Cataclysm Mage Simulators and Formulators - Page 5

Lhivera · 12/21/10, 8:11 PM

I'm running new output at the URL listed in the original post. The cleanup, bug fixes, and changes to priority lists over the past couple weeks have narrowed the range for the specs quite a bit; here's a sneak preview of the T11 372 Basic output:

26628 (100.0%): Fire
26317 (98.83%): Fire FFB
25097 (94.25%): Frost
25005 (93.90%): Frost FFB
24953 (93.71%): Arcane

Tyrian · 12/21/10, 8:24 PM

I like the look of those figures, but can we have some insight into the DPS variance? How much have the recent changes (if at all) changed/narrowed the large variance the Fire spec has. Do you view FFB-Frost being that close to Frostbolt-Frost as a concern? I was under the impression Blizzard wanted Frostbolt to be the clear choice for primary nuke.

Lhivera · 12/21/10, 9:20 PM

Fire's range is still pretty large, over 26% (vs about half that for Arcane and only about 11% for Frost). It's possible that's considered one of the trade-offs for higher mobility and cleave damage (Fire's advantage increases over both Arcane and Frost as you add just about any kind of raid event). Higher potential, lower reliability.

I don't know about Blizzard, but I considered it a significant issue when FFB was superior to Frostbolt in beta. I don't consider it an issue if it's equal to Frostbolt, as that allows people to play the spec either way. Choice is good.

Maje · 12/23/10, 3:58 AM

The numbers look ok, I need to take another look at Living Bomb after the change.

The formula for T3 Hot Streak is no longer correct from my perliminary tests it seems the graph has been shifted downwards, but there isn't enough information yet to regress a new function.

Combustion is a whole bag of worms, I have no idea how exactly it's calculated nor does anyone as far as I know, the most recent test was one by Zaldinar see (Cataclysm Fire Mage Compendium) but it hasn't shed much light on the subject except to verify that it's wonky.

Ignite is still horribly munched, statistically looking at WoL makes it out to be about the same percentage of the total DPS as Living Bomb, the closest approximation I found was to force aura_delay=1.5, horrible I know. In summation the munching is responsible for a loss of about 5+% of the total mage dps or about 25% of the damage done by ignite.

Now to more concrete information, the total mana pool looks to be incorrect in Simcraft (lower than live) and I think the stamina as well, I will have more information tonight.

Roywyn · 12/23/10, 9:33 AM

Originally Posted by Maje

Now to more concrete information, the total mana pool looks to be incorrect in Simcraft (lower than live) and I think the stamina as well, I will have more information tonight.

The simcraft health and mana numbers are definitely off there, mana should be as follows:

Mana = (17418 + 20*1 + (INT-20)*15 + 2126)*(1+0.02)*(1+0.05)

17418 is the base mana for (all races of) mages at 85. The first 20 points of INT provide 1 point of mana each, every point past 20 provides 15. 2126 is Arcane Brilliance. 1.02 and 1.05 are the Ember/Gnome multipliers, they are multiplicative for Gnomes with an Ember Diamond.

Health is similar. Note that the Armoury mouseover window is faulty. If you hover over your stamina entry, it'll say that your STA stamina provides 20*1+(STA-20)*10 HP, while it actually does provide 20*1+(STA-20)*14 HP. I'm not aware of direct health buffs, Blood Pact now provides stamina as well.

Health = 37113 + 20*1 + (STA-20)*14

I would suggest double checking those numbers as well. Base Health should be the same for any race at 85, I have not verified it for all races however.

talchas · 12/23/10, 1:27 PM

Originally Posted by Maje

Combustion is a whole bag of worms, I have no idea how exactly it's calculated nor does anyone as far as I know, the most recent test was one by Zaldinar see (Cataclysm Fire Mage Compendium) but it hasn't shed much light on the subject except to verify that it's wonky.

I've done a little testing myself, and "it's wonky" is a pretty good description.

A crit combustion tick does not proc ignite. It appears not to double-dip fire power for Ignite or LB.

First, combusting ignite does almost exactly what it says - (under constant mastery) an ignite doing X damage every 2s will combust into a dot doing X/2 damage per tick, for as many ticks as you get for your combustion. If you increase mastery between applying ignite and combusting it, you wind up with

2*combust_tick = ignite_tick/(1+old_mastery)*(1+new_mastery)

Boosting intellect in between does nothing, and boosting haste just gives you more combustion ticks. \

Thats the good news. For living bomb and pyroblast it appears to only check for the presence of the dot and then add damage as if you had cast the dot with your current stats. None of crit/hit/haste will affect the size of a normal combustion tick that I have found, but boosting intellect or mastery between the dot cast and the subsequent combustion will boost the combustion damage.

Now the bad news is that I did all of my testing with critical mass talented (although without the glyph). Without critical mass, I get (m being mastery, .2 base):

tick_without_CM=tick_with_CM/(1.15+m)*(1+m)

Which at least makes some sense, given thats what we get for LB.

I first ran a linear fit without CM, which gives me the worthless

combustion_tick*3=.6937185557795573*m*s+.2455850493829304*s+1297.028608363312*m+487.7207069840994

However, running instead a double-dipping-mastery fit gave me:

combustion_tick*3=(.2678930839158054*spellpower+518.6158840934165) * (1+mastery)^2

Accounting for CM this is:

combustion_tick*3=(.2678930839158054*spellpower+518.6158840934165) * (1+mastery)*(1+mastery+CM)

This makes a lot more sense and also fits extremely well - accounting for having to deal with rounding issues with both equations, you would probably need to have around .4 total mastery bonus to notice the difference, which I can't test with my gear. Assuming its the latter, note that according to simc and agreed to by my tests, we have living bomb as (after firepower and fire specialization):

LB_tick = (.300*spellpower + 518.93) * (1+mastery + CM)

EDIT: I just tested pyro, and again it double dips on mastery but has a low coefficient:

combustion_tick*3 ~= (.1005*spellpower + 283.9)* (1+mastery)^2
pyroblast_tick = (.112*spellpower + 283.6) * (1+mastery)

TLDR:
Combustion double dips mastery on LB/pyro, but not ignite. Does not double-dip haste. "Double dips" crit in the sense that more crit means more ignites to combust, but that is all. Does not proc ignite. SP and mastery procs affect combustion on its cast, not on dot cast. The SP scaling ignoring mastery is a bit low.

angayelle · 12/24/10, 5:15 AM

Talchas, do you mind if I use theses formulas into CombustionHelper to predict upcoming Combustion damage ? I haven't be able to get something remotely accurate so far.

talchas · 12/24/10, 1:40 PM

No problem. I should explicitly check that they just sum, but I'd be surprised if they didn't. I haven't checked FFB either, but my guess would be that it would behave like pyro/LB.

EDIT:
Ok, when LB and Pyro are summed up, the results seem to be about 3% high - possibly doubledipping Fire Power in this case? LB+Pyro+Ignite results in 3% higher LB+Pyro combust damage plus normal ignite-combust damage.

EDIT2:
Mastery procs and ignite, not too surprising:

combustion_tick * 2 = ignite_tick / ignite_cast_mastery * combustion_cast_mastery

As mentioned in the previous post, LB/pyro + mastery procs just use the combustion_cast_mastery throughout the computation.

FFB dot + combustion: does utterly trivial damage - 18-19 dps at 5563 sp, 13.68 mastery, with 3 stacks of the dot ticking for a total of 264 (at one stack combustion does even less). It doesn't seem to matter how strong the FFB dot was, but the variation in dot strength is small enough that I could well just not see it. LB+FFB and Pyro+FFB do not trigger the extra 3% damage that LB+Pyro does (LB+Pyro+FFB still is affected).

radikal · 12/28/10, 1:35 PM

Is there an easy way in simcraft to change deep freeze's behavior to the non-stun-immune version? I'm interested in stat scaling solely within deep freeze stuns. (It's a little messy as you need to first lock haste so that you can clip the 5s window -- but I'm curious how crit/mastery/int compare against only frozen targets -- under both the assumption that you lead with an EF bolt and when you don't I guess)

dedmonwakeen · 12/30/10, 12:43 PM

Originally Posted by radikal

Is there an easy way in simcraft to change deep freeze's behavior to the non-stun-immune version? I'm interested in stat scaling solely within deep freeze stuns. (It's a little messy as you need to first lock haste so that you can clip the 5s window -- but I'm curious how crit/mastery/int compare against only frozen targets -- under both the assumption that you lead with an EF bolt and when you don't I guess)

Not at the moment. However, if you open an Issue, then I (or someone) will make deep-freeze behavior depend upon target_level.

We are juggling tons of new development with only a few developers so the best way to ensure something gets done is to open Issue.

dedmonwakeen · 12/30/10, 12:44 PM

Originally Posted by Roywyn

The simcraft health and mana numbers are definitely off there.

These problems have been identified and will be addressed prior to next release, most likely sometime this evening.

EDIT: Fixes were committed (r6146) but missed the release.

Kavan · 01/03/11, 12:06 PM

I noticed that in simc pyroblast_hs does not have may_hot_streak set to true, only for the hard cast pyroblast. I was under assumption that both count towards hot streak. Can someone clarify if this is the actual behavior in game?

Maje · 01/04/11, 3:39 AM

Only hard cast Pyroblasts count towards HS, P! doesn't reset or increment the counter.

Lhivera · 01/04/11, 10:12 AM

Output linked in the original post is being updated right now using the latest build. I tweaked the Frost profiles to switch to Mage Armor a bit later, as the mana pool fix gave them some more breathing room. It's certainly possible that further improvements can now be made to all action lists.

Maje · 01/04/11, 11:09 AM

The T3 formula should probably be updated to the new finds see Cataclysm Fire Mage Compendium for more information. The current best approximation is -2.73c+0.95.