In the original post:
I'd imagine faster FrB will make it a loss sooner, seeing as how .2 seconds faster FrB is ~8.5% hast, but only for that spell individually.

The mana efficiency buff is only for pvp (we can't afford mana efficiency talent in pvp builds anyway). With the Resilience change matchs will last longuer. Mana is also currently a major issue if you are playing 2v2 with a healer (Hell I'm even playing arcane + 2P T3 for 1 minute cd evocate in 2v2...).

Overall very good changes for pvp, not sure if it will be a significiant boost for frost pve dps.

Oh, yes, but they knew before that patch came out that it wasn't going to work. The math went up almost instantly. There was a GC post about that somewhere early in the process, but I forget where it was.



Assuming I did the right thing in simcraft's mage.cpp file (I do not know the language, but the bit I tweaked seemed reasonably straightforward), the result is a DPS increase of approximately 5.3% (increased from 5926 with the original file to 6240 with my change). That's quite significant, but is of course nowhere near enough to bring Frost Mages back into progression raiding, apart from very occasional encounters designed to their strengths.

This also improves Frost's scaling with crit rating slightly, from 0.49 (about 48% less than the mean of the four major specs) to 0.53 (about 44% less than the mean).

As for the mana change, there are three possibilities:

1) This is just the first stage in a general decrease to the mana cost of all Mage spells.

2) The Empowered Frostbolt change increases Frost's burn rate by about 8%. Frost also could not receive compensation for the Replenishment nerf through a regen buff the way most other DPS caster specs did. These two facts may make an efficiency improvement useful after all.

3) It's purely a PvP change.


ETA: I forgot to reduce the mana costs in mage.cpp. Rerunning now, will update with corrected results.

ETA2: Mana cost changes made no difference in the results.

Also, the guy who posted after me seems to have removed his post, but he's right -- according to Blizzard's patch notes, just updated, the mortal strike effect is attached to Permafrost, not Empowered Frostbolt. This makes more sense, as Permafrost is actually accessible to other specs.

The absence of Blizzard in the list of mana cost reductions tells me that these changes were aimed squarely at PvP. You'd have a pretty difficult time making the case that Frost needed more PvE efficiency (relative to fire specs) in the first place, but not including Blizzard alongside a CoC mana cost reduction reeks of PvP-oriented changes to me.

As a minor point of interest -- including all coefficient-increasing talents and cast-time-reduction talents, the coefficient per second for Fireball, FFB and Frostbolt now comes to:

Fireball: (3.5 / 3.5 + 0.15) / 3 = 0.3833
FFB: (3.0 / 3.5 + 0.15) / 3 = 0.3357
Frostbolt: (3.0 / 3.5 * 0.95 + 0.1) / 2.3 = 0.3975

I again tweaked simcraft to remove the 5% coefficient penalty from Frostbolt. This pushes Frost within about 14% of Fire. If they were to remove the coefficient penalty, and find another 3-4% damage somewhere, Frost performance would be in a reasonable place; then it's just a question of improving playstyle a bit.

ETA: Further tweaking to the sc_mage.cpp file shows that, with the Empowered Frostbolt change, even using Ice Lance on a ghost charge is a (miniscule) DPS loss (using the T8 data file).

I'm waiting for someone to double-check this for me, but I believe that removing the 5% coefficient penalty and replacing the 3% personal crit on Winter's Chill with either +6% damage or +30% crit damage bonus (225.35% crits with CSD) would plant Frost right in the Fire minus 9-10% area.

ETA again: Here are the results from Simcraft. I removed Ice Lance from the rotation (which added about 5 DPS). "3.2 DPS" is current PTR w/Emp. Frostbolt change. "3/3.5 Coef" is that configuration, plus the removal of the 5% coefficient penalty. The next two columns start with the config for "3/3.5 Coef" and then add, respectively, 30% crit damage bonus or 6% damage. The results are pleasing to me.

The changes seem PvP-driven and will probably more than compensate for the resilience change. They could even be trying to compensate for the lack of a haste PvP set for mages, but then they would be missing the point that haste is both an offensive and defensive stat in PvP, because it reduces the GCD of blink, ice block, ice barrier, frost nova, spellsteal, etc and not just the cast time of frostbolt.

Let's hope that there's a bit more in the pipeline. If the changes really are purely PvP-founded, this might be all we get.

From a PvE point of view, the cast time reduction on frostbolt further devalues brain freeze as a talent. Either PvE frost mages are no longer assumed to spec BF or maybe there's still hope of getting extra fireball scaling as an additional part of the talent. (My suggestion of making BF fireballs scale with spirit still sounds reasonable.)

Running the new numbers on my frost DPS simulator using numbers for my gear level indicates that change is a 5.5% DPS increase for pure frostbolt spam.

Brain freeze value is difficult to determine, because the time effect of a GCD in a rotation isn't clear, but assuming there's a 100 ms penalty on each brain freeze, BF is currently a 2.2% DPS increase and devalues down to a 2% DPS increase. Calculating with 200 ms penalty, BF goes from a 1.7% DPS increase down to a 1.2% DPS increase. Mages claiming that BF is currently DPS neutral for them might find that BF is actually a DPS loss in 3.2. I guess YMMV.

Because the simulator ignores the water elemental and assuming the water elemental will not scale any better, the changes in the PTR result in a roughly 5% overall DPS increase for a raid-specced frost mage. Since the changed talent is a casting time reduction, a laggy server or a truly bad internet connection will of course distort these numbers.

Played around more with the SimCraft files today. Currently, SimCraft uses this 18/0/53 build for max DPS, using Brain Freeze fireballs, and casting Ice Lance on ghost charges.

In my experimentation, I first made the changes to Emp FB and mana costs, leaving the spec and rotation unchanged. Then I removed Ice Lance from the rotation. Then I removed Brain Freeze and Spell Impact from the spec, and maxed Student of the Mind (result: 19/0/50, 2 points leftover). Each change produced a small increase in DPS.

5922: Default
6241: 3.2 Empowered Frostbolt change
6244: Ice Lance removed from rotation
6270: Spec change to remove Brain Freeze and max SotM

I expected the change to "eliminate" ghost charge lances by making them a DPS loss; I was more surprised to find that trading Brain Freeze for SotM was an improvement. The change has really firmed up Frostbolt spam as the only way to play, and made it that much more difficult for Blizzard to achieve their desired PvE Ice Lance use.

ETA: This (Talent Calculator - World of Warcraft) appears to be the highest (stationary) DPS spec for Frost as of the current 3.2 changes (assuming simcraft accuracy). Three points leftover to take Brain Freeze for range, mobile damage, or Ice Barrier and something else for utility/survivability.

Lhivera's changes to sc_mage.cpp are in the most recent version of SimulationCraft r2808.
To access 3.2 changes, you need to set patch=3.2.0

After the frostbolt change: Another thing worth considering is whether the dpm loss from the crit chance reduction will make frost less attractive as a GV hard spec, not to mention being pushed under the GCD cap.

It looks like the mana reduction improvments will outweigh the small changes to crit. Even for stright Frostbolt spam it looks like basically 8% spell haste, but a cost of 12% less mana. Even with the loss of 4% crit - it should be a small DPM increase if not perfectly neutral.

Lhivera what glyphs are you using in your test? I was a long time follower of the last "raiding as frost" thread I don't recall what the consensus was on what glyph to sacrifce for the ice lance glyph. Was it molten armor?

Bestin, I seriously doubt Lhivera would sacrifice the molten armor glyph since in spirit-heavy Ulduar gear it's already 2-3% of your damage, and especially since he mentions SotM being a dps upgrade over brain freeze. So in testing ice lance rotations you would want frostbolt, molten armor and ice lance glyphs.

I'm not sure which other glyph you are thinking about, but if you're thinking of glyph of water elemental, remember that he (she?) only does 8% of your total dps. The uptime is also dependent on the fight length. The glyph may allow you to summon one or two extra times, or not. Even on a 15 minute YS fight, ignoring cold snap you'd have standard 5 water elementals, maybe 7 with cold as ice, and maybe 8 with cold as ice and glyph. So what with the movement on that fight and long duration, the water elemental glyph is still going to be hard-pushed to match the increase from molten armor glyph.

However, since Lhivera says we're certainly dropping ice lance in 3.2, it would be the next choice. So as Jak says above, the glyphs for a non-brain freeze, non-ice lance rotation (er, so that leaves frostbolt-spam) are: frostbolt, molten armor and water elemental.

I actually completely forgot to fiddle with the glyphs; all the previous tests were run with Fostbolt, Molten and Elemental. I just reran the test with just the existing PTR changes, using the Ice Lance glyph; the result was an increase over using the Water Elemental glyph and Frostbolt spam of about 0.29%.

Edit: Except I still need to test it when points have been shifted out of Spell Impact; 1 sec.

Edit again: Hm, I can't seem to convince simcraft to cast Ice Lance now. I'll try to investigate further later this weekend. It would probably be good if someone could try similar tests with Rawr (as a Mac user, that's not an option for me) and compare the results.

Icy Floes is a pretty dubious talent as well. The only DPS increase it gives us raiding frost mages is the chance to get in another Icy Veins - but this is very dependant on fight length.

With 3/3 IF the cooldown on IV is is 144 (2:24) seconds - without it the cooldown in 180 (3:00) seconds.

In a case where we're trying to cast as many Icy Veins as possible (ignoring the Water Elemental), we'd start the fight with IV (0:00) and then cast cold snap as soon as it ended and cast IV again (0:20). We'd then cast IV as soon as it came up off cooldown.

with 3/3 IF

0:00 - Fight starts - cast IV #1
0:20 - IV over, cast Cold Snap
0:21 - Cast IV #2
2:45 - Cast IV #3
5:09 - Cast IV #4
7:33 - Cast IV #5
8:20 - Cold Snap cooldown over - recast CS
8:21 - Cast IV #6
10:45 - Cast IV #7
13:09 - Cast IV #8
... continue to cast every 2:24

with 0/3 IF
0:00 - Fight starts - cast IV #1
0:20 - IV over, cast Cold Snap
0:21 - Cast IV #2
3:21 - Cast IV #3
6:21 - Cast IV #4
8:20 - Cold Snap cooldown over - recast CS
8:21 - Cast IV #5
11:21 - Cast IV #6
14:21 - Cast IV #7
...continue to cast every 3:00

So when we ignore the WE - if the fight length is between 2:45 and 3:21 then 3/3 IF has an advantage over 0/3 IF. There is another window from 5:09-6:21 where 3/3 is better than 0/3. And in fights over 7:33 in length 3/3 IF is better than 0/3.

But realistically - we don't ignore the WE. The WE is more of a DPS increase then IV so that's when we use cold snap. With a talented and glyphed WE it has a 1 min uptime and a 2 min cool down. So doing it again, but maximizing WE uptime with cold snap it looks like this:

with 3/3 IF

0:00 - Fight starts - cast WE #1 and IV #1
1:00 - WE over, cast Cold Snap
1:01 - Cast WE #2 and IV #2
3:01 - Cast WE #3
3:25 - Cast IV #3
5:01 - Cast WE #4
5:49 - Cast IV #4
7:01 - Cast WE #5
8:13 - Cast IV #5
9:01 - Cold Snap is ready, but WE is ready too - hold the Cold Snap until after WE is over
9:01 - Cast WE #6
10:01 - WE over, cast Cold Snap
10:02 - Cast WE #7 and IV #6
12:02 - Cast WE #8
12:26 - Cast IV #7
14:02 - Cast WE #8
... continue to cast WE every 2 min and IV every 2:24

with 0/3 IF

0:00 - Fight starts - cast WE #1 and IV #1
1:00 - WE over, cast Cold Snap
1:01 - Cast WE #2 and IV #2
3:01 - Cast WE #3
4:01 - Cast IV #3
5:01 - Cast WE #4
7:01 - Cast WE #5 and IV #4
9:01 - Cold Snap is ready, but WE is ready too - hold the Cold Snap until after WE is over
9:01 - Cast WE #6
10:01 - WE over, cast Cold Snap
10:02 - Cast WE #7 and IV #5
12:02 - Cast WE #8
13:02 - Cast IV #6
14:02 - Cast WE #8
... continue to cast WE every 2 min and IV every 3:00


So when we focus on the WE - if the fight length is between 3:24 and 4:01 then 3/3 IF has an advantage over 0/3 IF. There is another window from 5:49-7:01 where 3/3 is better than 0/3. And in fights over 8:13 in length 3/3 IF is better than 0/3.


How often you have fights that hit these windows is very dependant on your guild. It seems like when we were first going into Naxx we were pushing the enrage timer on a lot of bosses - we certainly weren't quicker than 3:24 very often. Toward the end 3.08 it seemed like most Naxx-25 fights were in the 3-6 min range. Now that we're still working through Ulduar it seems like we're back to pushing the enrange timers again.

Even when we're talking about a 10 minute fight where we know we will get an extra IV out of putting 3 points into IF - that's still only a 20% dps increase for 20 seconds out of a 10 min fight - an increase of only 0.67% for 3 talent points.

In a tree where we're short of DPS increasing talents I guess its better than nothing (especially in these 3.13 days where we're limited to 1/3 enduring winter). But this is probably the first points I'd drop for other things if I had to. Personally, I would take some non-dps talents like Arctic Reach and Ice Barrier before investing in Icy Floes. If you are focused on trash it may also be better to take 1 point in frostbite and 1 point in improved blizzard.

In 3.2 when we go back to 3/3 Enduring Winter and Brain Freeze is even more devalued it will probably be easier to drop Brain Freeze all togher in favor of Icy Floes.

*In doing some testing with RAWR - it does not appear to simulate Icy Floes at all. No matter what I would do to adjust the fight length there is no difference between 0/3 IF and 3/3 IF.

You should be using cold snap right after you use IV + WE, starting the CD on it a minute earlier. Also, talented Cold Snap's cooldown is only 6:24, so you'll pop it again right after you summon your elemental at the 7 minute mark.

I'm in the "fortunate" position of having my frost mage as an alt. The World of Warcraft Armory

So while I'm not doing wrecking ball BiS dps, I can still play the spec I enjoy the way I enjoy it and still pull top dps for the content (Naxx, heroics). That does make sense though and I'll swap the molten armor glyph back in for the WE one. I don't look forward to the day I scale beyond ice lance and bf, which I imagine will happen at or shortly after 3.2.

OK, it looks like this was a faulty assumption on my part. I read lots of discussion of ghost charge lances in conjunction with simcraft output, and assumed that it was actually simulating ghost charge lances. It does not appear to be doing so. I was mistaking natural variation between simulation runs for a change caused by removing Ice Lance from the action list -- it was just coincidence that the variation always gave a few DPS more to the runs in which I had made that change.

So I can't tell from Simcraft whether ghost charge lances are worth doing after the Empowered Frostbolt change or not. Someone will need to test that in Rawr, assuming Rawr can do it.

However, I've verified that Brain Freeze is now a DPS loss (except, perhaps, on ghost charges).

Output and config files used can be seen here: Simulationcraft Results

Rawr will be able to do it, but I don't believe Kavan updated the 3.2 frost changes yet.

Incidentally, for 3.1 calculations Rawr's "compute optimal frost cycles" suggests that it's a ~1dps increase to use brain freeze on the ghost-FoF as opposed to always using ice lance.

The "ghost charges" are supported. I piggy-backed on the "shatter_combo" code to implement it.

However.... While looking I realized that it does not perform any aura_gain() or aura_loss() calls so you can't tell what is going on when you generate a log via log=1.

I've cleaned up the code (functionality hasn't really changed, just better messaging, etc). It can be accessed via SVN r2835.

./simcraft config.simcraft log=1 output=config.txt

You will see it gaining Fingers of Frost and then gaining "Ghost Charge" when it falls off..... You can watch the sequence of actions, etc, to make sure the sim is behaviing in the expected manner.

BTW Lhivera, since you are doing some experimentation....... I never removed the code for Winters Grasp. You can specify any talent by name after the talents= line in the config. In this case: winters_grasp=2

I just checked the builds and no he has not updated RAWR's with PTR Build 10072 changes. The only update he's done to RAWR since 10072 was a partial implementation to the Silverlight Options menu (They are planning a web interface in addition to the stand-alone we all know and love).

OK, here we go. Thanks to dedmonwakeen for pointing out where I was going wrong. Here are the new results:

Simulationcraft Results

For those who don't want to link through:

7339 Fire (for point of reference)
6279 18/0/53 w/spell impact, using brain freeze and using ice lance on ghost charges, w/Ice Lance glyph
6199 18/0/53 w/spell impact, no brain freeze, using ice lance on ghost charges, w/Ice Lance glyph
6191 19/0/52 no spell impact, no brain freeze, no ice lance, w/Water Elemental glyph
6102 18/0/53 w/spell impact, using brain freeze, no ice lance, w/Water Elemental glyph

So the status quo remains. If you can execute ghost charge lances reliably every time, it's still best to do so by about 1.4% over the spec that I talked about upthread, using frostbolt spam.

The frostbolt crit % for the results:

66% (IL+BF) - 63.3% (no BF) - 64.9% (frostbolt spam) - 61.7% (BF)

6279 DPS is 1.3% more DPS than 6199 DPS, but you had 2.7% more frostbolt crits.
6191 DPS is 1.45% more DPS than 6102 DPS, but you had 3.2% more frostbolt crits.

What causes the differences in frostbolt crit chance between the simulations? It would seem to me that the frostbolt spam spec should have the highest frostbolt crit chance of all (thanks to SOTM), but it comes second.

What percentage of ice lances need to fail to hit a frozen target before they actually become a DPS loss? (This is why I don't rely on the ice lance glyph.)