is almost as good, and essentially free. This early in the content there are better pieces to buy with our limited VP than Fluid Death.

slippyjz · 01/01/11, 2:04 PM

Yeah, according to my math Tia's is only ~70dps behind Fluid Death....For my gear set anyhow

Tinderhoof · 01/01/11, 2:04 PM

Originally Posted by slippyjz

I'm curious about this, I have the card and was wondering if I should keep it or give it to a hunter in my guild. I can't find a lot of hard facts about it but my understanding is that it DOES scale with haste, or so i've read. You seem to have it listed very very low on the trinket list because you don't know what the PPM actually is. Does anyone know for sure? Also, you're listing the VP trinket Fluid Death as one of the top scoring trinkets, I've thought about buying it but I'm unsure. I've not seen any of the top druids (that i've looked at so far) using/buying it because of our low value on hit. Anyways, thanks for any advise/replies.

Edit: spelling >.<

Righthorn is correct about the amount of agility that Fluid Death gives us is the big draw. Something not mentioned is in Alarron's weighting he is assuming that 40% of that hit has been reforged into mastery. Same goes for other hit/expertise trinkets (Key, Heart, Eye). In each case thats at least 114 (128 for fluid death) mastery. On top of the agility it makes nothing else quite as good until we get into heroic raids.

Yawning · 01/01/11, 2:28 PM

Originally Posted by Kokomala

I see that on your blog you estimate 2 PPM, but from what I've been reading on Wowhead, I've seen tests up to 4 PPM - different class though. What would the rank be with 3 or 4 PPM in your trinket comparison?

Perhaps one our European members could test this out in a few hours and give some real results.

http://elitistjerks.com/f78/t110134-...4/#post1820734

1.0 PPM no ICD. Different classes bashing at the dummy with anything resembling a normal DPS rotation will see different raw results because the proc chance is calculated based on the base unhasted swing timer.

For the Rogue in the log, data matches a 3% proc rate after including specials, which is what one will expect with a 1.8 MH dagger.

Melee: 5789 Envenom: 433 Mutilate: 845 Backstab: 301 FoK: 736
Rupture: 2220 ticks (10 ticks/cast at 5 pt) ~220 finishers
Total triggers: 8324
Total procs: 276
Observed proc rate: 0.03315714
Expected proc rate (1.8 MH, 1.0 PPM): 1.8/60 = 0.03

Using SVN Mew with a T11 normal mode geared Druid, the proc is worth roughly 130 DPS (140 after adjusting for CoE on the mob). The value does change with gear/buffs (for example if you could somehow manage to get an absolutely phenomenal amount of haste, it would do a lot more), but it should be a reasonable estimate for now.

If I were a 2H plate DPS class or an Enhance Shaman I would be all over this. Not so great for a Feral Druid.

Alarron · 01/01/11, 6:07 PM

Answering a few questions:

DMC: Hurricane does scale with haste (as does all PPM items), but the lack of scaling i was referring to is in regards to mastery, as compared to a survival hunter or an enhancement shaman (who mastery makes the proc hit harder).If it's only 1 PPM, then it's even worse then I projected.

Use vs Proc Trinkets: As I mentioned in my blog post on the subject, this is something that is very much fight and gear dependent. Proc trinkets have an inherently higher potential uptime, but worse gear (especially haste) can reduce that somewhat. If a fight features a burn phase, use trinkets are also (probably) better. I like Unsolvable Riddle, personally; even though it's a bit lacking for a static fight, the 20s duration and 2m CD makes it sync well with glyphed Berserk and the Tol'Vir potion.

The Grog · 01/02/11, 6:37 AM

Originally Posted by righthorn

I would've expected Use trinkets to be generally better than proc trinkets. For one thing, despite Proc trinkets having 20% more maximum uptime compared to Use trinkets, in actual practice the gap is less, since trinkets do not immediately proc once the ICD is up.

But more importantly, because half our damage comes from Rake & Rip, more so than any other class our damage windows are very 'bursty'; instead of a constant stream of damaging attacks we have snapshots which disproportionately affect our DPS. Proper usage of 20s Use trinkets guarantees that you buff Rip & Rake twice every 2 minutes, which is something that you cannot reliably do with Proc trinkets beyond the opening attacks unless you sacrifice bleed uptime waiting for procs.

I would think that Skardyn's would be pretty good. Agi and Mastery are our best stats, and mastery only matters on dot refreshes. I'm not sure if overwriting Rip at the end of the duration would be worthwhile, but I suspect it is better than trinkets that proc haste or have static haste or hit/exp.

righthorn · 01/03/11, 1:34 AM

@Grog:

You can refresh DoTs early (aka 'clipping') without losing out on any tics. IIRC you have up to a 2 tic 'grace period' so for a 20s use trinket, you can Rip, Rake, Rake again, then Rip with 1s left on the Use (and 3s left on Rip).