Rawr.Mage - Page 107

Maje · 06/28/11, 4:48 AM

There is one thing that's missing in the mage module and that is Armor switching for Fire/Frost models, I wasn't convinced when Simc. output showed it to be a dps gain compared to evocation and I even tried dry calculating the mp5/dps for each mode and received very close results at least for full evocation, partial one's ie. 2 ticks exactly always showed a gain for evocation. A way to model it in rawr would be very useful.

Kavan · 06/28/11, 7:26 AM

Yea I've been planning to do that for a while, but there's some technical limitations. What's easy to do is to just treat mage armor and molten armor as two different states similar to cooldowns and optimize on that, but it wouldn't take into account any temporal restrictions. The problem is setting up a constraint for time lost on the gcd when switching. I have some ideas, but nothing quite workable yet. If anyone has any suggestions on how to tackle this I'd be interested to hear some ideas.

Mangara · 06/28/11, 9:37 AM

The state implementation would already give a good indication of how much time we want to spend in each armor, so it would be useful as a first step before working out the constraints. You also said earlier that a limited number of switches would be possible to implement and I believe that 2 switches (Molten initially -> Mage when mana gets too low -> Molten for final burn) would cover most fights. The maximum number of such switches could be a user option as well, right?

Hibbo · 06/28/11, 10:22 AM

Today I played arround with Simulationcraft and Rawr and recognized a big gap between the scale factor of Int in Rawr and Simulationcraft since 4.2. I tried to setup the same gear and fight duration in Rawr ( Rawr Repository: BIS 391 Simcraft Gear Profile)as in the Simulationcraft profile provided by Lhivera (Simulationcraft Results) but it still feels very odd to me and I can`t find the reason.

Scale Factor/Stat Value for Int in Simulationcraft: 3.6921
Scale Factor/Stat Value for Int in Rawr: 4.96

I already checked the cast time of AB, total Damage done of AB and multiplier of AB Stacks and everything is about the same. The only real difference i can find at the mana sources / ressource gains:

Simulationcraft:
Blessing of Might: 29179
Evocation: 251033
Initial Mana: 123770 (Flask: 4725 Food: 1417 listed seperated)
Mage Armor: 437483
Mana Gem: 45299
MoE: 27207
Replenishment: 60840
Mp5: 77922

Rawr:
Replenishment: 474751 (very high)
Evocation: 225229
Initial Mana: 128659
Mana Gem: 36308
Mp5: 29340
Mage Armor: missing but under Spell Cycles it is vissible that Mage Armor is used
MoE: missing

Kavan · 06/28/11, 2:12 PM

I haven't touched the mana sources graph since Wrath so there's probably some errors in it, but I'll take a look to make sure if any of those are actual calculations errors. Also the by spell breakdown isn't accurate for arcane at the moment because it doesn't take into account mana adept, another thing on my todo list.

Kavan · 07/01/11, 5:40 PM

Originally Posted by Maje

There is one thing that's missing in the mage module and that is Armor switching for Fire/Frost models, I wasn't convinced when Simc. output showed it to be a dps gain compared to evocation and I even tried dry calculating the mp5/dps for each mode and received very close results at least for full evocation, partial one's ie. 2 ticks exactly always showed a gain for evocation. A way to model it in rawr would be very useful.

I've implemented armor swapping without taking into account gcd loss. For one frost example that I had the dps changed from 31107 without armor swapping to 31149 with swapping. Considering it takes 2 gcds for a regen swap and back to molten that would have cost about 2 * 1.24 * 24k ~= 60k damage in the whole fight which is about 130 dps. So this would seem to confirm your suspicion that armor swapping is not optimal and instead one should use partial evocations.

Originally Posted by Hibbo

The only real difference i can find at the mana sources / ressource gains:

I've checked what is going on and the way it displays things is for Replenishment it shows all sources that are based on %mana, so what you're really seeing there is both Replenishment and Mage Armor. Also MoE is calculated as reduction in spell cost, that's why it doesn't show as a mana source. I'll see if I can clean things a bit to make it more relevant to current situation.

Pikenzi · 07/03/11, 7:13 PM

The stats on the Flickering Cowl cloth head are wrong in rawr. Here are the stats of a piece that dropped in our alt raid:

(Click to enlarge)

In RAWR this piece has 359 Int, 248 Crit, 248 Hit. The stats of the heroic version I expect to be wrong as well since they seem high in comparison to the tier.

Elge · 07/03/11, 7:42 PM

Originally Posted by Pikenzi

stats on the Flickering Cowl cloth head

This item gets random stats, exactly like the loot from Conclave and Al'Akir.

Chaosandfury · 07/03/11, 7:45 PM

Originally Posted by Pikenzi

The stats on the Flickering Cowl cloth head are wrong in rawr. Here are the stats of a piece that dropped in our alt raid:

(Click to enlarge)

In RAWR this piece has 359 Int, 248 Crit, 248 Hit. The stats of the heroic version I expect to be wrong as well since they seem high in comparison to the tier.

Me and Spart we're talking about this yesterday, the cowl does indeed have incorrect stats, saw a couple sims using 4pc&cowl, instead of 4pc& ragnaros gloves which is just absurd.

I Imagine it having 400 or slightly less intellect (same as that of the other helms that are 391) None of the other helms have a 30 int socket bonus and afaik this is included in the item budget.

Originally Posted by Elge

This item gets random stats, exactly like the loot from Conclave and Al'Akir.

He is referring to the intellect on the helm and the the #total for secondary stats.

Hinalover · 07/03/11, 9:02 PM

Originally Posted by Elge

This item gets random stats, exactly like the loot from Conclave and Al'Akir.

I made a point when I posted that list that I was using guestimate ids. I was especially leary about the items that had gem slots in them since those were giving me a hard time of figuring out the right ids. I currently have the correct ids in my local build and they should be posted in the next release. One point of interest, is that due to blizzards oddball way of rounding some of the ids are 1 point off in their secondary stats. I've mentioned this to Kavan since he was the one who originally set up the random enchant section of Rawr.

Tkmstr1009 · 07/09/11, 6:01 PM

For some reason Rawr is only showing surviveability for gear and rating my DPS/Damage as 0.00. It started doing this days ago and I can't for the life of me figure out how to fix this. Here's a screenshot.

Mindcrime · 07/09/11, 7:14 PM

Originally Posted by Tkmstr1009

For some reason Rawr is only showing surviveability for gear and rating my DPS/Damage as 0.00. It started doing this days ago and I can't for the life of me figure out how to fix this. Here's a screenshot.

Had the same issue. Problem for me was it could no longer automatically read my talent spec when loading my mage from Battle.net any more, so I had to manually input the spec. After that it worked fine again.

Kavan · 07/09/11, 9:27 PM

I identified a significant problem with the proc effects calculation in Arcane model. Basically the procs weren't properly interacting with Mana Adept. This will have some very big implications in some cases. This change should go live with the 4.2.1 release tomorrow.

ItaliaPerSempre · 07/12/11, 3:54 PM

I'm using the online version of Rawr as a fire mage and it's showing me that the heroic t11 shoulders,chest, hands, and legs are bis still, beating out even t12 heroic (t11 normals outrank t12 heroic at that). I was hoping to get a confirmation that this is either true or there's something wrong with the formula's and such.

Astrylian · 07/12/11, 4:00 PM

Originally Posted by ItaliaPerSempre

I'm using the online version of Rawr as a fire mage and it's showing me that the heroic t11 shoulders,chest, hands, and legs are bis still, beating out even t12 heroic (t11 normals outrank t12 heroic at that). I was hoping to get a confirmation that this is either true or there's something wrong with the formula's and such.

Rawr