Then there must be a problem with my Mage mana consumption/regeneration models.......

Or I just can't use the same priority list (arcane_blast,max=3/arcane_missiles,barrage=1/arcane_barrage) with Molten Armor....

Or I need to make the AI look at consumption/regen and properly transition into dps/dpm spell priorities....

The Mage module needs some attention for sure..... I'm afraid I've been primarily focused on extending support into the melee classes. Unfortunately. Priest/Druid/Warlock/Hunter are the only classes with dedicated support other than myself.....

What mana regen is the sample mage getting? BoW and Replenishment, I see. What about JoW? Mana Spring? Mana Tide? Evocation? Mana gems?

You might also try (arcane_blast,max=3/arcane_missiles). That's easier on mana than including Barrage.

Heroism is indeed modeled..... but after reading the forums the last few days, I believe I might need to change where it gets triggered.

Thanks.... I'll experiment more with different action priorities..... I'm afraid I just took the top one off Kavan's list....

To be brutally honest, part of the reason I haven't given much attention to sc_mage.cpp is because Rawr.Mage is so darn good. Kavan's formulation is not merely "good enough", I believe it to be more accurate than simulation: The formulation-reduction surrounding Combustion is (in my opinion) out-weighed by the inherent error in RNG simulation.

But..... I promise to do some more homework and see if I can improve the AI.

EDIT: Never answered the regen question: Everything you listed except Mana Spring and Mana Tide. I've included JoW, but I am no longer confident in my 25% proc (no cd) model.

How or why has 18/0/53 Mage DPS dropped almost 500 DPS, despite gear improvements? Is that merely based on the assumption of a 5% loss in global crit from the Winter's Scorch debuff, or is there something else I'm missing?

Didn't we have the same fail predictions for mages in Wotlk? We didn't turn out so bad all said and done. Mages are constantly top or near the top of the damage metres, despite predictions we wouldn't even be close.

Edit: The model has arcane mages with the same amount of haste and crit as frostfire mages, despite the fact arcane mages stack far more haste generally.

I should also point out that HAT rogues are notoriously hard to model. The ideal situation for a HAT rogue is to be in a group with four BM hunters stacking crit (IIRC), and that's pretty unlikely to happen in a "real world" raiding situation, especially not for each of the HAT rogues in the raid. It's a very interesting spec, but the design seems really problematic under these contrived conditions. They probably do need a nerf, but not as desperately as that simulation would suggest.

Sorry for going a bit off-topic, but at least this does serve as another cautionary note about reading too much into simulations, especially before the changes are finalized on the PTR. Chickens, hatching, ad nauseum.

To be fair here, when 3.0 went live the only reason our DPS wasn't terrible was because FFB got sledgehammered into relevancy near the end of beta. The predictions about our DPS being sub-par were entirely accurate up until Empower Fire got reworked to include FFB. It wasn't until 3.0.8 that the developers got their heads out of the sand with Arcane/Fireball specs and made non-FFB DPS work.

FWIW I fully expect the same thing to happen with rogues in 3.1 - bugged HaT aside they've been pretty bad, and so regardless of whether Blizzard intends for it to happen, rogues probably end up back on top for a while. If mage DPS falls off, for whatever reason, the same thing will happen to us.

That said though, it is completely silly to get worked up over DPS numbers when we know we have changes still coming. Until that time arguing over numbers is pointless.

removed.

M

Enthorn's point is that all of those changes have exactly zero effect on mages, so a drop in arcane DPS cannot be explained by the regen changes.

Another regen change:

MMO-Champion BlueTracker - Blessing of Wisdom & Mana Spring Totem change

That's 85 (or 91) mp/5 less in raids unless they double them, but "Additionally, we have been trying to tone down mana regeneration in large groups" implies that they don't want them any stronger than they are now. If they doubled it so there was no regen loss then they wouldn't meet that goal.

Good. It was weird as hell anyway to have bow and totems stack, despite the fact they were giving almost identical amount of mp5. Thing is, it is stupid to specifically require both a pally and a shaman in raids (particularly for 10mans) due to how poorly designed mage mana regen is in general. It sucks a bit on the short term, but for the long term it allows them to more properly balance things.