Maje

Kavan has updated JoW and the new spell costs, however the new version hasn't been released yet. I've compiled the changes, from what it looks like - fully raid buffed you'll have a little less mana overall (for me it's 0.80 mana gems vs 1.20 mana gems on 3 minutes), also JoW is a little less mana than Replenishment. On the plus side non-raid buffed without Replenishment/JoW you'll have more but you'll still be starved.

EDIT: The interesting thing is that with the mana cost reduction due to Master of Elements our average fireball cost got reduced by a bit less than what's listed (in absolute values). Fireball cost was 560 before the change and it became 515, MoE returns .3 of mana times your crit rating (lets assume 50% for now), which gives us an average cost of 476 before the change and 437 after which is 39 mana difference vs. 45 mana without MoE, in fact MoE reduced the mana saving by exactly it's value 0.3 * crit chance :)

 

Thegoodman

Worthless post, wrong thread. Lo Siento.

 

atliens

Is it necessary to cast Slow on bosses in a raid setting to trigger TtW? Or do other classes provide the snare (if so what are they)?

 

Pasture

It's necessary to cast Slow. It's the only snare that sticks on a boss to activate TtW. But casting Slow every 15 seconds in essence negates the dps increase so don't bother with TtW for Raids.

 

Vees

I'm really new to theorycrafting and mages, but I recently hit 70 with my mage and wanted to figure out the best DPS spec and rotation (with less than ideal equipment). What I found was that the Frost Mage, with a pet, seems to do better DPS than just about any other class. Since, I'm really new at this, I'm not sure I'm running simcraft peoperly, but my tests show that frost, with a pet, does much better than any other spec:specs):

Two minute fight
1 Mana Pot at -2.5K
1 Evocate at at -5K
1 Mana Gem at -2.5K
gear_stamina=219
gear_intellect=214
gear_spirit=70
gear_spell_power=420
gear_crit_rating=202
gear_hit_rating=30
gear_haste_rating=0
gear_mp5=21

Yields:

DPS Ranking:
3553 100.0% Raid
815 22.7% Mage_0_9_52 (Frost - Use Pet)
731 20.5% Mage_0_53_8 (Fire with Living Bomb)
717 20.0% Mage_0_50_11 (Fire with icy_veins)
686 19.2% Mage_51_10_0_Slow
634 17.7% Mage_0_9_52 (Frost - No Pet)

I've seen other simcraft run with much different result at higher gear levels. For my gear level though, frost seems to be the way to go.

 

Roywyn

That would be the issue. 420 spell power is rather mediocre at level 60, let alone level 70.
Frostbolts have good base damage but mediocre scaling, and especially the Water Elemental has rather bad scaling of your gear.
Which means that Frost is ahead when you're naked and fire will take over when you get more gear.
Fire also is incredibly reliant on a high crit chance, mostly from raid buffs.

Compare it to a prot spec with 700 Attack Power - it's probably the highest DPS at that level with green weapons due to high base damage on abilities.

The Blue Bar and you - the complete Fire Mage 2.4 mana compendium: http://elitistjerks.com/658230-post3191.html

DPS spec and class comparison in Naxxramas gear: http://code.google.com/p/simulationc...i/SampleOutput

And [Timbal's Focusing Crystal] doesn't proc on AM.
Neither does [The Egg of Mortal Essence] since 3.1.

 

Doroteasenjk

Yes, the burst nature of Frost is demonstrated in that length of fight. You have all of your cool down trinkets available, all of your Water Elemental damage, Icy Veins, followed by Cold Snap and repeat.

We had a fun Molten Core run 10 days ago. The top DPS was a frost mage who pulled close to 6k DPS on Ragnaros.

Real raid fights are substantially longer, with multiple phases, and lots of mobility. They don't usually let you stand and nuke for 2 minutes.

 

diag

I found that I couldnt get a dps close to what Rawr predicts, the best I could do is 500-700 dps lower than Rawr. Is that common to all mages?

 

Kavan

There are several things to check. First you have to take into account how your damage meter counts dps. Some will start counting time as soon as anyone in the raid engages in combat so in that case you would be missing some time. Also most fights require some amount of movement, to properly take this into account you should set the dps time option in Rawr. Last thing that will influence your dps is latency. Rawr uses 0.05 as default which might be a bit low for some players. Take a reasonably large combat log sample and calculate your average latency and use that to adjust the latency option in Rawr.

 

• Seonid

It is common to all simulations as it is just that, a simulation, not a prediction.
It also matters very much if the settings you fed into Rawr accurately match the actual encounter that you simulated.

 

solbergb

I find the absolute numbers in the simulation tools of less value than the relative numbers.

Ie, if I make this change, do the numbes in the simulation go up or down?

My actual numbers will vary based on my execution on a given day, fight dynamics that are often hard to simulate and for shorter fights, the number of crits I get vs my theoretical crit rate. It is fairly typical for mages to pick simple "tank and spank" fights with decent duration to compare to the theorycrafted numbers, because short fights or complicated fights tend to deviate from ideal performance more.

That said, you can look at the rotation predicted by the tool and compare it to what you actually did. A scorch/fireball/hot streak/living bomb rotation requires you to scorch enough, but not too much, to keep living bomb up as much as possible and to cast pyroblast every time hot streak procs. If the recording of the fight on WWS or some other tool shows hot streak auras without pyroblasts, or too many scorches per fireball cast, or not enough living bomb uptime then the problem is player skill related and not spec/equipment related.

 

eltigeron

That's all fine and dandy and very true. Skill plays an important role on how one's damage one day is gonna be.

However, regardless of skill, if you can afford the changes to your gear to promote the ideal numbers from a simulation which, in theory, says is the optimal DPS you can do given the conditions for a fight, than your optimal DPS can only go down from there, again in theory.

What theorycrafting and gear calculating does for people is set the limits as high as possible with the assumptions of no failure of one's rotation at all times, popping cooldowns at the most optimal times.

Now I don't encourage a person to live strictly by the theorycrafting because theorycrafting is just that. It won't always predict correctly what may go on considering the dynamics for a fight but it gives you a range or an idea on how or what role a change in gear does to you as a caster. Theorycrafting also will not define how one plays a mage and if a person lacks certain skills that the theorycrafting expects you have from the get go, and I'm talking about basics here, than it won't benefit you as much.

In that sense, it's one's responsibility to increase their skill as a mage. That's the first and foremost before worrying how one's gear or an item gemmed differently is going to improve/harm your play.

 

Doroteasenjk

Once upon a time, I used to design and implement data communications software and systems. The company I worked for competed with the likes of Cisco and AT&T a decade ago, and produced very competitive products in some very narrow market niches. Part of my job was to build analytical models of performance.

Building good models starts with some having good conceptual models of how data flows through a system, especially with crossing streams, and competition for scarce resources (CPU cycles, UARTs, bus cycles, etc.). The windows of opportunity of these scarce resources were often measured on scales ranging from fractions of microseconds through to seconds. Once a conceptual model was built, it could be used to build one or more practical models, similar to what we call theorycrafting. The practical models were then instantiated or implemented or measured to investigate specific cases. I built those practical model implementations in a combination of MS Excel, Perl code, and paper napkins -- LISP, SNOBOL, and even PROLOG were also used for some very specialized analyses. I'm glad to see more unified implementations in the RAWR & Magegraf & TCOM systems; they are much more useful than the collection of ad-hoc-ery that I had back in the day.

Anyway, back to my point and this thread. One area of analytical modeling that was required in my datacomm analysis stuff was how perturbations in timing affected data throughput. The analogy for mage theorycrafting is how interference with rotations affect DPS/DPM/total damage. From my simulation reading, this meant that you had to ask "how susceptible is this parameter to perturbations?"

From a paper napkin perspective, the haste rating input parameter is most susceptible to perturbations, especially those introduced by larger variations in lag; if you miss a window by being too fast ("Spell not ready"), or too slow (gaps in your spell casting) then haste rating to produce DPS gets wasted. Similar perturbations of mobility, push back, mana return, etc, have impacts on the other parameters as well, but I believe that haste is the most susceptible, hence sensitive, to uncontrollable environmental parameters.

The various modeling tools attempt to look at those environmental perturbations, and seem to do a good job. However, reporting of those results is lacking. Any tool that shows a result line is misleading to the casual eye; the result line is really a result smear and should have substantial size error bars on it. Now, those error bars cannot extend above the line because the line represents the maximum achievable result, so the candles should really hang below the line. The height (or depth) of the candle shows what you can expect in practice with a normal amount of lag, or movement or other environmental input.

 

Doroteasenjk

To complete my thought: The whole topic of Sensitivity Analysis can be very useful for helping with understanding the outputs of the simulation programs, and how to compare WWS or Recount to what RAWR tells you, and how to modify your play style and gearing. In one sense, the stats (input parameters) with the most sensitivity or are most perturbed by sensitivity are what you can control. Here is my reasoning.

Lag: Quartz is very good for allowing you to better time your spell casting. Good use of Quartz will reduce the impact of these sensitivities in practice, allowing you to reduce the size of the error bars. You need to do two things to improve your performance: find out how far into the red you need to delay to start your cast without "Spell not ready", and how to read combat logs to determine gaps so you can adjust how aggressively to pursue the red zone. I have seen no tools that help you with either one, but they seem to be the factors that distinguish excellent players from not-so-excellent players.

Lag Variability: Quartz works best when the lag is stable. If you are consistently at 145 ms lag, let's say, then Quartz gives you a very reliable red zone, and you can hit the start cast very well. However, fluctuating lags (varying by several hundred ms) aren't well handled by Quartz as predicting future lag from past lag can only go so far.

Quartz also does not show a red zone for instant casts; you end up trying to remember what the last cast bar looked like.

Mobility: Mobile fights really require you to "maximize your non-movement." Consider the Alar fight: I watch mages set up to nuke Alar on a particular platform and then run to the next platform on each shift. A better strategy is to position yourself between platforms so you can chain nuke for longer; a typical good rotation will have one less movement than a typical "bad" rotation.

Mobility and Situational Awareness: I don't just mean your own situational awareness; I mean that of your entire raid group. If you have people who consistently run up and stand on your toon because they aren't looking at the screen or they don't care or they don't know, then you will have to move more often to avoid collateral damage from air burst or burn or other tainting debuffs. If your raid group is very good, then you will get many more periods of stable stand-and-nuke; the non-elite of the raiding community will have to live with more frequent moves of a couple of yards because your neighbor cannot be bothered to watch his positioning.

Rotation Windows: Think of rotation windows as clockwork gears. The design of mage mechanics is like a piece of clock work with gears of different sizes and teeth spacing, all connected together with you cranking on one end and the boss dying on the other end. Some gears have large gaps between teeth (Living Bomb cool down) and some have teeth that appear randomly (Hot Streak) and engage on an unpredictable rate at a micro scale and a more predictable rate at a macro scale.

You can control your rotation windows by using Quartz debuff bars (assuming they ever get fixed), debuff indicators on your action bars (Dominos shows whether LB is still up), and watching your buffs/debuffs assiduously. The current state of add-ons that provide a true dashboard is pathetic and provides a good opportunity for a new type of add-on.

Value of Haste: Haste is a sawtooth function with respect to the rotation windows of any boss fight. However, instead of being a simple ramp-up on each inter-movement phase, there is a flat area followed by a ramp-up, followed by a drop back to the flat area on each movement or on each missed gear rotation. You can certainly map the plateau-plus-ramp as a flatter ramp, but I think that hides the reality, which is that there is no benefit to haste from some reductionist perspective. Only on a larger scale does haste really add to your total damage, and then only if you can control the lag, mobility, situational awareness, and rotation windows.

After writing up this and the previous posting, I've convinced myself that haste is vastly overrated for most mages, in most raids, for most bosses, with most computer and network setups most of the time. More importantly, haste actually has a quantized contribution to total damage delivered. The first quantum step allows you to transcend the ordinary game round trip delay. The next quantum step allows you to transcend the variability in communication speed (lag, latency). The next quantum step might allow you to transcend the plateau introduced by frequent moves when your movement is optimal. The next quantum step might allow you to transcend the sub-optimal movement patterns. I don't know how large those quantum steps are, or how they vary with lag, latency, mobility and windows, but I recommend some accomplished theorycrafters investigate this further.

The simulators don't show the bands and the band gaps, and I think that this confuses most people (including me) about what to target or allow for gearing purposes.

 

Duravi

I would argue that haste isn't "overrated", the models do weigh it correctly, but what you are saying is also true, due to communication realities it is truncated at certain step sizes and if you don't make it to the next step you lose some of the value, in reality I would guess this only adds up to a couple points of haste at worst tho. As far as round trip delay; you are queing your next cast to the server before your current cast (at its haste adjusted cast time) is even complete, unless that delay is longer than the time it takes your current spell to finish where is the lost time? One other reason it's not a smooth value is c/d stacking, if 1 more haste allowed you to get one extra fireball cast off in your 20s c/d stacking duration the dps value of x compared to x+1 haste obviously jumps up a step.

 

Doroteasenjk

The delays come in a number of flavors. Quartz attempts to predict the lag of this cast based on recent history. If your lag changes during the cast, then you can fire too early, and your game tells you the spell isn't ready, or too late, and you have missed an opportunity. The former is a little more expensive as it may take a round trip to allow your game client to unlock the spell again.

More importantly, you get different cool downs running asynchronously, since some are affected by haste and some aren't. Fireblast cool down is not affected by haste, but cast time for Fireball is. If you don't have enough points in Improved Fireblast, you will be waiting a noticeable amount of time to cast Fireblast, or skip it this round completely. Similarly, Living Bomb has a 12 second effective cool down (since casting it earlier wastes mana and time); if you have LB as part of your rotation, there is a level of haste rating such that adding 1 to your haste actually causes you to miss the window for casting a new LB, or waiting a period of time, not casting.

Similarly, because of the many fights with movement required, such that the fight is really broken up into a bunch of 20 second, let's say, fights, you will find that you have to have huge amounts of haste before your delivered damage actually increases. Now, the use of Fireblast or Arcane Barrage allow you to smooth things over so that you can cast during the move, but the movement time has to be <= the GCD to allow you to continue to chain cast.

You can test this if your simulator can be set up to require you to "move" for 2 seconds every 20 seconds or 40 seconds or 60 seconds. You will find that haste rating has a very pronounced step function, with no benefit at all for smaller values at shorter times. You need a minimum of 5% at 30 seconds between moves to make a damage difference, and 2.5% at 60 seconds.

 

vishal

Sorry if this has been answered anywhere but could someone please clarify if boomkin crit aura stacks with the buff that arcane focus gives or not.

thank you in advance.

 

dagee

I am completely confused about something that I hope you guys can help me out with. There is another mage in my guild (we are both frost mages 61 points into the tree). Gear wise we are very similar. He may out gear me by a little but if so it should be failry insignificant. What bothers me is that he out dps's me by a lot but for the life of me I cannot figure out why. On paper I think I have the better stats. When it comes to an actual fight though we are not close. I plugged both of our information in RAWR and according to it we should both be doing about the same damage (~1230 dps self buffed) which again confuses me since he is way under the hit cap which according to everything I have read should make a huge difference.

Here are our stats:
Mine
Hit: 178
Dmg: 1061
crit: 14.4
haste: 158
int: 438
spirit: 150

His
Hit: 104
dmg: 1044
crit: 19.2
haste: 93
int: 514
spirit: 206

While I have not yet started tracking our fights it looks to me that we both pretty much do the same thing: spam frost bolt and firebolt when the instant cast proc goes off. Both of use our pets as soon as they are off cooldown and for all boss fights get to use them twice. We are in the same group and get the same buffs during raids.

Am I wrong and thinking that I should be putting out more dps based on my stats. Does frost not need a lot of hit? Has hit changed so that it is less important and I should change my focus on crit?

I am including a link to my armory profile but not his since it is probably rude to post his without his permission: The World of Warcraft Armory

 

Kandir

Probably your friend is just playing more aggresively than you, and/or making smarter use of his cooldowns and such. Next time you play together, check out a dps measurement tool like Recount or WWS to get some statistics. How many frostbolts is he getting in the air? How often is he summoning his pet? How many times does he pop his trinket? Those are typical factors which set people apart, much more than gear actually.

 

blumpster

@dagee

If I had to guess it is cooldown management. He is more than likely stacking his better during the Bloodlust. Keys are getting your pet out before it, and using double icy veins / trinkets within the Bloodlust buff. Also if you want to do more damage ignore Brain Freeze, it is not an increase in DPS unless you take Ignite and sneak it into the third phantom charge of FoF. Yes your stats are better, and yes you still need hit capped. You are golden, 177 is hit cap for 3/3 elemental precision.

 

Maje

How did you come to the conclusion that Brain Freeze is a dps loss on regular conditions? Rawr at least for my PvE gear says it's a clear gain and that's not taking into acount that it's a free damage (0 mana) at the cost of a GCD, I don't see how Fireball with a cast time of GCD is a loss... I doubt the chance to proc FoF at the 1.5s (-) it takes to fire fireball can override it (this is without any fire talents).

And regarding Hit Cap, if you're raiding with a shadow priest or a moonkin you have 3% hit + 3% from EP means you need 11% which is around 138 rating, so if you have the buff change the trinket to something else.

 

dyelynn

as arcane focus is an individual buff and not a raid or group wide buff, yes, they stack.

 

dyelynn

On my dummy tests BF was about a 25dps loss. That's subject to making smart use of the spell, meaning, using the proc with FoF up for the increased crit chance, ect. With Frost spec, the fireball crits for 150% without ignite. The fireball also has no damage increasing talents to support it, unlike frostbolt. Obviously i'm offering anecdotal evidence and opinion, not RAWR math or wws reports, but simply put, the fact that it's free to cast doesn't really help me, because I've no mana problems on my mage with SCB, mana gems and WE... and the GCD it costs isn't quite just a GCD because chain casting frostbolt somewhat reduces latency to 0 and the only benefit the fireball would have would be in movement fights.

With 61 talents and a hugely bloated frost tree, I just don't think there are enough points to support BF as a dps increase worth picking up. Of course we're talking lvl 70 raiding spec because for pvp or lvl 80 specs the game changes entirely.

 

asys

Pre-3.0.2 my fire mage was quite high on the overall raid DPS but after the 3.0 patch things seems to be stalled. I've played around with my gear and spec quite alot, and looking at my dps on for instance Dr.Boom I cannot seem to feel else that my dps has been lowered.

I've used the following talent builds:
Talent Calculator - World of Warcraft
Talent Calculator - World of Warcraft

Armory link: The World of Warcraft Armory

My hit rating in raid would generally be around 165-177 with Draenei and SPs available in raid for extra hit.

My dps in a 5man last night was ~900 where as it used to be 1100+, and in raids I used to push 1500-1700+ dps on all out bosses like lurker.

While playing with rawr 2.0.3 it seems that it's reading the stats of the PVP neck wrong, and that it wants to enchant my offhand?! And also it seems to want to use crit gems instead of haste pushing crit to 44,91%, hit to 14,07% and 104 haste rating?

Has the mage dynamics really changed this much to make crit > haste > spelldmg? Using pure +crit gems vs +spell haste/dmg. Is RAWR wrong in it's assumption? I can easily say that I dont even do near the supposed 2605 dps RAWR thinks I should be doing.. 1450 today on Dr Boom.

Rotation usually used would be
2x Scorch (glyphed)
if Hotstreaks then pyroblast else fire ball, repeat until scorch is needed.

When reading the forums over the last few years the mage threads been quite nice and ones conclusions after reading usually managed to up my raid groups dps quite a few levels. Now, with 3.0.3 things are very scattered and it's hard finding the right spec's, gem's and enchants. Does anyone else feel the same? Have anyone found a good level 70 raiding spec that pre-SWP pushes a high dps yet? At 70 now I do see a lot of +hit 3000 (300 resist) from bosses in BT, even tho I am hit cap'd. And while reading around I cannot see a good reason for this either.

I know WOTLK is just hours away, but I would still like to understand where things changed so drastically at 70 that it would seem nerf our dps. I know I was very haste hungry and that this dont seem to be a good thing anymore, but with 132 haste, 171 spell hit and 230 spell crit rating I would think my dps should have been somewhat higher.

Anyone who can point in the right directions?

 

solbergb

in most of the current models for fire it is more like spellpower > haste ~= crit but it depends on exact gear levels and weightings. Prior to the patch it was more like spellpower ~= haste > crit. You may be overgeared for haste.

One thing that might be causing your performance issues vs theoretical performance is that the current fire rotation requires riding the global cooldown more, instead of chaincasting spells with cast times. In the latter case, you can usually reduce latency to pretty much nothing and get full benefit from haste. On instants though, knowing when to start after the GCD is up is pretty much an art....you'll lose much of the benefit of haste on your pyroblast procs if you can't start up the next fireball precisely 1.24 (or whatever) seconds after it's launched. Still most of your rotation is 3 second cast time spells, so that shouldn't be a massive decrease in DPS. Just a mild one.

Neither of the builds shown is going to be very good at trash, especially in the post 3.0 "we aoe all trash" situations, so I'm assuming you are focused on boss killing dps and not on overall raid dps when making this complaint.

Perhaps you're a victim of the RNG effect caused by slow cast spells combined with short duration fights. Have you checked your WWS stats to see the following.

1. Does the actual crit rate match your theoretical crit rate (if lower, you were just unlucky)
2. Did you cast as many pyroblasts as you received hot streak buffs? (should be identical...not lower, not higher)