Until recently, anyway. While everything I could find in English says "MA > FA most of the time", the Chinese-speaking NGA forums seem to say "FA > MA all the time", and even go as far as to talk about 14242 haste caps like Frost spec. Arcane Mages on top Asian WoL parses seem to be using Frost Armor as well. This directly contradicts the Western mage community. What gives?
The SimC sample profile has been using Frost Armor, mostly because it's pure single target. The margin is about 1% (~5K DPS). I decided to mess around for a bit, switching between Frost Armor and Mage Armor, varying between 1 to 3 targets. To my surprise, Frost Armor seems to beat out Mage Armor in every situation, and the gap grew on multiple targets. I didn't go further and try 4+ targets, because I don't think the AOE APL has been optimized well for those situations.
It seemed like a good opportunity to look a little further and see what Frost Armor and Mage Armor does for us, so let's do some simplified analysis. Frost Armor gives 7% spell haste. This is multiplicative, similar to the raid wide spell haste buff, and not in the form of 7% rating (ie. 2975 rating). To convert the value of FA into haste rating, we need to take into account the current haste rating of the mage:
Code: Select all
( ( 1 + 7% frost armor ) * ( 1 + gear haste % ) - (1 + gear haste % ) ) * 42500 rating per 100 percent haste = 2975 + ( 7% + gear haste rating )Code: Select all
( 3000 mastery from AM * (1 + PBoI amp %) )1. 3843 / 3291 ~= 1.168. 2T16H setups should use MA over FA only if mastery is at least 16.8% stronger than haste.
2. 3955 / 3291 ~= 1.214. 4T16H setups should use MA over FA only if mastery is at least 21.4% stronger than haste.
How big is this gap between mastery and haste? AMR can provide some context. AMR default weights for my arcane mage assigns 3.7 to haste. If MA were to be worth it, mastery would need to have a weight above 4.32 for 2T16H, and above 4.49 for 4T16H. That's much higher than the default 3.9. SimC give similar results - mastery is stronger than haste, but not quite 20% stronger:
That's a 5.92% margin - not enough for MA to be stronger. This gap also increases if you have higher amounts of haste on your gear due to unlucky drops.
But how about multitarget? The common belief seems to be that "Mastery scales Mage Bomb damage, Haste without meeting breakpoints does not". However, there are two problems with this logic:
1. Between breakpoints, bomb duration is shortened, which means you can apply it more often and get more overall ticks in a fight. Time spent casting high damage Mage Bombs is better than time spent casting filler ABs. The Increased number of bomb casts also means increased AM procs.
2. For NT, haste scales powerfully because you need 14242 haste to reach the 1s GCD. FA also guarantees a few extra ticks of NT because of the frequency of BPs.
3. AE and ABar gain a lot of DPET from getting closer to 1s GCD.
Just looking at the above, I don't feel like there's a very convincing case for why Mage Armor should be stronger than Frost Armor when more targets are added.
But what about the opener during Bloodlust? As you might expect, SimC says that the opener burst is a tiny bit weaker (~1%) with Frost Armor. The GCD capping during Bloodlust and Meta Gem stack durations hurts a tiny bit, but evens out over the duration when you're not GCD capped. And similar to the results from the Frost GCD analysis (http://altered-time.com/forum/viewtopic.php?f=5&t=291" target="_blank), you won't GCD cap on AB/AM with only Bloodlust or Meta Gem, unless you're above 15697 haste - a haste cap you're unlikely to hit unless you're using Kafa Beans.
Should we consider using Glyph of Armors and in-combat armor swapping? I did some hacking around on SimC to make this work, since the Glyph isn't implemented properly. I didn't do extensive testing, but it looks like the answer is "probably not". The damage gained from a mage armor opener seemed to break even at best with the GCD wasted on swapping.
Another way to consider this is through some napkin math: At our current raid buffed ~110% mastery levels, another 11% added by Mage Armor gives us a ( (1 + 110% + 11%) / (1 + 110%) - 1 ) = 5.23% damage increase overall, attenuated by our ability to stay near 100% mana. When you compare this to the 7% multiplicative haste with minor losses to GCD capping, it does makes sense that Frost Armor could come out ahead.
How does it work in actual raids? I ran SoO the last 2 weeks with Frost Armor, and it seemed to work pretty well. The faster pace does make the rotation and mana levels a little harder to manage, but the increased mobility is nice. With only 2 weeks of attempts and all the RNG, it's hard to tell if it gave me any actual DPS increase, though. I'd like to invite fellow Arcane Mages to try it out - running all 14/14H with Frost Armor instead of Mage Armor.
With all the math done, it doesn't look too good for Mage Armor. It also only gets worse with more haste from gear. The conclusion seems to be that Frost Armor is always better, because of the multiplicative effect.
Were we wrong in saying MA > FA except on pure single target, or are we missing something?
EDIT July 11th: Apparently I linked SimC scaling charts from when Frost Armor was already used.
