There are three factors in FD of which I am aware that are affecting the AI rotations to come out higher than the AS/AI rotations I suggest. These are:

- There is currently no option to hardcast AI during RFs and BL when not using it as the focus dump outside the CA phase. Using it instead of AS during dynamic haste effects is a large DPS increase.

- FD is currently over calculating focus regen during RFs. For instance, for my character, my base regen when under no dynamic haste effects is 5.149 fps. This is a correct number as calculated from 4 * 1.15 * 1.03 * 1.0867 = 5.149. It also correctly calculates regen during BL at 6.694 fps, which agrees with hand calcs of 4 *1.15 * 1.03 * 1.0867 * 1.3 = 6.694. However, it shows the regen during glyphed RF as a very high 16.22 fps, when it should be 4 *1.15 * 1.03 * 1.0867 * 1.5 = 7.723 + 4 fps from RR = 11.723. FD's number is about 4.5 fps too high, resulting in more AIs being performed than should be. There is no way focus-wise that we should be able to cast 3 AIs in a 12s CS cycle as is occurring. So for the RFs outside the CA range, using that extra focus for more AIs instead of more ASs (as is the current case in the AS focus dump case until the option to hardcast AIs during RFs is added) results in FD's number for the AI focus dump case is coming out higher than it should relatively. This bug has previously been reported and is being investigated.

- I am sure some of you started with characters with haste amounts tuned for the AI hardcast case and then just switched options and glyphs without retuning the character to optimal rotations and stats for this setup. Part of the problem is that the new ideal haste ranges have not been provided yet. But if you made the comparison this way, it probably skews the results in favor of the AI hardcast case.

After running some tests it appears that you need 950-1k haste from gear + 3/3 pathing+ 4pc t11 to support the glyphed CS rotation of CS-ASx2-SSx5[subbing AiS on proc for 5th SS] in a raiding environment with the 10% haste buff.
Obviously maybe a tad more to account for latency.
If youre not glyphing for CS you can drop this down to arround 850 ish[rough estimate may be off some]
Definitely alot better than hitting the 1383+ needed before lol

Either way its a tight rotation...maybe aim to be closer to 1k or slightly over to make it more manageable.

Focus didnt seem to be an issue in this rotation unless youre lucky with MMM procs which lead to about a half second delays in having enough focus to pop Chim or a second AS in some cases..but it was easily made up for and still supports using the Chim glyph so far.

Im running these tests with a 55 latency.


Edit
Damnit I put the wrong cast time #s in this then and had wow error out on me...and lost my player supplying the 10% buff.
The haste# range is correct though after abit more testing. A quick trip to FD shows the cast time#s on SS goal to be 1.3s or less with all buffs accounted for when not under any dynamic haste effects. Just from gear and with no buffs a target # on cast time would be 1.63 or less.

Whitefyst, I am just not getting the results you are regarding AiS vs AS using female dwarf even accounting for dynamic haste effects.

Disabling bloodlust, berserking, rapid fire, setting CA % to 0, swapping my prestor's for heroic essence of the cyclone and reforging appropriately still results in an AS dump being ~500 dps behind an AiS dump in a stand up fight.

I did a similar setup to what you did and got a 15 DPS difference in favor of hardcasting AI. Note that with the loss of point in BD and the FD fix for Sic'Em procs focus savings, that if you did the test with a Ferocity pet that 1 SA and 1 WH is now better. When doing FD with a CA phase, the frequent Sic'Em procs during that phase skews the WH uptime some.

I took a closer look at the FD data to see why its results did not mesh with my napkin math and SimulationCraft's results that have been reported. The most glaring thing that I observe is the autoshot modeling for losses from AI interruptions. In FD between the two cases, 27 AI were lost and only 37 autoshots gained. Thus, there was an average of 1.37 autoshots lost per AI cast. I had approximated that it would be on average 1.5 autoshots lost per AI cast with the actual value experienced in a fight depending on timing. With the 1.5 ratio, then 41 autoshots should have been lost. The DPS of 4 more autoshots is about 102 DPS. What needs to be added to that is the 1 to 2 more WQ procs and the 1-2 more GftT procs. This all works out to being a small DPS gain for my character.

Maybe my 1.5 autoshots lost per AI cast estimate is too high. If it is, then that will skew the results in favor of hardcasting AI. What do you all think?

One thing to take into account though is that this is ideal modeling. In reality you will have times where your AI casts experience some pushback, in which case more autos (and WQ and GftT procs) will be lost.

Now I will readily point out that a lot of us are around the inflexion point where the two choices are roughly equal and depending your setup and gear, one may work out slightly better than the other when doing the ideal simulation. However, with all of the movement required in combat, even brief ones to move out of fires, and all the interrupting effects, if the two cases are roughly equal in an ideal simulation, I believe that you will find the AS case better in practice.

A final factor is that using AS applies the MfD effect. This will not always provide a benefit depending on the situation, but it can in some.

Anyway, when I get to that part of the guide to update, I will make sure to point out that in general we are at the tipping point and that one case may be a little better than the other for some folks and that it is up to the individual to decide for their setup and playstyle which focus dump (AI or AS) is best for them. However, my first pass through the rotation section will only be for using AS as a focus dump outside the CA phase and when under no dynamic haste so that I can get what I think is the generally recommended case out first. Then I may go back and support the AI hardcast case during those situations.

1.5 autoshots seems to be a pretty good estimate based on the Simcraft results, maybe even a little low. The all Aimed Shot rotation got off an auto-shot every 3.49s while the AS outside CA rotation got one off every 2.42s, so 44% more autoshots for the AS rotation. Note that this is for the entire fight duration including periods under dynamic haste effects when you would lose less than 1.5 autos per Aimed Shot so the actual number of auto-shots lost during the "normal" portion of the rotation is likely 1.5-1.6.

I know you said you were testing whether or not to use Chimera Shot during the careful aim phase. With the CS Glyph, would you still just use CS at the last possible moment to extend the sting during CA? I can't see using it off CD during CA as a Dps increase.

If one was going to attempt the AS subsitution for AI during the CS phase, has anyone looked at how it might affect the Termination phase if you talent for it?

I would assume the AS focus dump might make taking Termination valuable (ie not wasted focus) but I am not sure. Assuming the rotation that Noleafclover is proposing should we avoid Termination?

With the fix that just went into FD to remove the excess focus generation RF was providing, the DPS value of the RF glyph seems to go down, to a point where the AI glyph shows as better (so primes are arcane, steady and aimed). This is still probably not really correct since FD still isn't able to hardcast AI during RF.

Yes, I expected that the switch to AS as the focus dump during most of the fight would greatly diminish its value, although it should be a little higher as you suggest once FD allows AI hardcasts during RFs when not the docus dump. This is mentioned in the updated glyph section of the OP and part of the reason that I now list the CS glyph as probably the best third option, although its preferrential.

The Aimed Shot glyph is a bad option for the last glyph, at least last I checked. During the CA phase and RFs, you should have plenty of focus to support the AIs without it I believe (still need to verify). Then for most of the fight you are only doing an MMM proc AI every 18-20s on average.

Just for reference- my current implementation on FD for loss of autoshots from aimed shot uses two factors:
1) I estimate the loss of the mid-cast autoshot from starting a fresh aimed shot to be .5 autoshots
2) I tally up the cast time from each hardcasted aimed shot, and subtract the number of autoshots that would be cast during that time based on the average expected autoshot cast speed

I'm not really sure if this is the best method, but at the time it seemed the most logical approach. The biggest issue with this approach is I'm not actually tracking how fast the autoshots are when the interruptions are taking place, I'm just using the average case autoshot. With a more dynamic use of AI, on average the autoshot cast speed is going to be shorter when you're actually using it, so that would throw off the numbers somewhat- since without the interruptions you'd actually be casting more autos during that hasted period. When you hardcast aimed over the entire fight the estimate would be closer to actual practice.

I'm considering spending some time working manual autoshot tracking into the simulation instead of the current estimated shots method (since it looks as if we will be living with this aimed shot situation for some time). That requires some significant coding though so I'm not sure how soon I could get around to it.

If I'm understanding you correctly Zeherah, I don't think that the autoshot cast time during interruptions for AiS matters regardless of haste until AiS gets pushed under 1 second per cast. While the autoshots are firing faster, the AiS is casting faster as well, and the number of autoshots lost per AiS cast should remain constant over time.

In modeling this behavior, it should be as simple as multiplying the number of hard cast AiS by the average autoshots lost per cast to come up with the total number of lost autoshots each sim.

However if you're only modeling 0.5 autoshots lost per AiS cast, that seems quite low. On average it should be (AiS cast time)/(weapon speed) + 0.5. For my own 2.9 speed bow (2.9/2.9) + 0.5 = 1.5 lost autoshots per AiS.

This is what I factor into my hand calculations and how I came up with the 1.5 number I use. Most hunter weapon speeds are between 2.8 and 3s. With base AI cast time of 2.9s, the autoshot cast frequency will always be about the same as the AI cast time.

I agree that this simple change should increase the accuracy, especially if you explictly account for the weapon speed as D3thray suggests.

I've been pondering whether careful aim is actually really a worthwhile talent. I can understand that maybe when killing adds etc.. it might be worthwhile. However, on a number of fights that I've been on there has been a considerable amount of time where we stop dps and literally do nothing awaiting the perfect time to transition. Chimaeron and Maloriak being examples of both.

Increased dps in the first 20% of a fight I just can't see being a make or break thing on fights. If it was on the last 20% i could see a point to the talent, but the first. Has anyone else got some thoughts on this?

We've touched on that before, but as MM it seems moot since we don't really have a choice about it - I can't get down to Master Marksman and Rapid Recuperation without either spending points in either CA or Termination (or spending points in much worse talents like Resistance is Futile). You can argue that if CA is useless, Termination might be more useful, but most people don't like the idea of Termination (means not doing a rotation that we optimized for the middle 60% of the fight, gives bonus focus at a point of the fight we're using focus free shots, doesn't reduce the number of StS we need to do to keep ISS up, etc).

Personally even if it is a fight where we hold DPS to wait for a transition, I think having to do that is better than scrambling to make a transition in time.

For other specs like SV, debating whether to take CA or not is a better question, since you're actually choosing between things like CA and Frenzy or CA and SE.

Careful Aim is also worthwhile with regards to meeting Enrage timers on some bosses.

While it's not helping for a large portion of the fight it is a huge boost for that first section which especially on first kills can make the difference at the end.

Without the 4 piece T11 bonus I'm finding that you would need about 1260 haste rating from gear and 3/3 Pathing to get 5 Steady Shots and 2 Arcane Shots between Chimera Shot without the CS glyph. That leaves you with a 1.4 second Steady Shot cast time assuming the 10% haste buff while outside dynamic haste effects. It also works out pretty well during Rapid Fire as it gives you almost exactly a 1 second Steady Shot cast time. That also means that if you get an MMM proc you are only waiting 0.4 seconds to fire CS.

If you are close to 359 item level and don't yet have the 4 piece bonus I think this is a pretty good amount of haste to shoot for if you plan on using the Arcane Shot focus dump. Now that FD has been updated to allow you to hardcast AiS when under dynamic haste effects I'm playing with the settings and finding that 1264 haste with my gear seems to be about the sweet spot when using the settings I used. If you'd like a look search FD public settings for:

359 - no 4 piece

I found that disabling CS and SrS during CA was a little over 70 dps gain. I also found that hardcasting AiS during both RF and Bloodlust were dps gains. I wasn't sure that using it during BL would be as beneficial, but it was about a 259 dps gain.

For those of you that use FD, I have determined at least two items in FD that affect the outcome of the decision of whether to use CS and SrS during the CA phase.

My hand calculations was showing that it is more beneficial (although slightly) to use CS and SrS during the CA phase, but many people were reporting that FD was showing the opposite. The two factors that I have found so far that affect the result are:
1) The setting to save focus for Chimera Shot. With how this setting requires you to have 94 focus before firing an AI, having it set can result in both shotting less AIs than you could have and wasting focus. However, turning this setting off is not ideal either since it can result in you not having enough focus to cast CS in time and your SrS falling off.
2) The latency. FD applies the latency to your instant casts but not your cast time shots. This ends up favoring casting AI over CS. However, if latency is applied to both types of shots, or if your latency is really low, then CS is favored again.

At 0 latency and with the save focus for Chimera Shot setting turned off, I got about a 750 DPS increase with casting CS and SrS over the CA phase when doing a test of the CA phase only.

If latency is indeed an issue that affects instant cast shots only and cannot be eliminated with the shot queuing, then that would end up favoring not casting CS and SrS in the CA phase with the current gear set.

EDIT: Okay, I got educated on the fact that latency does still apply to shots that take a GCD or less to perform. Thus, this does still favor the AI only rotation during the CA phase, but only by just a little. However, I still recommend using CS and SrS since I believe they are better options in real game situations that involve movement and interruptions/push backs of cast shots. If two options are relatively equal and one is cast time and the other instant, the instant choice is better.


The quoted comment is not a statement of how much haste you should have, it is just a clarification of what I mean when I say haste from gear. I added this clarification since I got a large number of posts asking for clarification on what "haste from gear" meant and whether or not it included Pathing.

The 1260 haste rating from gear that you list for the case without the T11 4-set is consistent with the 9.75% that I recommend as is your statement about using AI as the focus dump during RF and BL.

Although FD may show disabling CS and SrS during the CA phase as a gain, I recommend actually using them during the CA phase per the analysis in the OP and per the discussion earlier in this post concerning FD modeling.

Fiddling around with the spreadsheet comparing Arcane as a focus dump sub-80% to straight Steady spam sub-80%, and it's showing a ~400 dps gain in favour of Steady spam in 372 gear.

Since I can't create custom weapons, I added 2500 AP and the dps gap widened even further in favour of Steady spam.

While Arcane itself gains more damage per AP/weapon dps than Steady, the damage Aimed gains per AP/weapon dps is much higher than Arcane. So it seems that the extra Aimed! procs that Steady spam grants more than offsets any dps gained by replacing Steady with Arcane to dump excess focus. Steady also gains from addition haste at certain points, and also from crit a second time via Piercing Shots.

I wouldn't be surprised if all MM hunters end up dropping Arcane for static single-target rotations in T12 gear. Blizzard will have to increase Arcane damage by ~15-25% to make Arcane worth using as a focus dump.

Your results only show part of the story.

It is true in the no CA case that when going from a AS focus dump rotation to an SS spam only case between CS casts that hunter DPS will increase. In FD with, the increase in hunter DPS is 140 with my current gear and 148 in full 372 gear (pretty constant). This increase is due to the additional MMM AI procs as you state; however, this does not tell the whole story.

The loss of using AS results in significantly less Sic'Em procs greatly reducing pet WH uptime and pet DPS. At my current gear, the loss in WH uptime (using 1 SA and 1 WH in both cases since it was best) is 44% with a 310 loss in pet DPS. Thus, its a 170 DPS in total hunter and pet DPS. At the 372 gear set, the loss in WH uptime (1 SA and 1WH is still best for both), is 45% with a loss of 387 pet DPS (higher crit is more Sic'Ems lost). This resulted in a 239 loss of total hunter and pet DPS.

Hence, with just considering this factor, if it is a fight where you pet can't be used well or is dead, spamming SSs is better than casting AS as the focus dump. However, for most fights where our pet is active at least 50% of the time, you are better off using AS.

Another problem with this analysis is that dynamic haste effects favor SS instead of AS since the SS cast time is reduced while the AS cast time remains constant. This doesn't factor in that in either rotation that while under dynamic haste effects that casting hasted AIs is better than casting just hasted SSs (which may be capped at the GCD) or ASs. Factoring in this reduces the benefit of the SS only case relative to the AS case.

Is this with AS glyphed?

I show a loss of 32% WH uptime (down from 90%) with a 1/2 + 1/2 specc'd pet. Dropping Steady is still a nearly 400 dps gain, even after a ~180 dps loss from the pet.

You didn't really explain anything. Did you swap the Arcane glyph for Rapid Fire? Did you drop a point from Pathing for BD? At lower gear levels the gap nearly disappears, but it seems to increase with improved weapons and higher AP.

I was only comparing Arcane to Steady. I'm well aware that at sufficient haste casted Aimed surpasses both.

By design, a focus dump should be a non-trivial dps increase over non-stop focus capping. But Arcane is underpowered as a focus dump.

In my analysis and response to you, I did the following, which I either explained or felt were obvious choices:
- Looked at two cases - my current gear set without the T11 4-set and at about a 350sh ilevel and a full 372 gear set with the T11 4-set
- I did not change any talents beyond untalenting CA and putting those points into a useless talent like CB
- Between the two cases, I changed 1 glyph. For the AS focus dump case, I used the AS glyph since it makes no sense to use AS as a focus dump without that glyph or to compare an unglyphed AS to a glyphed SS. For the SS case, I used the RF glyph (I did also check the AI glyph, but it did not do much.). The third glyph is both cases was CS.
- I reforged haste as neccessary to improve each case
- I set up FD to cast AI when its cast time is less than 1.8s (during RF and BL)

And yes, as a focus dump AS is rather weak when compared to SS, but when you glyph for AS, as you should if you use it, it is still a small overall DPS increase that seems to get a little larger with gear scaling instead of smaller as is the case if you do not glyph it. The straight hunter DPS component may scale slightly negatively with gear increase, but the pet DPS component scales positively more since your pet does more DPS at higher gear and more crits result in higher WH uptimes and DPS.

And the reason that hasted AI was mentioned is because if you are going to compare the AS versus SS cases against each other, you need to compare them with how you would actually ideally use them. For the SS only case, you would not still be casting SSs only during RFs since that is a waste, you would use AIs too, and that should be factored into the comparison is all that I was saying.

Since you seemed to be making a recommendation to use SS instead as AS outside the CA phase, I was only trying to make sure that you accounted for all the factors when making that recommendation. Since it appeared that you did not and that my analysis showed the situation to be the opposite case, although still very close, I am obligated to explain why I do not recommend what you do.

I will readily agree though that I think AS needs a buff to make it truly a worthwhile focus dump; however, even without a buff, it is still better to use overall in most situations.