[TC] SimC Analysis of Haste vs Mastery at 588 BiS

[Komma]

As of 5.4.8, there's a lot of debate on high end fire mage stat priority. Generally, everyone agrees that Crit > all, but after that there really isn't much consensus of what works better and when. Now that SimC is doing a decent job, we can answer some of these questions.

Up until now there have been several main suggestions, as documented by Kver's advanced guide:
1. Traditional build: Crit > Haste > Mastery
2. 2T16 5-stack GCD cap build: Crit > Haste to 14100 (GCD cap - 5 stack 2T16) > Mastery
3. Mastery build: Crit > Haste to 9623 or 11.6K > Mastery

Since we all agree that Crit > all, the goal is to find a balance between Haste and Mastery. We started with the new SimC BiS FIre profile, which gives ~20K Crit/~11K Haste/~11K Mastery. Here is a reforge plot, indicating what would happen if we traded haste for mastery in either direction:

250K iterations, 450 seconds, single target. Haste/Mastery values are PBoI amplification.

Since we did say that it's BiS Fire, it comes as no surprise that 11K Haste/11K Mastery gives us pretty close to the best results. What's interesting is that between 9K Haste/13K Mastery and 13K Haste/9K Mastery, there is practically no difference in their value!

450 seconds is 7.5 minutes, which is a lot longer than your average BiS Fire mage takes for the first 12/14H bosses. We ran this again for a shorter fight:

25K iterations, 200 seconds, single target. Haste/Mastery values are PBoI amplification.

As many have said, mastery is significantly stronger for shorter fight durations. In our graph, trading mastery for haste gives obvious DPS losses, while trading haste for mastery gives us more DPS until roughly ~9K haste.


From the above we can draw a few conclusions:
1. The Standard and 2T16 GCD cap builds are outdated for high end raiding. Haste becomes weaker than mastery long before the GCD cap is achieved.
2. On long fights, haste and mastery achieve parity between 9K-13K. Haste does not pull ahead.
3. On short fights, beyond ~9K haste, mastery>haste.


Here are the gearing strategies we recommend for high end (584+) fire mages:
1. Get as much crit as you can, same as always.
2. Get a minimum of 9K-10K haste. This allows you to meet the earlier breakpoints, such as +6 Glyphed Combustion (9095 haste), and +3 Living Bomb (13163 haste, achievable with 5 2T16 stacks).
3. For the remainder you can either:
a) Keep haste and mastery roughly equal. This will give you more mobility and flexibility due to shorter casts, and you'll have an easier time keeping up 2T16 stacks.
b) Dump all the rest into mastery. This will give you a strong opening burst, and powerful ignite/combustion cleaves on stacked targets.


Special thanks to Frosted, Dutchmagoz and Kver for helping acquire these results.

[Dutchmagoz]

Keep in mind that this was all done on a BiS 588ilvl profile. When I ran the same tests for my character (~585 ilvl with suboptimum trinkets/weapon) the results were still practically the same, but slightly favouring haste > mastery still at 9k, and the perfect ratio came somewhere around 10-10.5k haste -> mastery.

Don't apply the discussed logic on characters without at least 580ilvl and 18k+ crit, we will get to optimum gear scenarios for those in the future.

[Garrod]

At 581 equipped, I'm seeing about the same. What blew me away was that there was a 50k dps difference between 11663 build and dropping my haste down to 10k and evening it out with mastery. Interested to see how I do on our heroic clear today with it.

[Frosted]

Keep in mind that this was all done on a BiS 588ilvl profile. When I ran the same tests for my character (~585 ilvl with suboptimum trinkets/weapon) the results were still practically the same, but slightly favouring haste > mastery still at 9k, and the perfect ratio came somewhere around 10-10.5k haste -> mastery.

Don't apply the discussed logic on characters without at least 580ilvl and 18k+ crit, we will get to optimum gear scenarios for those in the future.

I had the same results from my profile, which has significantly lower crit than both Dutch and the BiS profile.

With decent Crit, you probably only need a bare minimum amount of haste necessary to keep up a fast enough fireball/GCD time in order to continue pushing out Pyro!s at a fast enough rate to keep your 2pc bonus up - which is then providing the rest of the haste you would need. I'd be interested in seeing what happens to these plots sans the 2pc (just to confirm if it's the 2pc or some intrinsic property of fire that is effecting haste/mastery values. Back in ToT mastery was considered higher than haste due to Combustion. This may not have been entirely true, it's contribution as a throughput stat may have been a bigger reason and we just never knew it)

[Boomop]

I'm @ 16.2 Crit. See pretty much the same results. There's a Haste bump somewhere around ~11.5k. Basically, Fire always wants more Crit, but there are Haste humps (for lack of a better word).

The difference between 14k and 11k seems to result in less Pyros, but that dmg is made up for by Mastery. The 2pc has relatively similar uptimes, although the feel of the spec is def different between the two. That would also loosely translate into 14.1k = better sustained. 11.5k = better CD burst.

Messed around on the SIMs with a buddies toon that had 19.5k Crit and 14.5k Haste. Same deal. Haste bump around ~11.5k, just transposing Haste and Mastery. Overall single target DPS difference between them is minimal. ~1600.

[Dutchmagoz]

Simc disagrees with 14.1k being better sustained. Even on very long fights (10minutes) simming 10.4k haste still does more DPS than 14.1k haste.

[Boomop]

Simc disagrees with 14.1k being better sustained. Even on very long fights (10minutes) simming 10.4k haste still does more DPS than 14.1k haste.

Think you missed what I was saying above. The lower Haste value shifts more dmg into our CDs. That tends to be more beneficial in raiding.

Wasn't talking about sustained DPS over the length of the entire fight, rather sustained DPS between the CDs -- basically DPS that is not a result of CDs. Should have called it instantaneous sustained DPS :p. The same way you can travel at 70 miles an hour even if you only drive two miles :p.

[Methusula]

I haven't seen this discussed in irc, but have you guys done any comparisons of simc 2pc uptime vs actual parses? Obviously, haste is somewhat devalued with higher 2pc uptime, so analysis of realistic uptimes could be relevant

[Frosted]

I haven't seen this discussed in irc, but have you guys done any comparisons of simc 2pc uptime vs actual parses? Obviously, haste is somewhat devalued with higher 2pc uptime, so analysis of realistic uptimes could be relevant

I'm not aware of a way to parse out the uptimes of the individual buff stack states from current logs in WCL/WoL. If someone knows a way, it would certainly be worth a shot. I think currently, all 5 buff states are bunched together.

[Methusula]

You click buffs-Potent Flames, and there is a handy graph plotting out the stacks in WCL.

Here's an example:http://www.warcraftlogs.com/reports/z61 ... ity=145254

[Dutchmagoz]

That doesn't show the uptimes on each stack though, you have to manually select all timeframes and add them ujp and divide them over the duration of the fight, which is quite inaccurate.

[Komma]

The impression I get from writing the APL is that fire is actually much more vulnerable to disruptions than you might expect. The 5 second window with 2T16 actually creates an extremely high skill cap with regards to reaction times and HU/HS planning. This combined with things like trinket pyro chains, aura delays, spell latencies and reacting to spell travel time delays means that I expect the sim to have quite a bit higher 2T16 uptime than you would see in regular parses.

The high skill cap probably explains the early preference for haste, and scorch-based fire players. Both of these make the rotation significantly easier to master and play optimally. As we move towards end of expansion and players become significantly more familiar with the rotation and tricks though, mastery-heavy fire specs are starting to dominate - just as the sims suggest.