WotLK - Complete Mage Compendium (3.3.3 live) - Page 71

Kavan · 09/04/08, 5:49 PM

From my testing AM is bugged currently regarding Arcane Potency. If you do single casts the 30% will be applied to the whole AM that procs the CC, and not to the one that consumes CC. However when you chain cast AM potency only applies to one wave/missile of AM. It seems like their code removes potency when AM channel ends and due to how client/server comm works it seems that end of last wave is processed after new wave is started and thus shortcircuits the buff.

Faxmonkey · 09/04/08, 6:11 PM

Originally Posted by Xentropy

Another reason I know these full resists were not table-based is that due to the new 10%-step partial resist tables the chances of a large partial resist are far smaller below the top end of resist anyway. I received a bell-curve of partial resists centered around 30%, with few 10% or 50% resists, and none I noticed over 60% other than the full resists on frostbolt. If there truly was a table with some small chance of a full resist, there should have been some amount of 80-90% resists as well trailing off toward them, and ice lance should have experienced them too. Ice lance was my "control", basically, a frost damaging spell without a binary component.

And it isn't simply a matter of binary spells getting compensated for their chance to full resist by getting reduced partial resists, since the amount of damage partially resisted by frostbolt and icelance was similar. Frostbolt simply gets an EXTRA penalty for being binary, at the moment. That's why I classified it as a bug and not just mentioned it as a quirk of the system. It seems to me an oversight.

I think I must have misunderstood some aspects of your original post. For instance I thought that you were saying ha you had experienced a 30% overall mitigation not counting full resists. I assumed meant plenty of partial resists above 30% in order to offset the ones below.

Xentropy · 09/04/08, 6:21 PM

It did. But with a finer degree of steps of partial resist possible, the bell curve is much steeper than it used to be. No longer will the 6% resists against bosses mean ever seeing a 50% resist; instead it seems to be approximately 45% no resist, half 10% resist, and 5% 20% resist (or similar).

With frostbolts and ice lances, I was seeing about 10% of the time 10% or 50% resist, 20% of the time 20% or 40% resist, and 40% of the time 30% resist. I don't remember ever seeing a full non-resist, and 60% resists were extremely rare. All of them fell in the 10-60% range, giving an average of bit over 30%. No resists of 70%, 80%, or 90% were ever seen in testing, so it wouldn't make sense, even ignoring the icelance data, for 9.5% of casts to show up as 100% resists without *any* data points in the 70-90% range. Granted this was only 200 data points, but I find it highly unlikely further testing would suddenly see >10% of the hits be 70-90% resisted, and all it takes is one person seeing a full resist from ice lance to invalidate the theory that only spells with binary components can do so, so if anyone would like to do more data collection, please have at it.

dedmonwakeen · 09/04/08, 6:23 PM

SimulationCraft r225 available for download......

I've discovered the reason why my sim was valuing FFB so high: I was not combining Burnout and Ice Shards correctly.

It does make me wonder..... Perhaps Blizzard should combine Burnout and Ice Shards in a multiplicative manner instead of the way in which Spell Power and Ice Shards are combined. It is possible that would create an end-end-game place for FFB beyond just gimmicks: Essentially, FFB would eventually become more effective than Fireball provided the height of gearing was itemized for sufficient crit.

Oh..... the inclusion of JoW solves the Fire Mage mana issues quite easily:

Xentropy · 09/04/08, 6:44 PM

Very interesting. I figured out part of what's penalizing frost so harshly in Simulationcraft compared to, say, magegraf. The lag simulation will of course hurt faster casts more than slower ones since it's functionally adding a static amount to the cast time of each spell. Lowering lag to 0-100ms from the default 150-250ms and correcting frost's talent tree (it doesn't have frost channeling, so it's evocating, whereas with frost channeling it does not evocate), a 360-second fight still results in 5% downtime for fire. Switching to mage armor saves it an evocation and keeps it downtime-free, while increasing its dps for a 6 minute fight. I also again doubled hit ratings from 200 to 400 to better approach the cap. I get the following, which makes frost look a lot more viable:

This still doesn't account for FoF-combos which would again add about 50dps to frost. It seems those with bad connections may specifically need to avoid frost at the time-being, as it'll affect both the viability of FoF-combos and penalize frost most heavily by adding cast time to more overall casts.

I LOVE your program, dedmon. It makes tweaking little things and seeing their results easy. The only thing I'd like to see would be (and this may already exist, I'm just not sure how to set the configuration file up to do it) a way to give every build the same stats without having to change the numbers in multiple places, so I could see how gearing in various ways affects the various specs. For instance, I'd expect crit scaling to be highest with frostfire, followed by fire, then frost, then arcane, but to change crit for them all right now requires a search and replace rather than just one line change.

dedmonwakeen · 09/04/08, 7:26 PM

Originally Posted by Xentropy

This still doesn't account for FoF-combos which would again add about 50dps to frost. It seems those with bad connections may specifically need to avoid frost at the time-being, as it'll affect both the viability of FoF-combos and penalize frost most heavily by adding cast time to more overall casts.

I finally have a design for FoF-combos I'm happy with and I'll be implementing it later tonight.

I LOVE your program, dedmon. It makes tweaking little things and seeing their results easy. The only thing I'd like to see would be (and this may already exist, I'm just not sure how to set the configuration file up to do it) a way to give every build the same stats without having to change the numbers in multiple places, so I could see how gearing in various ways affects the various specs. For instance, I'd expect crit scaling to be highest with frostfire, followed by fire, then frost, then arcane, but to change crit for them all right now requires a search and replace rather than just one line change.

At present there is no way to specify "default" gear settings. I'm not sure the most convenient method to implement: class-based defaults, lump casters together, etc.....

One thing I like to do is to use enchant_xxx on the command line. This enables me to boost one or more stats without editing the config file:

simcraft input=raid_wotlk.txt active=Mage_FrostFire enchant_spell_crit_rating=200

aliengrey · 09/04/08, 9:47 PM

I hate to derail the conversation going on here but can someone in the beta confirm that we get new rank 14 fireball and frostbolt at level 70? thanks

Finkum · 09/04/08, 10:06 PM

I hate to derail the conversation going on here but can someone in the beta confirm that we get new rank 14 fireball and frostbolt at level 70? thanks

Yes, you do (for Beta at least, who knows for patch 3.0).

dedmonwakeen · 09/05/08, 12:36 AM

Originally Posted by Xentropy

This still doesn't account for FoF-combos.

Support for "shatter combos" now exists in the simulation engine. The DPET numbers for IL seem awfully high...... probably needs investigation.

Player=Mage_Frost  DPS=3521.3  DPR=19.3  RPS=182.3/171.6  (mana)
  Core Stats:  strength=44  agility=50  stamina=726  intellect=820  spirit=831  health=9800  mana=14443
  Spell Stats:  power=2000  hit=7.6%  crit=23.6%  penetration=0  haste=10.9%  mp5=91
  Attack Stats:  power=34  hit=0.0%  crit=4.0%  expertise=0.0  penetration=0  haste=10.9%
  Actions:
    fire_ball             Count= 16.4|18.2sec  DPE=4472| 7%  DPET=3242  DPR= inf  Miss=1.2%  Hit=3671  CritHit=5507|5510|46.1%
    frost_bolt            Count=118.1| 2.5sec  DPE=6545|73%  DPET=3030  DPR=15.3  Miss=1.3%  Hit=4130  CritHit=8261|8265|59.8%
    ice_lance             Count= 10.2|29.3sec  DPE=6742| 7%  DPET=4871  DPR=33.8  Miss=1.7%  Hit=3481  CritHit=6971|6975|95.1%
    ignite                Count=  0.0| 0.0sec  DPE= inf| 2%  DPET= inf  DPR= inf  Miss=nan%  Tick=1245
   water_elemental
    water_bolt            Count= 81.0|3007.8sec  DPE=1464|11%  DPET= 639  DPR= inf  Miss=12.0%  Hit=1531  CritHit=2297|2306|15.4%

Vontre · 09/05/08, 12:57 AM

Where is Deep Freeze?

The huge discrepancy between magegraf and your sim is concerning. Do you have more detailed records of what's going on?

Roywyn · 09/05/08, 2:35 AM

Originally Posted by dedmonwakeen

Support for "shatter combos" now exists in the simulation engine. The DPET numbers for IL seem awfully high...... probably needs investigation.

Not sure, check coefficients maybe?
Ice Lance scales like a 0.5s spell, so that when tripled, it scales like a 1.5s spell.

In my sheets, using Ice Lance as 3rd charge of FoF (when DF is on cooldown) is still a DPS gain, but a very very minor one only. Damage per cast time (GCD) of a FoF Ice Lance is only slightly better that the DPS of the whole frost cycle.

The 3rd FoF charge only gives a very noticable benefit with deep freeze.

epiphenom · 09/05/08, 2:42 AM

Originally Posted by dedmonwakeen

Support for "shatter combos" now exists in the simulation engine. The DPET numbers for IL seem awfully high...... probably needs investigation.

If I'm understanding the code section correctly, hits of 3481 and crits of 6971 are way off. Perhaps you're coefficienting it at 43%, when it's actually (1.5/3.5)/3 to account for the tripling?

Edit: Roywyn beat me to it. Serves me right for leaving the window open.

dedmonwakeen · 09/05/08, 4:47 AM

Originally Posted by Roywyn

Not sure, check coefficients maybe?
Ice Lance scales like a 0.5s spell, so that when tripled, it scales like a 1.5s spell.

In my sheets, using Ice Lance as 3rd charge of FoF (when DF is on cooldown) is still a DPS gain, but a very very minor one only. Damage per cast time (GCD) of a FoF Ice Lance is only slightly better that the DPS of the whole frost cycle.

The 3rd FoF charge only gives a very noticable benefit with deep freeze.

Bah...... I need to go back to elementary school....... (1.5/3) != 1.0

Detailed output:

Player=Mage_Frost  DPS=3452.3  DPR=18.9  RPS=183.0/171.9  (mana)
  Core Stats:  strength=44  agility=50  stamina=726  intellect=820  spirit=831  health=9800  mana=14443
  Spell Stats:  power=2000  hit=7.6%  crit=23.6%  penetration=0  haste=10.9%  mp5=91
  Attack Stats:  power=34  hit=0.0%  crit=4.0%  expertise=0.0  penetration=0  haste=10.9%
  Actions:
    deep_freeze           Count=  6.2|48.1sec  DPE=4924| 3%  DPET=3561  DPR=21.6  Miss=1.5%  Hit=2537  CritHit=5076|5077|95.5%
    fire_ball             Count= 16.4|18.1sec  DPE=4473| 7%  DPET=3243  DPR= inf  Miss=1.3%  Hit=3671  CritHit=5507|5510|46.3%
    frost_bolt            Count=117.2| 2.5sec  DPE=6539|75%  DPET=3027  DPR=15.3  Miss=1.4%  Hit=4130  CritHit=8261|8265|59.7%
    ice_lance             Count=  3.8|78.7sec  DPE=4272| 2%  DPET=3086  DPR=21.4  Miss=1.4%  Hit=2194  CritHit=4389|4391|96.0%
    ignite                Count=  0.0| 0.0sec  DPE= inf| 2%  DPET= inf  DPR= inf  Miss=nan%  Tick=1245
   water_elemental
    water_bolt            Count= 81.0|3008.4sec  DPE=1465|12%  DPET= 639  DPR= inf  Miss=12.0%  Hit=1530  CritHit=2298|2310|15.4%

Priority list:

flask,type=pure_death
molten_armor
mana_gem,trigger=3500
mana_potion,trigger=2500
deep_freeze,fb_priority=1
ice_lance,frozen=1,fb_priority=1
evocation,trigger=10000
water_elemental
icy_veins
cold_snap
fire_ball,brain_freeze=1
frost_bolt

EDIT: SVN code repository updated..... I'll put out a new download tomorrow morning.

Valestra · 09/05/08, 5:18 AM

What kind of speccs did you use for Arcane Spam and Arcane (which sits awefully low).
Done any calculations with 34% Haste+ and waeving ArBarr into AB rotations and/or Missle Barrages?
I'll never specc somthing which is 1 button mashing only again. It's so damn mind numbing.

dedmonwakeen · 09/05/08, 5:31 AM

Originally Posted by Valestra

What kind of speccs did you use for Arcane Spam and Arcane (which sits awefully low).
Done any calculations with 34% Haste+ and waeving ArBarr into AB rotations and/or Missle Barrages?
I'll never specc somthing which is 1 button mashing only again. It's so damn mind numbing.

The vanilla Arcane spell priority list was coded as if AB-spam was not sustainable:

arcane_missiles,barrage=1
arcane_barrage
arcane_blast

It was more interesting with the old version of Moonkin Haste...... now it is a bit boring.

The Arcane Spam priority list is just arcane_power/arcane_blast with Mage Armor. Mana is easy..... no need to even Evocate. Obviously busted.