Rawr (v2.3.23 released on 9/12) - Page 61

Bashui · 07/28/08, 11:26 AM

There were several problems with the 14.1 bear threat model, and this one should be much more realistic. On testing that I did on my own it was within a few percent of threat reported by in game addons. It still doesn't account for block, so it is still a bit of an over estimate.

Astrylian · 07/28/08, 11:44 AM

Re Haste in RestoSham, I noticed that too, and I think it's a bug. I'm hoping we can get it fixed for 15.1.
Re Resistance from MotW, I'll take a look at that and try to get it fixed for b15.1.
Re TPS in Bear, yeah, this is "righter", the previous version only considered 5% boss avoidance or something minuscule. It should include parry. It massively overestimated threat in 14.1 (though it was proportionally correct, so at least up to its overly-small exp cap, it was accurate for comparisons).
Re Total Mitigation being slightly different, yeah, there was a minor error in the calcs for this (not capping one of the possible swing outcomes' effect on total damage), which I corrected. Not a big deal, but you will see very slightly different values between 14.1 and 15.0.
Re Survival being different... Hmm, I don't recall anything that should make Survival any different... Perhaps there's a buff different, or we had the wrong value for one buff and we fixed in in 15.0? Can you compare the basic stats and see if you can tell what changed? The only two things that effect Survival are Health and Mitigation (armor).

• frmorrison · 07/28/08, 12:33 PM

Originally Posted by Zedd

I added 1minute as fight and gave myself 500mp5 from a SP. It still suggests mp5 gems over haste

Rawr does that for the healadin module as well, even adding lots of mana buffs.

I just delete straight mp5 gemmed items from the item comparison list.

ixnorp · 07/28/08, 12:41 PM

Rawr HolyPriest b15.0
This entire model seems fundamentally inaccurate. I double checked to make sure that Rawr has all of my gemming, enchanting, and items correct. I also double checked the listed values of Spirit and +Heal for all pieces of gear I am wearing to see if they are accurate and they are. With no buffs, Rawr reports my healing as 2418 while I have 2420 on the live server. Rawr lists 615 spirit, WoW lists 622.

Changing my race from Night Elf to Human increases my Spirit to 638 in Rawr. The Human Spirit applies to total spirit and not just base spirit.

Unbuffed, my character sheet lists my Holy spell crit as 11.66% while Rawr lists 12.30%

Unbuffed, I have 10500 mana. Rawr lists 9930.

My character sheet lists 717 oFSR, 257 iFSR. Rawr lists 711 and 255.

Since the base stats are wrong, all the derived calculations are going to be wrong as well. It also doesn't seem to place all that much emphasis on Regen. The "Time in FSR" slider doesn't really do much since time to OOM doesn't seem to be modeled yet. Redeemer's Alchemist Stone is listed as worse than Essence of the Martyr since it doesn't seem to have the regen portion modeled.

Astrylian · 07/28/08, 1:21 PM

Thanks very much for the feedback. This is only the first public release of Rawr.HolyPriest, so there are going to be some slight inaccuracies like that, and we'll do our best to get these fixed.

kalbear · 07/28/08, 2:08 PM

Re Total Mitigation being slightly different, yeah, there was a minor error in the calcs for this (not capping one of the possible swing outcomes' effect on total damage), which I corrected. Not a big deal, but you will see very slightly different values between 14.1 and 15.0.

I don't think that's what the original poster was talking about. I'm seeing something similar. When I do my character with just normal raidbuffs, no crushless mobs/dual-wielders, it heavily overvalues avoidance over mitigation relative to the prior model. I'm curious what changed to flip the values so much in favor of avoidance. The dodge + miss is still only 61% in this case. What changed in this calculation between the two models to favor avoidance so much more?

ETA: even odder is the behavior when I use 'enforce metagem requirements'. In that case, all the avoidance recommendations are reversed entirely and we get back to full stam. I'm sure there's an explanation for this, but the behavior is very counterintuitive.

Astrylian · 07/28/08, 3:06 PM

First, "Enforce Metagem Requirements": The only thing this does is, if turned on, requires you to meet the requirements of your metagem in order to get its bonus. If turned off, you just always get the metagem bonus. Obviously, if you have an 18sta metagem, and don't meet the requirements of it, any item with a blue gem for a slot where you don't have a blue gem, is going to be rated alot higher. That's probably what you're seeing, and is accurate, and hasn't changed.

Second, the difference between 14.1 and 15.0 mitigation: I see no difference that "heavily overvalues" avoidance. Can you give examples? The only difference is as follows:
The order of items on the combat table was off for crits vs hits; you couldn't become unhittable without becoming uncrittable in 14.1. That's corrected in 15.0; after avoidance finishes pushing hits off the table, it will push crits off as well. I doubt you're seeing this change, because you have to have at least 97.4% avoidance for this to have any effect.

While debugging though this just now, though, I discovered one additional bug that I'll have fixed for 15.1, but again is not causing any sweeping changes. Negative chance to crit was rolling over onto hits, up to 1.2%. If you had 3.6% anticrit, giving you a -1% chance to be crit, that was acting as an extra 1% hit avoidance.

EDIT: So others can double-check my math, here's the math that will be used in b15.1, in pseudo-code:

cappedCritReduction = max(2.6, critReduction)
avoidance = dodge + miss
crits = min( max(0, 100 - avoidance), 2.6 - cappedCritReduction)
crushes = max( 0, min(15, 100 - (crits + avoidance)) - crushReduction)    //For 73s. Set to 0 for 72s and lower
hits = max( 0, 100 - (crits + crushes + avoidance))

damageFromCrits = crits * (100 - cappedMitigation) * 0.02
damageFromCrushes = crushes * (100 - cappedMitigation) * 0.015
damageFromHits = hits * (100 - cappedMitigation) * 0.01
DamageTaken = damageFromCrits + damageFromCrushes + damageFromHits

kalbear · 07/28/08, 3:16 PM

First, "Enforce Metagem Requirements": The only thing this does is, if turned on, requires you to meet the requirements of your metagem in order to get its bonus. If turned off, you just always get the metagem bonus. Obviously, if you have an 18sta metagem, and don't meet the requirements of it, any item with a blue gem for a slot where you don't have a blue gem, is going to be rated alot higher. That's probably what you're seeing, and is accurate, and hasn't changed.

No, that's not it. My gear currently is fine with a PED and the gems to use it. When I do not use the enforce metagem, it recommends I go to a Relentless and gem for agi everywhere. When I do enforce it, it recommends I stay with stam. So in this case, it's saying I should stay with my current gear since I meet the requirements, but if I could get the agility I should go for it? That seems a bit odd, and it's especially surprising that it affects all gear choices, not just the helm. I would expect that if I were on the cusp that much, agility would be favored regardless of what my metagem was. And it's not like I don't have enough red gems for the relentless as it is.

To sum up:

when I have enforce metagems on, all my gear is recommended to use stamina gems or stam/x depending on the slot bonus. I already have met the metagem requirements wearing my current gear, so this is essentially saying 'stack stam'.

When I have enforce metagems off, it recommends switching to a relentless, gemming everywhere with agility, and not recommending a single yellow gem anywhere.

Also, if I load a feral druid in with the holy priest model, I get the 'unhandled object' error.

rayous · 07/28/08, 3:28 PM

I have found that the Tree section of rawr does not include a large amount of gear, and doesnt seem to be rating healing. It does seem to be rating mana regen, but thats it.

Astrylian · 07/28/08, 3:36 PM

"Enforce Metagem Requirements": The only thing this does is, if turned on, requires you to meet the requirements of your metagem in order to get its bonus. If turned off, you just always get the metagem bonus.

That *completely* explains your situation, look closely.

When you have it off, your metagem's bonus is included regardless of what other gems you have, so it rates gear with agi gems higher than your gear with stam gems, because you're not going to 'lose' your metagem bonus by dropping blue gems. And it rates helms with relentless metas higher, because you're going to get the effect of either one, and 12agi is better than 18sta for you.

When you have it on, your metagem bonus is only included if you have the gems to support it, so it rates gear with agi gems lower than your current gear with stam gems, because making that swap, even though 10agi may be better than 15sta, you'd also lose your metagem bonus if you lost that blue gem, and 10agi is certainly not better than 33sta. Same with the metagems, you don't have the gems to support the relentless, so obviously 15sta is better than nothing.

Bottom line is that Enforce Metagem Requirements is the truly accurate mode, but it's an option that you can turn off in order to see 'expected' results, to compare items without respect for their effects on metagems.

It sounds like in your case, agi is better than stam, and you may be better off making the switch to relentless, and making the required gem changes. The Optimizer is the intended feature for doing this in one step, and finding the best way to do it, while also allowing for requring a certain level of survival for the bosses you're fighting, with the healers you have.

EDIT: "Also, if I load a feral druid in with the holy priest model, I get the 'unhandled object' error." -- I'll get that fixed, thanks.

Promii · 07/28/08, 3:43 PM

I'm normally feral but I noticed when trying out my healing gear in the Tree model that it seems to be massively overvaluing spirit or undervaluing +heal.

I tried loading one of the full-time restos from my guild and it's rating [Sparkling Star of Elune] higher than [Teardrop Crimson Spinel] for a druid with 2100 +heal and 260 in combat regen.

kalbear · 07/28/08, 3:44 PM

When you have it on, your metagem bonus is only included if you have the gems to support it, so it rates gear with agi gems lower than your current gear with stam gems, because making that swap, even though 10agi may be better than 15sta, you'd also lose your metagem bonus if you lost that blue gem

That's where I'm saying it's wrong. I have at least 6 blue gems in my current setup. When I say 'enforce metagems' it recommends I remove NONE. (in fact, it recommends I add even more blue gems to the mix). If I say 'do not enforce metagems" it recommends I remove all of them (and add a relentless). That's my point here; I'd understand it if it were recommending going all agility (or even agi/stam gems) in places, but it's not doing that; it's recommending staying with the current blue gems no matter how many I have. That seems very odd to me. There may be a reasonable explanation, but as I said it's very counterintuitive - and it's not the same behavior that was in 14.1 with the same gear.

Astrylian · 07/28/08, 3:46 PM

I have found that the Tree section of rawr does not include a large amount of gear, and doesnt seem to be rating healing. It does seem to be rating mana regen, but thats it.

Can you give some examples? It does seem to be for me, but I'm not much of a Tree, I'll pass that onto the Tree dev. I know how it currently rates haste is kind of odd, but that's to be expected to some extent, due to how the value of haste has several cutoff points for fitting more casts into your cycle.

EDIT: Kalbear, please send me the character file in question, and some specific examples, and I can give you the exact answers for those situations.

Mijae · 07/28/08, 3:55 PM

Originally Posted by Astrylian

While debugging though this just now, though, I discovered one additional bug that I'll have fixed for 15.1, but again is not causing any sweeping changes. Negative chance to crit was rolling over onto hits, up to 1.2%. If you had 3.6% anticrit, giving you a -1% chance to be crit, that was acting as an extra 1% hit avoidance.

EDIT: So others can double-check my math, here's the math that will be used in b15.1, in pseudo-code:

cappedCritReduction = max(2.6, critReduction)
avoidance = dodge + miss
crits = min( max(0, 100 - avoidance), 2.6 - cappedCritReduction)
crushes = max( 0, min(15, 100 - (crits + avoidance)) - crushReduction)    //For 73s. Set to 0 for 72s and lower
hits = max( 0, 100 - (crits + crushes + avoidance))

damageFromCrits = crits * (100 - cappedMitigation) * 0.02
damageFromCrushes = crushes * (100 - cappedMitigation) * 0.015
damageFromHits = hits * (100 - cappedMitigation) * 0.01
DamageTaken = damageFromCrits + damageFromCrushes + damageFromHits

While already correct, wouldn't it be cleaner to format all the calculations similar? Changing the crits formula would mean you don't need a "cappedCritReduction" variable.

crits = max( 0, min(2.6, 100 - avoidance) - critReduction)

Hopefully someone should be over the crit reduction cap, but if not you should technically also model damage reduction on crits from resilience as well.

EDIT: You'll also want that 2.6 crit chance adjustable for non-73s as well.

Astrylian · 07/28/08, 4:36 PM

Ah, yeah, I could prolly invert that to be the same as the others. But yes, the 2.6's are adjusted by level, I just put 2.6 in there for the pseudo-code.

And true, should probably model the reduction in damage from crits from resil... Perhaps later.