Originally Posted by Rivkah
|
First, great work on your recent updates to FD. Thanks again for all of the hard work on the site. I really rely on it.
I took your linked MM simulation and made some modifications to increase its DPS by about 770.
Note that your link currently does 85641 as is.
I got it up to 86411 with this saved setting:
Zeherah's Hunter DPS Analyzer.
The changes I made for this simulation and the reasons are:
- In the Shot Priority, I moved KS before GT. KS currently does a little more single target damage than GT and costs 15 less focus to use. Furthermore, with KS having a shorter CD with "back-to-back" uses, delaying it for GT has a larger impact than delaying GT for it.
- In the Shot Priority, I moved Stampede to after RF. This ensures that the first Stampede is fully hasted and that the second Stampede becomes available closer to when the 4th RF becomes available when Readiness gets off CD. In your sim, the first Stampede is done about 3.5s before RF and 4.7s before Readiness meaning that it comes off CD for its second use about 8s before the RF resulting in the RF not fully overlapping Stampede. With the altered priority, the first Stampede is cast with RF and is fully hasted as desired. The second is cast about 2.3s before the 4th RF, which means that RF does full overlap with Stampede. Personally, I would wait the 2 to 3 seconds to cast Stampede after RF to be safe in case Stampede "snap shots" stats at it start. (I assume it does not but I do not know for sure. Have you tested this?)
- Changed the setting for "Minimum focus to cast Arcane Shot" from 67 to 61. Either way, this setting basically results in saving focus for CS without having it listed in the "Save Focus For" setting. The reason that I use 61 is that the cost of casting an AS then a CS is 20 + 45 - 4.5 (approximate focus regen over the AS GCD) = 60.5, which I round up to 61. This fine tuning allows the sim to cast a 4th consecutive AS in some situations where DB is active that the original case did not while still maintaining the CS cycle. This results in a little better use of focus with less focus capping since 3 ASs cost 60 focus but the focus regen over the 3 ASs and the 2 SSs is a little over 60 when not including latency and even a little higher with accounting for latency. Lowering this setting below 61 results in not having enough focus to cast CS on come cycles and pushes some CS cycles back since 1 or 2 SSs are needed. The Sim clearly illustrates that. Changing the setting to 61 saved enough focus to allow 6 more ASs to replace 4 less SSs.
- I did 2 regemming:
-- The first is the shoulders, which has a blue socket and a +60 agi socket bonus. Your setting puts a Delicate (+160 agi) in this socket and forgoes the socket bonus. I was confounded by this at first since it seems clear that putting in a Glinting (+80 agi and +160 hit) would be better since with the socket bonus it is -20 agi for +160 hit, which theoretically is clearly in favor of the hit. Then I realized why you did this, which is that with all of the large number of stats that it was near impossible to make this change and stay only a little bit over the hit cap. All of the options I tried that were over the cap, all were over by more than the 14 you had and was a DPS loss. I almost gave up on the gem change until I decided to see how going a little under the hit cap would affect things. With the 14 hit over cap to start and the 160 from the gem, we were 174 over cap. There are two options for removing 180 hit. The first is undoing the haste to hit reforge on the shoulders themselves. The second is the glove enchant being changed from 180 hit to 180 crit. Both options result in being 6 below the hit cap (-0.02%) but are DPS gains.
-- The boots have a +120 haste socket bonus with a yellow socket. Once again you used a Delicate (+160 agi). Using a Deft (+80 agi and +160 haste) provides a trade off of -80 agi for +280 haste, which with being a 3.5 ratio generally favors the haste. Now since this case is already at close to an ideal haste amount, adding this much haste is a loss, but some of the additional haste can be converted to crit by changing the Ring 1 reforge for a DPS gain.
- I also messed around with reforges for a bit to increase DPS more; however, some of these increases may not be real and just shot shifting.
- I also changed the "Maximum speed to cast Aimed Shot (suppressing arcane shot)" settingfrom the Cata 1.8s value to 1.4s. This did not affect the results since all AIs previously being casted were all below 1.4s. To illustrate how around 1.4s is a reasonable value, if you lower it to 1.3s, which makes the the sim cast AS instead of AI over the 2 RFs outside the CA phase, you see a minor DPS loss of only a few hundred DPS.
- One correction over previous statements that I have made is that it does now result in maximum MM DPS to overlap BL and RF affects, preferrably during the CA phase. The reason for this is the low maximum speed to cast AI. While in Cata where AIs during BL were well below the cutoff point of 1.8s resulting in a huge DPS gain, AIs during BL in MoP are currently too slow to be worthwhile casting AI. However, the increased speed to AI during the CA phase where due to AIs always critting and the cutoof point being much higher, it is a huge DPS gain currently to have faster AIs and to cast more of them. The MoP version of FD was shwoing this from the start, but I just did not believe that that could be the case until I had theorycrafted it more. What also effects the situation is that we do a lot less SSs too. Hence, we haste a lot less SSs than we used to by chaining BL and RF instead of stacking them to below the GCD.
EDIT: After looking through the shot logs and comparing it with target dummy testing, I realized another needed settings change in the Shot Priority. This is moving AI below AS. I know that this may seem counterintuitive since AI does so mcuh more damage than AS, but this priority ordering along with other settings like suppressing AS during CA phase, maximum speed to cast AI, and minimum focus to cast AS makes everything work as desired.
The settings for suppressing AS during the CA phase and the maximum speed to cast AI and suppress AS already control when we use AS or AI hard casts as the focus dump and make their use mutually exclusive such that the Shot Priority list has no impacts on the decision.
The Shot Priority list only affects the priority of the MMM AI procs versus using AS. We already know that when we get a MMM AI proc that we want to use it before the next SS cast that is not the second in a MM pair to maintain SF and before the 10s window before the buff drops. Since we should already be performing an SS pair every 10s to maintain SF, then we should always be using MMM AI procs before the expiration. This means that we do not have to use MMM AI procs right away and can put it lower in the list of other shots with CDs so that we delay the MMM AI cast (with no penalty for doing so) to avoid delaying the other shot's CD by a GCD.
The same is true for AS although it does not have a CD. MMM AI procs do not have to be cast before an AS. Whether the AS or MMM AI proc is cast first does not matter from the perspective of using the proc before it expires. By having AI higher in the priority list than AS, then MMM AI procs will always be cast before AS even if we just got done with a SS pair and will cap focus during the MMM AI GCD. By switching the priority, this situation is avoided. If we are high on focus, then AS will be cast first to help avoid focus overcapping and provide better focus utilization. if we are low enough in focus, then the Minimum Focus to Use AS setting will be in effect and will cause the MMM AI proc to be used instead of the AS. The is more ideal and better models how the shots should be cast in combat.