http://elitistjerks.com/f75/t30655-w...m_3_1_updated/ would have been the place to look...Roywyn keeps lots of good info in that first post. But yes, it's a 45s ICD on the 2 piece bonus.

Hey guys,
I't written on all walls in orgrimar and stormwind that spell damage > crit > haste for fire specs.
I'm using the FB/TTW spec and i've seen a rather strange thing going on. As far as i know spell damage scales linearly, the spell gets a % from total spellpower based on the cast time. I think it's (cast_time * 100 / 3.5 sec) with normalisation to 1.5secs and 7 secs. What i seem to find out that most stats tend to bend into a curve and not a linear function when reaching some decent values.
On the other way i can compare myself with another mage in raid, same spec, we trade focuses, that has less 120spell damage but more 100 haste rating and the 4 t7 set bonus. His dps output can easily surpass mine at about 3-400dps while we are doing the same rotations and getting same buffs. I know a lot of dps coming from mages can be purely luck related, i've seen runs with crit -> normal -> crit -> normal so no HS in 2 minutes, but i suppose it all gets to a distribution in the end that should be expanded on a time-span and you get what you loose here later.

The question that remains, can we consider that ~2000k spell power unbuffed, it's worth to give up 111 spell power for 100 haste, though officially the math 1 haste ~ 0.6 spell power sais it's a bad ideea? I would try to model the math but i'm lazy, would need to take in account that more spells -> more crit chances -> more HS procs probability, that is not easy additions and substractions, plus the haste brute value actually scales up with the spell power. I think there is a similar relation between haste and crit.

I think a good explanation would be that for example your 1st 100 haste rating has a bigger impact then your last 100 to reach 400 especially at decent to very high spell power values.

rawr seems to be easily giving up 60-70 spell power for the same amount of haste or crit.

What i can imagine is a 4-dimensional function DPS(spell_power, crit_rating, haste_rating) with a constant TIME in it that is actually a form where we could see which combination brings the best output for a specific spell. What i want to see is the moment where 1 crit rating will be actually worth less then 1 haste rating and they are both more important then 1 spell power even if general rules say that sp>crit>haste for fire.

Does this make any sense to anyone, think it's worth investigating? I can try some matlab testing for this but i'm curious what you guys thing about this before i invest some time in this.

I'll apologise in advance in case this question has been answered before, i did not really looked through all forum posts. If info is somewhere there even a nice keyword for search would be appreciated.

Really, trust in rawr.

You dont provide anything but some of your own assumptions. Saying that you have X more spelldmg and he has X more spellhaste, and you loosing to him in a DPS race doesnt mean jack shit unless we can see some parses from the fight.

Several things can be the reason why he did better than you and one could be that he is simply better at positioning himself and casts generally more spells than you do - times his LBs and Hotstreaks better etc etc. Whatever really.

You should also note, that for a Fireball spec you'd wanna go with the Spellpower > Haste > Crit mindset rather than swapping crit and haste. FFB would favor Crit slightly however.

Also, you'll find the breakeven points for the diffrent stats if you use Rawr aswell.

I know that the comparison between the 2 mages has little relevancy because of too many variables involved but it was only a starting point. What i am more interested is that relation between power/haste/crit. It's more of a curiosity if the general rules can loose their meaning at limits and if there is such a case when that haste being better then crit can become the other way around for some values.

My "math" problem right now is that functions look like this:
HasteToDPS(haste_rating, crit_rating, spell_power,time=const),
CritToDPS(haste_rating, crit_rating, spell_power,time=const)
SpellPowerToDPS(haste_rating, crit_rating, spell_power,time=const)
OverallDPS( Haste, Crit, Spell_power, time=const )
get to too many dimensions to display them. A friend just suggested me to plot the surfaces for a set of constant Spell Powers and see if it makes any sense.

Although, if i got this right, it's a simplified version of rawr mage modeling that can show some conclusions regarding the math we use for gear choice.
We have sheets to transform crit/haste to power and all that to damage, but when does the general rules (power>haste>crit) actually fail, or they never do?

btw this is not a issue of recovering some 300dps difference, most fights rarely work with nuking, but the mechanics should be interesting to now.

"Also, you'll find the breakeven points for the diffrent stats if you use Rawr aswell."
You mean i can find out the actual answer to this problem by using rawr? How? Not really feasible to alter with ratio modifiers the stats 100*100*100 times.

Also the haste>crit for ttw is a new thing for me, thanks for that.

Why don't you just use Rawr's graph feature? It produces graphs that show how your dps will scale with each additional point of haste, crit, sp, int etc, if you are interested in seeing the scaling at extreme limits. Unless I'm way off the mark on what you're trying to do here.

That is close to what i am looking for but it's one degree down. Instead of lines on graphs i would like to obtain a surfaces in space, which shows the dps scaling depending on variation of 3 stats rather then 2. It would be a spell against crit against haste and i would be looking for mountain picks on the surfaces, although as i've seen the graphs in rawr everything seems to be pretty linear and the most probable output would be a an inclined surface mostly higher on the haste slope, getting higher as the spell power grows higher. Imagine you start from a point with 0 haste and 500 crit and a point with 500 haste and 0 crit, with variations to (0 crit 0 haste) to (500 crit 500 haste). Having peaks would mean that there are some points of interesting balance where the stats combined give a more overall then if a value of one is brought down while the other is raised by same value, or just place of turning, where the haste slope would actually go for a small part of the graph beyond the crit slope.

I hope i am not talking chinnese here

The number's i've given were just random values. I was trying to find some information about the way these stats are interacting, not about upgrades rawr suggests, i know RAWR is usually right. I'll follow my research on the field and let you know if i ever reached any conclusions. This discussion has already brought me an "infraction" point for not being able to express myself in the righteous terms, which seems to be enforced by the feedback i've received. I hope things will be clear when i'll have the space surfaces i'm talking about. I've seen some people posting some math formulas on other topics and i though i hit the jackpot. I'm sorry for creating any confusion.

I think the point being made is that Rawr computes stat equivalence values (or rather, computes the DPS gain for +1 spell power, +1 crit, +1 haste, etc., which is used as a basis of comparison--using such computations, you can figure out whether X spell power > Y haste or not).

To answer your more theoretical question, dotout, the standard average DPS function looks like this:



Where h is hit, q represents flat multipliers on damage, m is the base damage, r is the spell power coefficient, d is spell power, b is the crit bonus (crit multiplier - 1), c is crit chance, z is haste. (Hit, crit, and haste must be expressed as decimals, not percentages.) t is the base casting time.

This expresses the average DPS of a single spell. For rotations, we must have a weighted average of individual spells' DPS (weighted by time spent casting those spells). For things like Hot Streak, the weights themselves are a function of crit (the more crit you have, the more time is spent casting Pyroblast, not Fireball/Frostfire Bolt). This, in general, can lead to complicated nonlinear scaling, but as you can see, for any given stat, DPS of a given spell increases linearly as that stat increases linearly. For combinations of stats, DPS increases as a polynomial trend of the same order as the number of stats being increased (two stats = quadratic scaling, 3 stats = cubic, etc.). Again, overall DPS of a rotation is much, much more complicated.

The other thing you should consider is how you norm this coordinate space. In my opinion, the only natural norm is the itemization budget norm (a ln(2)/ln(1.5) norm). This is useful in that, for stat combinations of constant itemization cost, you can find (in theory) the best DPS for the least itemization cost.

I have not seen any specific posts on this, but the topic of maximizing damage on General Vezax has sparked in my head. I saw a video of the hard mode kill of him and the mages all went frost and went for efficiency since you can not drink.

Would taking a build 19/0/52 be more beneficial than a trandition 19/52/0 or 0/18/53 build be better? Efficiency spec (Pros) Less time spent in saronite thus more dps up-time, (cons) less DPS.

19/0/52 is most likely the best spec for hard mode General Vezax. The reason it is used is not so that you spend less time in saronite and thus more dps up-time but because for hard mode you cannot use saronite and hence are limited to 22k mana (or however much mana you have raid buffed) for the whole fight.

I don't think I asked my question properly. I meant to ask, would it be more total damage done in non-hard mode to off spec with that efficient frost spec than to use a traditional spec?

For quite a few people, Frost does more total damage in Vezax hard-mode. Try Frostfire and see if you think you can make it.

In general, Frost and Frostfire both are very efficient specs, with Frostfire being at least equal in terms of DPM. However, on General Vezax, Frostfire loses Master of Elements and 50% Spirit regen from Pyromaniac, while Frost loses nothing of its own talents. (There is the raid-wide loss of Replenishment et al., but that's for everyone and every spec.)

I've done the fight as FFB before and I could not last the whole time without using a saronite cloud. I can go oom in 2-3 crystals at TTW/Fireball, 3-4 as FFB. I should really count how many crystal clouds they use in the video but I'm getting ready to sleep soon. >.<

As I tried to explain to you with my last comment, the hard mode for General Vezax (which you said you saw the video for) means NOT BREAKING ANY saronite crystals until they form into an Animus you have to kill. This means you have no saronite clouds for the hard mode, and hence 0 mana regen for the WHOLE fight.

This is the reason frost is used, frost is the most mana efficient spec and like Roywyn said, all of the frost efficiency talents are active in the General Vezax fight whereas some fire talents arn't. If you watched the video closely you would have seen that some of the mages wen't oom by the end of the fight and were wanding (and that was even as frost), so I do not see how frostfire would last in that fight since it may be more efficient than fire but still nowhere near as efficient as frost.

This, I can not answer 100%, but I am leaning towards not picking up Arcane Mind. The reason being that from what I picked up from the video, there were 4 mages in the raid, all frost. 2 of the mages went oom shortly after the Animus died and the other 2 mages lasted until Vezax's death, barely. I don't see wanding in the last ~10% of Vezax's health as a problem because they would have done a lot of the dps on the Animus when it was needed (which is one of the main parts of the fight).

In my opinion, having another 5% (or slightly more) dps for the Animus would save the healers more mana than Vezax dying a little bit slower in the last ~10% because you are wanding (which might not even happen since 2 of the 4 mages lasted the whole fight without Arcane Mind.

Again, most of what i've said is just speculation on what i've picked up from watching the video so if Kyth or someone else from Fusion ends up dropping by, i'm sure they can correct me, but i'm fairly sure this is accurate.

Sorry. In my original post I failed to include using that spec for NON-hard mode to try and get more total damage. Wondering if it would be advantageous for not having to use any saronite clouds. =P

I don't know if it's different for other mages, but I am far more limited by threat early on and far more limited by finding shadow crashes later on. Trying to cast outside of a shadow crash is usually a waste of mana.

I've been playing around with Rawr for a while now, and at first, when I didn't have Shadow Crash factored in, it suggested 25.0.46 for a build. It's a variation on the Frost build that Kyth and Fusion mages were using; it picks up Arcane Mind, which I mentioned earlier (in replacement of Chilled to the Bone).

However, that was also at a 300 duration fight, and their actual fight is listed as being just under 8 minutes long on WWS, so I set the fight duration to 480 seconds. Then I went in and set the Cost Multiplier to 0.3. The Disable Mana Sources feature has always been enabled though.

After making those changes, Rawr drops points in Arcane completely and instead goes deep Fire and does a straight Frostfire Bolt spam with Hot Streak. It takes Cold Snap over Living Bomb as well. The relative stat values aren't quite as dramatic as they were (without shadow crash), but they're still generally the same:

Intellect: 2.20
Spell Power: 1.56
Crit Rating: 1.12
Haste Rating: 0.83
Spirit: 0.62

Intellect is at 1727 (it was over 2000), and starting mana pool is at 28,893 (it was at 33k). The two trinkets suggested are [Pandora's Plea] and [Flare of the Heavens]. All the gems are [Luminous Monarch Topaz] and [Brilliant Dragon's Eye]. Also, although it does use Pyroblast on Hot Streaks, it does not use T8 4-piece, only 2-piece set bonus.

Is there any discussion on the tier 8 4 piece bonus? I just completed it last night and noticed it has a very low proc rate, to the point that when I first tried it on a training dummy, I emptied two mana bars and didn't see it proc once and thought it was broken. It doesn't seem like a very large DPS increase compared to the tier 7 4 piece. It's just more flaky RNG on top of how unreliable Hot Streak is.

For hardmore vez, I'm leaning towards some sort of 28/0/43 (to 32/0/39) type spec.

TBH, I'm not sure WE is worth using. It seems like the optimal way of using it is to summon as passive right before a crash is coming, let it follow you to the safe zone, then attacking with it while it is out of the crash area. Even doing this, WE is generally not more than 3-4% of your total damage.

I have horrible PvE gear (s5 pieces in some slots lol) but as the 19/0/52 type specs, mana was not a problem and I was not spending every single BF. (I don't have 4pc t8 and I like to try to save BFs for when I need to move for crash; obviously, you spend them immediately if mana isn't looking great) The only situation where mana is a potential problem as these specs is when you are forced to nuke into the animus without shadow crash (which with bad luck happens). However, if you get a drought of crashes during animus, it seems like generally a wipe anyways, so I'm not sure it's worth obsessing over.

You CAN dark rune and it is pretty good -- don't use it early though. =p

It's a bit hard to value BF in the fight, but I bet it's a bit better than TC would indicate. (though one might counterargue that DPS b4 animus doesn't matter)

Doing the fight as FFB seems quite doable, and perhaps as good as these frost specs in the sense that you'll have dmg when it matters. (But probably worse for learning...)

It doesn't seem terribly useful to me to obsess terribly about having a HUGE mana pool especially as some deep frost spec. It's not like you can really convert that mana into damage in a meaningful way.

If you read http://elitistjerks.com/f75/t30655-w...m_3_1_updated/ it states that there is a 20% proc rate without an internal cooldown.

Another big advantage is to have ice barrier against the animus phase. After your shadow protection potion runs out, each barrier will stop another 6 dots or so from being applied. That's a huge help for healing.

What struck me is that Kyth chose to glyph water elemental, I can't understand that. Rawr is saying very clearly that glyphing for and using ice lance (on ghost charge) will be a whole lot better. It's a cheap spell and seems to add theoretically ~100dps over FrBFB.

Have a look at how Fusion did hard mode. They stack all dps casters and disc priests on one area, and all move out and in again for each crash. This way you'll always have a crash available, and everyone knows where to move.

I am not pretty sure about this but when standing in shadowcrashes your casting time is roughly 1.1seconds, depending on how much haste you have. Thats pretty close to the hastecap. This is also pretty much ice lance's global cooldown. At this point spamming frostbolt should be better, shouldn't it?