[Discussion] Patch 6.2 PTR

[Frosted]

Ok, so, continuing where my previous jade post left us.

First off, the fundamental reason why haste gained so much value has not changed. It is still centered around the idea of breakpoints, with some residual added value coming from haste giving you more pets/4pc uptime. The only question is now the magnitude of the gains haste makes, since the pets are now no longer an overwhelming force of single target destruction.

We start out simple. Compare a mastery optimized profile to a haste optimized profile. These profiles are posted a few posts back. The "mastery" profile is my old profile which I optimized for mastery gear/enchants, and the haste profile is the one Deathshinobi optimized for haste gear/enchants.



At first glace, mastery seems to be winning. But I wanted to get something a bit more rigorous than that. Keeping breakpoints in mind, I thought that what could be happening was that the haste profile is too far beyond the breakpoint - that it's essentially gone into territory where haste has lost a lot of its value; so I looked at DPS scaling plots for mastery/int/haste.



Here I used the mastery profile, because I was most interested in seeing what happened to haste as we approach, and then pass, the breakponint - which the haste profile may not have let me do. We can see the breakpoint really clearly (indicated with the arrow). It's not nearly as strong as it once was, which makes sense, but it's still a clear jump in DPS. Seeing this, I remain confident that gearing for that 20% breakpoint is correct. (The exact value here in the scaling sim is around 1224, but that from eyeballing the graph - I think using the actual swingspeed vs haste formula is a better way to determine the actual value, which is the 1300 I cited earlier).

Interestingly, haste continues to have a high value for a brief period, before declining below int and mastery. Remember here that you're mainly interested in the slope of the lines. Just because the haste scaled DPS value is above the mastery does NOT mean it is actually scaling better still. That is what is actually telling you how the DPS is changing with the stat increase. And the slope for haste is looking as if it decreases after the breakpoint - so there is some value, slightly beyond the breakpoint, which it's worth it to drop everything else to gear for, and then beyond that it loses value below int/mastery again.

To ensure nothing wonky was happening, I extended the stat scaling out another 400 rating:



The decline in the value of haste is more obvious here, as we see the addition of raw int overtake just stacking haste. The conclusion from this is: It is NOT worth it is blindly stack haste. Haste > * is not acceptable, we must keep in mind the ilvl of the gear (and thus the amount of int involved) when making decision. Therefore, explicitly stating haste (20% breakpoint) haste > Int > haste/mastery? is mandatory in this case.

To go further into figuring out exactly how much haste we're allowed to gear for, I used reforge plots.



This what what happens to your DPS as you trade off, point for point, mastery and haste. The breakpoint is extremely obvious here, as is the continued value of haste..but at the end of the +haste/-mastery range, haste still has value. So I extended it further.



The clear trend of haste > mastery only extends a few hundred rating after the breakpoint. Then, it fluctuates/flatlines a bit before finally diving down sharply.

Keeping this in mind, I replaced all the enchants on the mastery gear with haste. If the reforge plot was right, this should be a straight DPS increase (as we're not involving any int trade-offs here, just simple mastery for haste).

Code: Select all

63035 50.7% Mage_Arcane_T18N_Mastery_haste_mix 61379 49.3% Mage_Arcane_T18N_Mastery

! Large DPS gains. Enchants alone were enough to push the mastery profile over the breakpoint, but not venture into areas where I'm losing DPS for gearing haste. I went further and started replacing, item by item, any mastery item for any haste item. Eventually I came to this:

Code: Select all

head=demon_princes_ascendant_crown,id=124159 neck=choker_of_forbidden_indulgence,id=124391,enchant=gift_of_haste shoulders=mantle_of_the_arcanic_conclave,id=124177 back=cloak_of_hideous_unity,id=124138,enchant=gift_of_haste chest=robe_of_the_arcanic_conclave,id=124171 wrists=powdersinged_bracers,id=124183 hands=gloves_of_the_arcanic_conclave,id=124154 waist=demonbuckle_sash_of_argus,id=124200 legs=pantaloons_of_the_arcanic_conclave,id=124165 feet=bloody_daggerheeled_pumps,id=124149 finger1=loathful_encrusted_band,id=124192,enchant=gift_of_haste finger2=spellbound_runic_band_of_the_allseeing_eye,id=118306,enchant=gift_of_haste trinket1=tome_of_shifting_words,id=124516 trinket2=prophecy_of_fear,id=124230 main_hand=edict_of_argus,id=124382,enchant=mark_of_warsong

Which is the best performing profile, above gearing straight haste or mastery, at each ilvl snapshot (normal/heoric/mythic). Trading off any additional mastery items for haste items is a DPS loss, because the haste items are almost always a lower ilvl than the mastery items, and the value of haste as dropped to where the haste gains can no longer support the loss of intellect.

Here is the stat scaling for this profile (keep in mind, this is a general feel for stats. It has a hard time seeing the decayed value of haste as it's continuously stacked and cannot handle breakpoints well):



Here are the talents you want:



Here are your trinkets:



Note: You really want prophecy. Like. You REALLY want it. I haven't gone into trinket specifics here (much confuse, many variable), but prophecy loves haste. and haste loves prophecy. Plus it's insane on multi-target when it procs and you dump a bunch of AMs into it. You want it.


The only odd variable in all of this is the pet delay. I posted it earlier in the thread, but you seem to need at least 25% haste to begin seeing reductions in pet delay. Only the full haste profile can obtain this. The profile which mixes haste/mastery can only achieve ~23% at mythic item levels. So I think this leads to two potential gearing schemes.

1) Conservative Haste Gearing - Haste 20% (1300rating) > Int > Mastery > Haste > Mult > Crit > Vers
2) Aggressive Haste Gearing - Haste 20% (1300rating) > Int > Haste (~25-26%) > Mastery > Haste > Mult > Crit > Vers

Conservative Haste Gearing is essentially "going with the sim." Because it cannot quantify the effects of haste delay, you are betting against it being worth the DPS tradeoffs the sim would see in order to get it.

Aggressive Haste Gearing is gearing for pet delay reductions, you are betting that the pets are worth any DPS loss the sim may see due to whatever you did to get that haste.

If you want my personal opinion, I'd go with #2. I think that if you're smart about it and a bit lucky, you can use gems to get yourself to the 25-26% haste rating, and never have to compromise the above profile I posted. I do not advocate tanking your int/ilvl to reach that haste level though. I do not believe it is worth it.

[Fleks]

Awesome work thanks but one question,are we now using the Classtrinket in the hastebuild aswell?

[Frosted]

For the profile I made towards the end it uses class trinket + Prophecy as its BiS.

[Frosted]

Extended upgrade path. (I think someone else posted it earlier? Showing that nothing really changed.) Get the pets.

[Frosted]

Spec Comparisons

Normal


Heroic



I'm not gonna post mythic cause it's using the legendary.

Keep in mind, the new WJ just happened, and I haven't done shit for work on the Frost profile in comparison to the amount of time I've put into Arcane optimization. The WJ almost certainly requires APL upgrades (which honestly might have to wait until people raid with it, as I have 0 clue how to handle it right now and I don't think Komma has gotten too far with it yet).

Also, we're not modeling the icicle+doom nova bug. That's I think ~4-5% more DPS for Frost? If it doesn't get fixed, which is not shown above.

[Griepen]

Extended upgrade path. (I think someone else posted it earlier? Showing that nothing really changed.) Get the pets.

It this after the 40% nerf?

[Frosted]

Everything I just posted was after the 40% nerf.

[Griepen]

Awesome. Thanks, just wanted to make sure.

[Bashlow]

I am unsure about the trinkets we should go for now, Frosted`s sim said Insignia and gaze , now we see prophecy and archi trinket

[Mortiferus]

Frosted, we all need to buy you a drink. Or 7 drinks. Incredible work.

I extrapolate from this that, as we gear up in HFC, it's not worth switching current T17 enchants from Mastery to Haste. At least, not until we get T18 2PC. Unless I'm missing something, switching the T17M profile to Haste enchants is simming as a DPS loss. I also extrapolate that it's going to make gems interesting. Obviously none of this matters once we're all in the BiS gear, but as we get a piece here and a piece there, I think this will make the decision structure far more difficult than it was last tier.

[Magegodx]

I'm quite sure that people has the same questions. So I'll post 'em here!

1: Class trinket is the way to go as Arcane?

2: Is 2-set / 4-set as Fire considering good?

[Rabona]

Awesome work Frosted, thank you so much<3

[Frosted]

I'm quite sure that people has the same questions. So I'll post 'em here!

1: Class trinket is the way to go as Arcane?

2: Is 2-set / 4-set as Fire considering good?

1) Yes, I believe the post addresses that.

2) Is it good? Yeah, it's alright. I posted spec comparisons above, they include the set bonuses.

[Frosted]

Frosted, we all need to buy you a drink. Or 7 drinks. Incredible work.

I extrapolate from this that, as we gear up in HFC, it's not worth switching current T17 enchants from Mastery to Haste. At least, not until we get T18 2PC. Unless I'm missing something, switching the T17M profile to Haste enchants is simming as a DPS loss. I also extrapolate that it's going to make gems interesting. Obviously none of this matters once we're all in the BiS gear, but as we get a piece here and a piece there, I think this will make the decision structure far more difficult than it was last tier.

Yeah, I think we'll want to swap only once we get 2pc.

I agree, progression gearing is not going to be simple at all. Honestly, Jebbage had a good idea in the IRC that we may want to start using AMR again to help us with deciding.

[Raencloud]

Frosted, I believe I have a slightly better BiS gear list (+400 dps). The changes are:

(feet) Bloody Dagger-Heeled Pumps -> Dessicated Soulrender Slippers
(wrists) Powder-Singed Bracers -> Contemptuous Wristguards


I also was hoping you could help me understand why unglyphing AP is producing some noticable DPS gains (1200-1400 dps when using UM+OP and UM+PC). As expected I see a decrease in average AP uptime when unglyphed, which I attribute to the fight length variances and unglyphed AP losing 1-15 seconds of AP uptime on any given fight. Still though, this seems to be producing a DPS gain despite a lower total uptime. This is a stark contrast compared to SN+PC which show a slight dps decrease when unglyphing AP. Thoughts?

[Vynestra]

I'm still trying to digest that post. What is the mastery break point where we start losing dps if we trade it for haste?

I understand before 1300 you can trade int and haste barring no other factors would cause you to lose stats unevenly. But once you hit that -- then you want haste over any 2ndaries til 25-26%. In which case you stop going for haste and back to mastery. But is there a point before 25-26% haste that trading haste for mastery can be bad?

And will you try to factor the ring in (despite not knowing all the interactions) to see if haste gearing is better or worse with the ring? To see if it's worth it to try go for the haste gearing? Because if the ring finds mastery more valuable it would be a few weeks of haste then back to mastery.



EDIT: Frosted clarified on these in IRC chat.

And some of them were clear in the post but I didn't read them correctly lol.

[Frosted]

Frosted, I believe I have a slightly better BiS gear list (+400 dps). The changes are:

(feet) Bloody Dagger-Heeled Pumps -> Dessicated Soulrender Slippers
(wrists) Powder-Singed Bracers -> Contemptuous Wristguards


I also was hoping you could help me understand why unglyphing AP is producing some noticable DPS gains (1200-1400 dps when using UM+OP and UM+PC). As expected I see a decrease in average AP uptime when unglyphed, which I attribute to the fight length variances and unglyphed AP losing 1-15 seconds of AP uptime on any given fight. Still though, this seems to be producing a DPS gain despite a lower total uptime. This is a stark contrast compared to SN+PC which show a slight dps decrease when unglyphing AP. Thoughts?

Huh. That's odd, I know I looked through all the bracers last night and chose the one in the profile above because it performed best. I must have made a mistake, thanks! Also double checked the other foot item - getting the same results. I honestly didn't even see that item, I've been told it was a trash drop though so that makes sense.

Thanks for taking the time to look into improving the profile . I'll update it in SimC here in a bit.

As for the second part of your question, I'm not sure. I'm a bit busy trying to get stuff together to update all of the spec SimC threads. Once I finish I'll give it some thought. Have you tried removing the archi trinket and looking at if that is the culprit?

[Raencloud]

The Mage trinket is the culprit, I just don't understand why that is. The mage trinket is also the main reason why OP becomes viable as a choice (as we've seen UM_RoP_OP is 2nd highest DPS below UM_RoP_PC).

I also noticed I forgot to switch my food choices from Sleeper Sushi to Buttered Sturgeon, but doing so did affect any gear choices that I saw (Sturgeon is better than Sushi though). It's probably also noteworthy that the necklace slot is extremely close (between Choker of Forbidden Indulgence and Vial of Immiscible Liquid). I see Choker as barely beating Vial when using PC (less than 50 dps change), but Vial as superior when using OP (closer to 100 dps change).

[Muoric]

Huh. That's odd, I know I looked through all the bracers last night and chose the one in the profile above because it performed best. I must have made a mistake, thanks! Also double checked the other foot item - getting the same results. I honestly didn't even see that item, I've been told it was a trash drop though so that makes sense.

Thanks for taking the time to look into improving the profile . I'll update it in SimC here in a bit.

First of all thanks Frosted for all the work.

I looked at the bracers also myself but missed those boots also. And without the boots the haste level on the normal gear would have been below the 20% threshold which based on your own analysis is the big one.

Was looking at the heroic level myself and wondered about the bracers but didn't have time to do proper sims to back my thoughts and didn't want to blindly trust normal stat weights and assumptions that mastery should overcome (or at least catch up) haste after the 20% threshold. Now adding the boots to the profile it's a whole different matter with all the haste on them.

[Muoric]

I also was hoping you could help me understand why unglyphing AP is producing some noticable DPS gains (1200-1400 dps when using UM+OP and UM+PC). As expected I see a decrease in average AP uptime when unglyphed, which I attribute to the fight length variances and unglyphed AP losing 1-15 seconds of AP uptime on any given fight. Still though, this seems to be producing a DPS gain despite a lower total uptime. This is a stark contrast compared to SN+PC which show a slight dps decrease when unglyphing AP. Thoughts?

This is just a thought but could it be possible that with the class trinket glyphed AP lasts too long for our burn phase? So we might gain a bit of AP uptime, but enough of it goes to evo etc that it's no longer worth it. I doubt I'm good enough with simc but checking how many AB casts are empowered with AP in both cases should tell us if that's the case.