Melee probably don't get 100%, but warriors hitting more often results in more rage to burn on abilities which generates more yellow damage as well. Rogues also have a talent that grants energy on some successful offhand attacks, which results in being able to spam Sinister Strike more often and therefore more yellow damage.

There might be a model in the warrior/rogue TC threads.

I can't remember whether mana ticks every 1 seconds or two seconds now, however let's assume you have 600 intellect and 400 spirit with raid buffs. By the new formula, you would regenerate 91.385 OOFSR every second, or 182 every 2 seconds. Let's be generous and assume AM->Scorch gets you a full two-second tick. Normally you would get 60% of that value. AM->Scorch let's you reap the full value, which is a net gain of 73 mana each cycle. This is less than the mana you spent dumping into an arcane missiles followed by a scorch instead of 3 frostbolts. Furthermore, both arcane missiles and scorch do a fair bit less dps, with their respective talents, than a 40/0/21 frostbolt. So using AM instead of frostbolt causes you to lose both mana and dps. The only thing you gain is pushback resist, which may or may not help you avoid a cycle interruption which costs 1 second of dps time. In other word you MIGHT cause your 13.3 second cycle to take 14.3 seconds instead (value reduced accordingly with haste).

I guess we'll have to agree to disagree then. In the TC I've done when there's both COS and COE and no winter's chill then ABx3+FrBx3 with 40/0/21 is almost the same as ABAMx3+Sc with 48+/0/11. When there is only COS the AM cycle is better. If there is winter's chill then FrB cycle is better.

A warrior under heroism will already be generating enough rage to use every GCD available to him on damage. The only way its going to increase his dps is if he is executing, and executing is the best use of rage when at low rage and scales worse the higher in rage you go. Same with a rogue. They are pretty much filling every gcd as well. Combat potency under haste is not going to make the number of yellow attacks go up any large degree.

I think we have to consider spellpush back also.

40 arcane / 21 frost is good DPS, but has absolutely zero talents preventing spell pushback. It will do a lot worse for those fights. So, the variability of the DPS of this build will be higher.

>48 arcane where talents are put into AM, means that for such spell pushback fights, you have a nuke you can spam that will be totally unaffected by spell pushback. That's pretty big.

A compromise would be to have Deep arcane/fire. With points into burning soul. So, you can fall back on fireball as a good DPM spell. And its dps is relatively high too, and it has 70% immunity from pushback. Then when mana permit,s just spam Arcane blast the rest of the way.

I always wonder why people say Teron is a good or fair DPS measure when arcane has such disadvantages in that fight because of the frequent spellpushback. In that kind of fight, or say like bloodboil, maybe just spamming AM might be better than using AB.

Vontre, I looked at your mana consumption. And I think you don't theorycraft a fully well played arcane spec while for fire and frost, it is truely nearly at 100% play already.

The key thing lies in mana consumption. A well played arcane mage tries to end the fight at zero mana. Any fight where you end up with tons of mana, half mana, or more means you could have done far more damage but you didn't. When I input my charecter's numbers into your simulator and check mana consumption. For the arcane specs, I alaways end up with lots of mana left.

One arcane spec variant had me standing at over 9000 mana at the end of the fight, another had me at over 6000 mana.

For fire or frost specs its probably no issue, because they cannot convert any excess mana further anyway. But for arcane, it really is a big issue.

I would like to suggest one variant spell rotation that turns everything on its head. For theorycrafting, how about we assume that arcane will intentionally try and AP, AB, and AB spam all the way until it is nearly oom (like left 1000 mana). eat gems, potions, evoc all as early as possible, then just keep ABing until nearly oom. THEN, it just sticks to fireballing or frost bolting, or some combination of AB rotations till the end of the fight.

This ensures that it will be casting throughout the fight, plus it would have used up most of its mana as well.

One of your mana consumption routines for arcane spec had it evocating near the end of the 5 minute fight, resulting in nearly 9k mana left.

Also, about the stacking int part. Its controversy. But you were saying stacking int is less useful because its wasted on short fights. But what's the definition of a short fight to arcane spec? Even with goodly high int like 700 or 800 int, AB spam can only be kept up for at most around 2 minutes. Going by that. We have a long way to go before stacking int causes us to "waste" int because few raid boss fights will be less than 2 minutes.

Please don't get me wrong. I deeply respect what you have done. But I think its a lot harder to simulate a well played arcane mage than it is to simulate a well played fire spec or frost spec because of the ability of arcane mages to convert excess mana into damage.

What's the point of cold snap AP? Icy Veins AP is such a beautiful thing to stack, and keep paired, especially if you AB spam every second for its duration.

We have a chance to test if heroism/bloodlust are considered haste in their effects on the GCD for spell casting yet? I'm operating on the assumption they will not, but I have yet to see anything definitive.

You simulation seems to be consuming mana potions and mana gems at the rate of one of each per minute, and use more than 3 charges of the highest rank of mana gems, which makes the arcane specs able to spam more arcane blasts than they can in reality.

Also, the destro lock never uses mana potions, and uses life tap to stay at full mana. In reality, the warlock would be better off ending as close to 0 mana as possible instead, he would spend mana until hitting 0, and only start life tapping then. The simulation uses 21 life taps on a 5 minute fight with the parameters I used, it should be possible to get by with far less than that.

The entire change would be nearly pointless if they did not.

The reason the simulator leaves excess mana is because it was difficult to predict how much mana I would to have left to spam AB for the full bloodlust duration. However, I left it as is for two reasons. One, arcane spec was already handily beating out fire in the tests, and two I believe it would more accurately represent human play, as trying to balance mana consumption around 45 seconds of ~3500mp5 burn is likely to result in a heavy margin of error. I like the idea of a simple burn it to the ground and spam frostbolt for the rest of the fight idea, actually, because it significantly reduces human error, so I may give that a try tomorrow.

One other spanner to throw in for more optimised arcane DPS routines in a 5 min fight. You can only use arcane power twice. You can only use potions and gems twice. So, the first one near the start of the fight is easy when you have more than enough mana to burn.

But the second gem and potion should not be used once the cooldown is up. This is because full AB spam would leave arcane spec with near zero mana and going into high DPM mode casting(casting frostbolt or fireball) after just 2 minutes. So, the best thing to do, is to intentionally wait until the 3 minute plus mark, when 3 things have all happend.

a) AP cooldown is up.
b) gem cooldown also up
c) potion cooldown also up.

Then, arcane spec can activate everything all together and go into a brief Arcane powered burn mode for another 15 to 20 seconds before going back into high DPM mode once mana has been burned down to a low again. It ensures maximum bang for buck because AP powered Arcane blasts are good.

So, graphs of arcane should see 3 humps instead of two. One hump at the start, powered by arcane power AB spam. Another one sometime after the 3 minute mark powered again by AP. Then one final one at the end, powered by blood lust/haste/drums, etc etc.

There should also be a plataeu. A first 2 minute where average DPS is significantly higher, followed by a drop off to a lower average DPS phase as it goes into high DPM mode. (But still with the three humps I mentioned).

The beauty of such a rountine is that it ensures that arcane spec wold have used all of the mana it had at its disposal as much as it could. It would always end any fight over 3.5 minutes with nearly zero mana. And most of that mana would have been used on AB spam to maximise DPS.

As good as a CD management system that will entice 3 peaks sounds, I am more interested in hearing Kavan in a more detailed explanation of his last post.

You support that ABx3 AMx3 SCx1 is an upgrade from the ABx3 and FBx3 as long as you don’t get a CoE and this of course implicitly states that it will also do much better on pushback fights… Which, overall, also implies it is a better damage spec since the 40/0/21 would at best deal the same amount of damage … I would love to see some numbers on that.

I also don't understand why bother with scorch if you're going to channel 3 missile anyway...
Maybe I misunderstood your post though.

[EDIT] Unless you plan to use AB AM AB AM AB AM scorch as a mana regenerative tool when you're not ramping up AB.

Is there a point? The difference of one less fireball and two more scorches over 300m is so nominal it's rather inconsequential don't you think? It's hardly going to fluctuate more than 5dps. Accuracy is great and all, but no need to go so overboard, particularly when we're looking at two specs of 200+dps difference.

Edit: On the argument of AB*3/Fbolt*3 vs. (AM/AB)*3/Sc I ran some numbers and came up with them costing 78.9mana/sec and 61.3mana/sec given (a) kavan's stats with only self-buff and DivSpi+BoK and (b)no resists on any spells.

As for the DPS difference, YMMV because it's too dependent on too many factors, not least of which is whether you gear int or spell and COE or not. The mana difference however, is non trivial. The latter also ensures you manage to effectively CC-AM while the former simply adds 3%crit.

I have not tried yet to examine the mana drain of 3*AB/3*Fball in 40/18/3 given 3/3 Master of Elements, though irrespectively I expect the FBall rotation to be higher due to imp. Scorch which is most likely to be present from fellow firemages.

Also bear in mind, as has been said before, any non-steady-state AB rotation (ie: only cast one, or chain-cast at max debuff) suffers from that the server delays the incoming confirmation of "you have AB debuff" so if you're successfully casting as fast as possible you will pay for the debuffed spell but cast at the previous debuff speed making for a distinct time loss.

Edit II: Genocidal: The rotation is AB AM AB AM AB AM Sc. The point behind it is (a) maximizing OO5SR, (b) Capitalizing on Clearcasting AM: If it procs off an AM, there's plenty of time to skip the next AB and AM instead, and if it procs off a AB or Sc there's an AM after it anyway. Secondly, as I said above, AB*3 does not in fact give the correct ABs although it costs the correct ones. Careful analysis will in fact show, that due to new casting changes you'll start casting the lower AB by the time you get the confirmation of the faster one so 3*AB isn't in fact as fast as it should be on paper.

Clearly (AB/AM)*3/Sc has a much lower AB-to-other ratio than either AB*3/AM/Sc or AB*3/Fball*3 or Ab*3/Fbolt*3 but whether or not this is as bad as it seems is still under examination.

I just plugged in the gear from 2.4 that is available so far in a spreadsheet just to see the stats and how would a mage be with all the new stuff. The results are Here without taking any 2xT5 bonus or 4xT6 bonus, just for an overview here.

Plugged this in Vontre's simulator and got some really weird stuff... someone has an idea from where that comes?

The simulator tries to stack cooldowns correctly...which will increase your dps.

Look at mana consumption... and also only AM spam used... why does it calculates that AM spam is better than AB spam? even though with AM spam it won't use any mana if a shadowpriest is added with that kind of mp5 while casting...

The thing I wanted to see is how would be ignoring the 2xT5 and 4xT6 for Arcane in 2.4 and if it would be viable having a boost on mp5 and int by about 50% with the new items. But simply at 529 mp5 while casting and a mana pool over 18k AB spam during AP is better dps than AM given that it will get the benefits of haste in 2.4, but in the simulator AB is nowhere to be seen...

During a (AB/AM)x3/Sc rotation, isn't the AB still going to be a 4th stage AB, assuming you are coming down off AB spam? Unless I'm missing something, wouldn't this defeat the whole point of a rotation?

Considering the problems with AB cycles not timing correctly according to some posters, I used the alternative arcane method of blast spamming until you're almost oom and then frostbolting or missiling the rest of the way down. Keep in mind for every comparison I've assumed 1 pushback for every 45 seconds. I made considerable improvements to the simulation logic.

For instance, I discovered that a glitch was causing evocation to have a cooldown of 4 minutes instead of 8. Whoops (I'm surprised none of you caught the two evocates in the mana regen log =p).

Ok, so that changes things considerably and we're looking at a more reasonable picture here. Basically long story short, arcane is a reasonable alternative to fire if you're lacking CoE, and can almost keep up with fire + CoE, however deep fire spec + CoE is still the best possible damage. If you have malediction on your CoS, and CoE normal, it's pretty much a wash between the two.

If you trade hit rating for spell damage successfully such that you are gaining about 100 spell damage in place of superfluous hit rating, AND you are not getting CoE, arcane/arcane is your best bet. Considering you should really be using this build to make up for a lack of CoE anyway, I'd have to say 48/0/11+2 is a good candidate for best arcane spec. However my comparisons weren't terribly in depth, I was basing the stat choices off the item formula rather than actual items. Your results may vary.

Additionally, I noticed trading spell damage for int and as such raising int to 700 increased damage slightly, however further increasing by another 200 intellect caused a sharper decline in overall damage. Looks like there's a sweet spot for intellect/spell damage ratio using this build.

If the updated calculations are correct, and blizz doesn't change them, I think they about hit the money on arcane spec.

I'm glad for things:
1) I was a spriest in tier 5 so I have no tier 5 2pc on Nuru
2) For some reason the Warlocks have been putting up CoE lately over CoR. I can't make sense of it, but the melee haven't said anything.

It's not that I'm against Arcane, it just seems like a significant amount of extra effort for the gains described thus far, even with the 2.4 changes.

No, that's why the Sc is in there. It dropsthe ramped AB.

well with new calculator fully raid buffed with my current gear. this includes coe and cos, ele shammy and sp. We always run with 1 deep ice mage for wc as well.

40/0/21 1919 dps (+1371 mana)
2/48/11 1741 dps (+7172 mana)

I also use serpant coil braid. When I start my dps cycle I eat a mana gem for the damage bonus after I get down just a little, 2 to 3 cast into the fight. I also start with arcane power. I never use icy veins when casting arcane blast beacause of being against the wall of gcd. I know gcd wont be an issue in 2.4 but, my frost bolt just does more damage , especially when I crit, which is alot. I use all my cooldowns when they are up, 5 seconds into a cool down is either wasted dps or mana. Evocation is used based on mana situation, fight progression and how I look on damage meter :P, If i need a boost on dps meters, I go into full blown arcane blast mode.

I been trying to fine tune my rotation last couple of days. I look at my situation all the time and constantly observing my mana pool, cooldowns, boss health and change it all the time, so trying to stick to a rotation is almost impossible. The 40/0/21 spec is probably the most interactive spec I have played.

Hm, I plugged my T6 (currently frost) profile in to see how things work and scale. 300s fight, standard options.

Numbers are CSIVFire spec, and Arcane/Frost (40-0-21)

Baseline: 2166, 2387
+20 int: 2170, 2433; +4, +46
+20 haste: 2190, 2389; +24, +2
+24 dmg: 2189, 2408; +23; +21

I'm very surprised about the excessive scaling of intellect. I remember int beating crit and competing with dmg, but not that far ahead.

What I don't get though is the non-scaling with haste.
Yes, AB spam works by burning mana, which haste has no influence on. But increasing the DPS of the DPM rotation should have a very noticable impact on the overall DPS.
AB spam is inefficient burn. Haste rating is efficient burn.

I don't have the numbers right now, but might check them tomorrow.
From what I roughly remember, frostbolt spam has a DPM of 6-10. Replacing those by AB spam gives a conversion of 1-2, maybe 3 additional damage for every point of mana burnt. The math takes damage/time/mana differences, so it's pretty volatile to errors.

The point is that haste allows some more mana to be burnt with your DPM rotation. Yes, this does leave a bit less mana at the end, but because AB is such a bad mana converter, haste should have some significant impact. More than the 2 DPS listed.
I guess those 2 DPS come from a faster evocation


I know that I don't have the numbers for mana at hand, but I'm at a loss trying to explain those results.