Thanks for the data. I count only 8 SPELL_PERIODIC_DAMAGE events in the log you posted, rather than 9. That is consistent with the analysis in the sim thread: ES acts like other Cata dots in that it clips and extends the current dot after its next pending tick. In your log, the last tick of the second shot was clipped. There was a discussion early in Cata about whether clipping was worth getting ES on CD. Even then, the numbers were close between ES ES ES, ES ES AS ES, and ES CS ES AS ES. I'm running similar sims with the new dot behavior, but the data isn't quite together yet.

I did some testing on this after hearing bout the dot clipping change.

I found that if I spammed explosive shot to que it up and make sure it fired *just* as the GCD popped I would only get 8 shots (second one was always clipping the first es). But If I waited to see the GCD pop up and then hit the shot (simply adding in human response time) then there was no clipping. It seems there is a slight time delay when the first exp is not yet registered as a dot yet and so gets clipped.

The 3rd one showed no change between spamming to que and waiting. Seems at that point the game accepts exp as a dot. It just needs that half second (or less) of delay between the first and second shot.

So that 0.2 seconds or so it adds on seems to be enough to keep me from clipping. And only seems relevant when dealing with the gap between the first and 2nd shot.

Perhaps someone else could also test this so we can see if results are consistent?

Imo it's a nice should be a dps boost as it ends up being extra explosive shot during the fight. ex-cs-ex-sc-ex puts ex on CD around 3 seconds after after 3x ex would. So theoretically you'd end up with one extra ex every 2 L&L procs over someone filtering in CS.

I'd really love to see the math comparing the 2 though. I'm not so good with the math part>.>

I've been playing around with FD to see if anything regarding stat weight / shot priority / talent value has changed between t11 and t12. I had a theory that the set bonus might change the relative value of mastery, because with both the 2- and 4-set bonus, a larger percentage of our damage will be elemental rather than physical. This turned out to be false, according to FD anyway. The value of mastery climbs slightly but it still remains our worst stat.

But I did find out something pretty interesting.

FD tells me something strange today: once you reach a certain ilvl pretty close to 370, kill command becomes favored as sv's focus dump over arcane shot. Using an armory import of myself with a couple key upgrades added manually, I would gain ~70 dps by focus dumping with kill command, and replacing glyph of ArS with glyph of KC. It appears to provide a larger dps boost as RAP scales.

FD still favors all pet focus talents over improved kill command - probably because kill command already has such a high crit rate. Another 10% crit would only give KC about 7% more damage overall.

Now, this may all be wrong. I'd love to hear an explanation as to why this result is incorrect, how I've configured FD wrong, or how FD's sim is a bit off.

Kill command has a much higher RAP coefficient than arcane, but arcane factors in weapon damage.
I could imagine either scaling more rapidly with ilvl, depending which of RAP/weapon damage scale faster:
Arcane - 100% weapon damage plus (RAP * 0.0483)+289
Kill Command - (849 + (RAP * 0.516))*1

edit:
Testing this, KC is 40 focus while arc is only 22. KC is behind arc in DPF, but ahead in DPCT. KC has a much higher crit rate.


My shot priority and settings are listed in public as 'asmo - t12' although this particular setting is actually mostly t11 gear.

edit #2, some more thorough testing to compare dpf/dpct of arcane shot and KC to cobra shot:
cobra: 3,694,593 dmg. 223 shots, 94.6% crit. avg: 16,567. DPCT: 12,005 (inside CA)
--- normalize this to 31% crit ---
(x 1.31 / 1.946) = 2,487,110 dmg. avg: 11,152. DPCT: 8081 (outside CA)
arcane: 892,031 dmg. 79 shots, 31.6% crit. avg: 11291. DPF: 513 DPCT: 11291 (delta 3210)
KC: 1,067,958 dmg. 82 hits. 41.5% crit. avg: 13023. DPF: 326 (64%) DPCT: 13023 (delta 4942, 154%)

Basically this was just a reality check on the sim's results. You dump focus less efficiently with KC, but for a larger dps gain. It's very hard to use these kinds of numbers to really make a call one way or another: you get 50% additional damage over CS with KC, than by using ArS. But your focus dump is 36% less efficient. Basically i'm saying from this standpoint it looks feasible, so I think it's reasonable to trust the simulator this time.

It is also worth noting that because ArS has a lower DPCT than CoS during CA, it is simply never worth casting ArS during CA phase. This in the past has severely limited the usefulness of triggering RF right at the beginning of any bossfight, since there was no good way to dump focus during RF/CA. KC on the other hand has higher DPCT than CoS, even during CA (although only by about 7%). So, if KC is the preferred focus dump anyway, RF during CA will be a bigger dps boost than before.


Disclaimer: All of this computed against a theoretical gearset that I don't actually have, and tested on my 364 gearset with t11 4-set. During the actual parse/test I used wolf pet; the crit rates would change substantially with cat or any other pet.

I've been playing with shot sequences for lock and load proc to leverage this dot change, but I wasn't gettign the right number of dot ticks. After a few iterations on my own, Caltiom reminded me of how to avoid clipping the dot. I eventually ended up with:

actions+=/explosive_shot,non_consecutive=0,if=dot.explosive_shot.remains<=(tick_time+travel_tim e)&!in_flight

But I was still getting only 7 ticks instead of 9 ticks. It turns out that there is a (simc) bug in the refresh of dots that tick when applied (i.e., on their the zeroth tick). Most dots don't tick immediately, and most dots that do tick immediately have a cd that limits their refresh occurrences. Thus, ES during LnL may be the only place where this is an issue. I did a local hack (add 1 to num_ticks in action_t::travel for a ticking dot that is supposed to tick on zero) and ran a few sims. That gives me the correct number of ticks (9) in the sim, and is indeed a minor dps improvement over the default profile:

DPS Ranking:
32215 50.0% ES_ES_AS_ES
32184 50.0% Hunter_SV_T12H
31776 33.0% ES_ES_ES <---unverified

However, in examining the output, the resulting shot sequence for the putative "ES_ES_AS_ES" during LnL is not typically ES ES AS ES; it's more like ES ES AS AS ES or ES ES CS AS ES. This is because the remaining dot duration after the second shot is much more than a second. To illustrate, I'll assume a shot timing of 1.1 seconds (gcd plus a little latency) and a travel time of 1 second.

Event	Remaining ticks	Time	Notes
ES 1 Fired	0	0
ES 1 Applied	3	1
ES Tick 1a	2	1
ES 2 Fired	2	1.1
ES Tick 1b	1	2
ES 2 Applied	4	2.1
AS 1 Fired	4	2.2
ES Tick 1c	3	3
ES 3 Fired	3	3.3
ES Tick 2a	2	4
ES 3 Applied	4	4.3	we just clipped tick 2c here
ES Tick 2b	1	5
ES Tick 3a	1	6
ES Tick 3b	1	7
ES Tick 3c	1	8

By comparison, the action rule above avoids clipping, and thus results in a different shot sequence. Both the sequence I get above and the actual ES ES AS ES sequences (and a few others) are worth comparing, but I thought it worth posting the preliminary data.

The interesting realization from the sim is that the delay is necessary. The timing of the first two ES ES is such that any latency at all is sufficient to ensure that they don't clip. The second ES must land just a little more than a GCD after the first one in order to extend the ES dot out the full extra three ticks. However, as in the table of the message you quoted, the third ES must land after tick 2b (at 5 seconds) in order not to clip. With just a single AS interposed, the third ES lands a full 0.8 seconds early (including latency on each shot), clipping at least one tick.

Since there should be a gap of 3 seconds between the second and third ES (one of which is spent as the GCD of the second ES), the ideal might be to squeeze in two AS (for ES ES AS AS ES), but that requires almost full focus. That's why the simulation mixes AS and CS between those last two shots: a single shot is insufficient, and there's usually not enough focus to afford two AS shots and still have enough focus for the follow-up ES (though I did see it get lucky once). Having enough focus for that in general is unlikely given that this is responding to a proc that could happen at any time. I could see playing with timing variations: at high haste, push back the third ES a little to squeeze in two shots. At low haste, just fire a CS and wait the 0.2 seconds or so for ES to become available.

The current profile is a little bit stale; I have it for reference below. Then by default I switched it to ferocity pet and One With Nature rather than Careful Aim (so that's the default below). In a previous note some time ago, I showed that further dropping SicEm for CA and 2 Shark Attack on the pet (instead of one SA one Wild Hunt) was a dps increase. I tried it with the ES variants as well (note that "other" below is what I refer to as "CS AS" previously; it's really just whatever the normal priority puts in). Here's some results from those experiments:

DPS	Description
32,904	ES_ES_other_ES with CA
32,584	ES_ES_other_ES with SicEm
32,550	Hunter_SV_T12H updated with a ferocity pet and OwN instead of CA
32,542	ES_CS_ES_ES
32,451	ES_ES_AS_ES (or as close as I could get)
32,430	Hunter_SV_T12H
32,141	ES_ES_ES

For the T12H, ES ES ES is the worst of the lot, often losing two ES ticks. The best so far is attached below, and is almost 600 more dps than the current profile. More interesting is that it is a slightly different but straightforward strategy for humans to apply. No that it also includes some small dps improvement from the "no AS while in CA range" rule. (Edit: different strategies may work better for different gear sets).

A minor note: the ES action is designed to fire ES as soon as it could while not clipping the dot. That also correctly avoids clipping if the shot just before the LnL was an explosive shot. In that case, the first ES extends the ES from before the LnL, then there's a 3 second gap, and then the second pair of ES.

I've been fooling around with Zeherah's Hunter DPS Analyzer trying to optimize my gear reforging (as well as compiling some BiS gear setups). I've been going by the following guidelines:

1) Reforge each item to maximize Critical Strike rating (E.G. Flamewaker's Legguards: 111 Hit -> Crit as opposed to 84 Haste -> Crit because the prior provides greater overall Critical Strike rating).
2) Reforge everything else to maximize Hit Rating (should bring one WAY over hit cap)
3) Figure out which pieces to reduce Hit rating to ensure I'm as close to 961 as possible (reforge out of hit to haste).
4) Now I figure out how much haste I can remove from gear (haste to mastery) and figure out what haste/mastery ratio provides highest DPS (according to the calculation).



This process is quite time-consuming but--looking at Rawr, Reforgenator, Reforgelite, and Mr. Robot--I find I end up with a higher damage overall than the things they suggest. The process essentially optimizes Critical Strike, then Hit Cap, then Haste "plateau," and finally throws all else to Mastery (our least favored stat to my understanding).

Items other than the Flamewaker Legguards (from step 1) I find I don't mess around with include: Firearrow Belt (77 Haste -> Crit); Lancer's Greaves (75 Haste -> Crit); Hide-Bound Chains (69 Haste -> Crit); Ancient Petrified Seed (153 Mastery -> Crit); Fluid Death (128 Hit -> Crit); Dreadfire Drape (55 Hit -> Crit). Again, my reasoning behind this is to optimize Critical Strike Rating.

My current gear setup (load US-Medivh Zardon) I'm at 10 hit rating over cap and I cannot find a way to reduce this to exactly capped. I was thinking--is my optimizing technique off base? I might be able to get exactly 961 hit rating if I do step 1 differently but--from how I see it--though I'm +10 hit rating over cap--I would essentially be reducing that hit but also critical strike rating. I'm not exactly sure what would happen but--in my head--it would be something like -10 hit, +haste/mastery, and -crit (regardless of what I do because I'd be reducing optimized critical strike rating).

-somewhat change in topic-

On a somewhat unrelated note--the first post on the MM thread tells us suggested settings to run Zeherah's Analyzer for the MM spec. It would be amazing if the first post in this thread had a suggested spreadsheet run settings. I've just been keeping it defaults and best raid buffs/debuffs for DPS comparison but I'm not sure this option best simulates real raid settings. Perhaps I'm looking too much into optimization and relying too much on the analyzer but--the way I see it--two hunters playing at the same skill level, one optimized and the other not, the optimized should do more damage.

It's probably time to at least talk about SV in 4.3 and I've done a bit of math using my own gear to find new haste points with 2pcT13.

Essentially the 2pcT13 allows a very strict rotation to emerge at very low haste values (~100 haste rating for this soft plateau).

ExS, BA, CoSx3 ... (ExS, AS, CoSx3)x3

(I call it a 3+1+1 for brevity)

Handle LnL as normal.

Above 100 haste rating, haste overtakes mastery in value for single target damage up to a hard plateau of over ~4100 haste rating.

------------------
The 4pcT13 also presents a rotation you can fall into when above a secondary soft haste plateau of ~1390 haste rating during the 4pc proc or under any other dynamic haste effect of at least 30% in magnitude.

ExS, AS, CoSx4 ... ExS, ASx2, CoSx3

(A 3.5+1.5+1 rotation)

BA may be substituted for any AS in this sequence.

I doubt he claims SV is competitive with top tier DPS, it's just not as bad as many people think.

There is not a lot of things player could change to optimize their DPS as SV currently and there is nothing that could make it go through the roof with current gear.

Taken from a Blue response on the forums

"Please teach me Blizzard!
When looking at the big picture of class design and performance, the answer is kind of both. I think the point is that specs not played nearly as often as others aren't necessarily underpowered. Our data may show a much smaller percentage of players using certain specs, though a good amount of the people who do use them competitively tend to do very well.

Here's a follow-up from Celestalon: Unholy and Survival are definitely underpowered in 4.2, and are getting buffed to compensate in 4.3. However, the point that Xelnath was trying to make, I think, is that they’re not as far behind as people think they are. With just a couple minor buffs, these specs are able to be competitive, in at least some encounters. Subtlety is a good example of this already: Subtlety is a very difficult rotation to play, and suffers quite a lot of drawbacks. But in some encounters, those drawbacks can be mitigated, such that it becomes the optimal spec. For example, the top Subtlety rogues are beating the top Combat and Assassination rogues on Baleroc, since that fight plays to Subtlety’s strengths. Will Subtlety win on every fight? No, probably not. But they’re not nearly as far behind as most players think, and sometimes can even be better. We’re hoping the changes to Survival and Unholy (and Destruction, and Beast Mastery) in 4.3 will help push those specs into a similar situation. We really hope people give these specs a chance. (Blue Tracker / Official Forums)

I find this to be a bit of double talk. One designer says Survival is not being played to its max potential while another states Survival was underpowered and is getting a buff. Suffice it to say, I don't think the community is missing the "optimal" priority/gem/reforging as Blizz initially indicated. What I do think is Blizz has a value in their head that they want the class to perform at and Survival is within that % deviation (low side).

What they're saying is true. They made a HUGE mistake by making it sound like the player base is dumb and doing it wrong. You get a value for each spec on a tank and spank fight. We know this, they know this. But what they're saying is that we, as the community, are looking att his in a vacuum. Furthermore, 90% of the player base is stupid and/or bad. So, when you take a sim based on a defunct type of fight and add in anecdotal evidence, the whole community goes "chicken little" and screams bloody murder. Is SV below MM? Yes, of course it is and it is getting buffed in the next patch. But, overall, it isn't in such a poor position that it is totally unplayable. The community MAKES it unplayable. Combine the mentality with a lot of poor players and posters paroting stuff that they neither discovered on their own or understand and you've got a really bad situation that the devs can't correlate.

There are actually quite a few reasons to raid Survival. The rotation requires less attention which allows you to pay more attention to fight mechanics and proper execution rather than which of your abilities you should be using next. It also brings one of the more important raid buffs, which is far more important for your raid if you don't otherwise have it than the difference between Marks and Survival personal dps.

Also, you are saying that the higher skill cap of Marks is the reason for it's higher theoretical dps, which isn't true. Marks has a higher variation in dps when you factor in player skill, but that has nothing to do with it's potential dps in comparison to Survival.