Jump to content


Photo

Hunter DPS Analyzer (MoP edition)


  • Please log in to reply
176 replies to this topic

#1 Rivkah

Rivkah

    Great Tiger

  • Members
  • 838 posts

Posted 11 August 2012 - 07:09 AM

This thread is for discussion of the Hunter DPS Analyzer (also known as FemaleDwarf.com). The Mists of Pandaria version designed for level 90 is now available at www.femaledwarf.com. The Cataclysm version is no longer available.

Private messages are preferred for bug reports- please accompany any dps discrepancy reports with the name or link of your saved setting so I can test effectively (also specify whether your issue relates to the beta or live version). More general feature discussions can go here.

You can see the frequently asked questions for common questions, version history, planned features, etc. Please be sure to read the known issues list before reporting an issue. Note that it may take a day or two for new patch changes to be reflected- you can check the version history to get details on the most recent updates.

You can also get announcements of the latest changes to the site from Twitter.

#2 Whitefyst

Whitefyst

    Great Tiger

  • Members
  • 771 posts

Posted 18 August 2012 - 11:26 PM

If anyone wants to start with my settings as a base for their own tests, this is what I've been using:

MM: Zeherah's Hunter DPS Analyzer


I finally had a chance to look at your MM simulation. The initial DPS was 96992.

Here are some comments I have for increasing the MM numbers or possible bugs that need to be fixed.

1) PS not being calculated correctly:

It does not appear that PS is being calculated correctly, which greatly hurts MM DPS. During some 30s CA phase simulations, the only shots that trigger it are SS and AI, and both are at 35.6% +75% =111.6% crit chance. Thus, their crit chance should be capped with doing 100% crits.

With 8 AIs at 154849 damage each and 10 SSs at 23844 damage each, that equates to 443170 PS damage, which averages out to 14772 DPS over the 30s, not the 4924 DPS reported, which is 1/3 the amount. Now of course, some of the PS damage was lost at the end, but I do not think FD models that, and I do not think that almost 300K (about 68%) of the PS damage was still left to tick off at the end of the period.

For a whole 7.5 min run, 1555 DPS from PS is really low. Just the 44 AIs alone and without factoring in the CA phase all crits results in 1424 DPS from PS. Also not including the CA phase, the 152 SSs result in about 750 DPS. The 43 CSs result in about 1484 DPS from PS. I get a total of 3658 PS DPS from my calculations with not including the CA benefits.

2) Rapid Recuperation not incorporated correctly

It does not appear that RR is being added correctly. At the 5111 haste rating in the Sim, that is 12.026% haste from gear resulting in a 4.481 focus regen. FD shows this value correctly. RF is 40% haste, resulting in a regen of 6.273 for those without RR. With RR providing 12 focus every 3 secs, that is equivalent to 4 FPS, resulting in 10.273 (which is consistent with Cata), but FD shows 8.273. It appears that FD is adding in only half the amount. MoP Wowhead still shows me the 12 focus every 3s, so unless this ability has been nerfed (and WoWhead is not showing the nerf) to 6 focus every 3 secs, FD is missing about half the focus from RR.

3) Aspect of the Iron Hawk not implemented

Both WoWhead and the WoW talent calculators show Aspect of the Iron Hawk also providing 10% AP. The ability is red on the FD talent calculator showing no DPS effect, and when selected or unselected, the DPS does not change. Obviously, this affects all of the hunter specs.

4)Should not stack RF and BL

The Sim starts not only with the back to back RFs but also Heroism. This is not ideal and ends up being a lot of wasted haste for MM since SS cast time is at 0.713s and AI cast time is at 1.034s. Just like in Cata, ideally MMs do not want to stack RFs with BL. If the raid is starting the fight with BL, then start the RFs after. Here are some improvements with moving the BL to different locations.

a) After the RFs are done at about 32s (7% of the fight) in: 96167, +232 DPS
- In this case, there is a few second overlap with RF, but it maximizes hasted CA time.
B) After the RFs are done at about 36s (8% of the fight) in: 95608, -336 DPS
- Not as good since part of the CA phase was unhasted
c) At the start of the KS phase: 95136, -808 DPS
- Unlike in Cata, BL in this phase is not as useful since we are only casting 2 SS instead of 4 and since with all of the focus free instant cast KSs and other abilities like GT, we have less cast time shots affected by haste.

Note that for a shorter fight where the CA phase is fully covered by RF that BL is fine at any other time in the fight other than the KS phase. It is only where the CA phase is longer than the 30s back to back RF duration that it is ideal to chain BL after (or vice versa) to maximize hasted CA time.

5) Unideal CA phase shot priority

When looking at the CA phase only of 30s duration, a few items are not ideal with the starting rotation, these include:

a) Starting with 2 SSs to get the Steady Focus buff. This wastes all of the focus generated by the 2 SSs - which is roughly 37 focus. Furthermore, after the 2 SSs all the sim uses is all instant cast abilities that do not take advantage of the SF, RF-Stampede-MoC-GT-DB-Readiness-GT-DB. This requires a new SS pair to be cast to refresh SF once all these instant abilities are done. In fact, SF actually falls off briefly during the second SS cast to refresh it. The only items that take advantage of the additional haste from the SF are your autoshots and Stampede. The first AI cast is not until 13s into the CA phase. Only 8 AIs are used the whole time, with 2 of them being instant cast MMM AIs.

Personally, if it is best to blow all those instant cast CDs at the start, I would not bother casting the first SS pair and Stampede until after since Stampede is not affected by Readiness and since the full 20s will still be affected by RF haste. However, I could not figure out anyway to delay the casts of the SSs initially in FD without changing ISS behavior to Standard Priority, which is a bad option since it does not guarantee SS pairs to get SF.

B) Other focus is being wasted as the start as well, even with delaying the first 2 SSs casts. RF is cast increasing our regen, but then the sim casts Stampede and DB, both which cost no focus. Thus, we lose the 16.5 (current FD) to 20.5 focus regened over those casts (not including the regen over the latency). Hence, I suggest having MoC cast before RF, Stampede, and DB so the focus regen will be better used.

Making this change in the Sim has no effect for this CA test case, but it makes sense.

c) Currently, after casting RF, 7 instant abilities are cast that are not affected by RF, except for Stampede. These were listed previously in a). To better take advantage of both RF and Readiness, if it is worthwhile to use all of these abilities, then I suggest moving the MoC, DB, and GT initial casts before the RF cast.

Thus, the order changes from: RF-Stampede-MoC-GT-DB-Readiness-GT-DB
To: MoC-GT-DB-RF-Stampede-Readiness-GT-DB

This can allow RF to affect more cast time shots. Making this change in the sim resulted in no change in DPS since the same shots were cast; however, RF still had additional time left after the end of the 30s simulation, so making this change in the CA phase should be beneficial.

One thing not considered here is trinket uses/procs and other buffs like Blood Fury stacking and their durations. The ideal order will be affect some when factoring these in.

d) Just like we defer casting SrS and CS during the CA phase, we need to evaluate whether it is beneficial to cast these other abilities during the CA phase. Below is an analysis of each of the abilities currently being used in the CA phase in the sim. Also note that each ability used other than AIs and SSs also slightly reduce the chance for MMM AI Procs.

- Stampede: Not necessary to use during the CA phase from the benefits of CA, but still beneficial to use due to the haste from RF and stacking of trinket procs and racials like Blood Fury. Although with costing no focus, it does not necessarily need to be cast right away but can wait until after some focus has been spent.
- GT: 121328 at a 15 focus cost and GCD
- MoC: 458277 at 60 focus cost and a GCD
- DB: 148849 at 0 focus cost, a GCD, but some focus regen.
- AI: 154263 *1.3 PS = 200542 at 50 focus and a 1.345 cast time with SF and RF or a GCD if a MMM AI proc.

I would contend that GT probably is not beneficial over the CA phase since its damage is considerably lower than AIs and since it costs focus. Note that if cast GT is beneficial, then casting CS would be beneficial to since its damage when factoring in the SrS refresh is much higher. The 1 cast of MoC is beneficial. The DB casts are probably beneficial as well since they do high damage, cost no focus, and regen focus, allowing more AIs and less SSs.

The initial DPS with all of these abilities included is 172365.

If I exclude GT, MoC, and DB from the CA phase, I get 146013 DPS for a loss of 26351 DPS in the Sim. Note that 2 more AIs and 4 more SSs were cast, so with fixing the PS problem, the DPS of this case will be higher.

If I add back in MoC, I get 157540 DPS for a loss of 14825 DPS. 1 more AI and 4 SSs are cast, so better PS modeling will help this case.

If I also then add back in BD, I get 164578 DPS for a loss of 7787 DPS, but with 1 more AI and only 1 more SS due to the additional focus. PS alone fixes probably cannot make up that difference, but is with fixing RR for an additional 60 focus and 1 more AI should make not using GT in the CA phase a good choice.

e) Since GT is close to being beneficial during the CA phase, I decided to check SrS and CS. Adding them to the priority with no GT was a +1148 DPS gain over not having CS and SrS.

With GT included, it was only a 3 DPS gain, but with PS being calculated correctly, it would be an even higher gain.

Will have to reevaluate after the PS and RR fixes are done, but I assume that not including GT during CA would be ideal and that possibly including CS and SrS may be beneficial.

f) Anyway it would be good to possibly include some additional disable while CA is active options for GT and DB. I assume that DB will always be good, but might as well make an option for it to be safe. To be complete, should also probably have options for Blink Strike, Lynx Rush, Powershot, and Barrage in case those spec abilities are chosen. Might as well add an option for MoC as well.

6) Move RF Lower in the Priority

Moving RF to just before Readiness in the priority list is a small 9 DPS gain. Nothing big.

7) CS wait time is too large

Waiting a whole 0.5s to cast CS is too large.That is half a GCD. Considering all of the abilities that we have that do comparable damage to CS, waiting that long can be a huge DPS loss. I suggest halfing the amount to 0.2 to 0.3s. This is what I did in Cata. FD seems to favor this with about 180 DPS gain.

8) Stat priorites

Since abilities are still being modified, it is hard to exactly determine which stats are best, especially what the ideal amount of haste is. Since it seems that initially in MoP that an AS rotation is the best option and that somewhere between a 2 SS and 3 SS rotation is focused balanced. This information kind of drives the ideal haste amount (not considering latency). Several factors to consider:

a) A 3 SS cycle is tight with a 1.33s cast time. This requires an unreasonably high 18.58% haste from gear in early MoP.
B) A 2 SS cycle is tight with a 1.5s cast time. This requires an easy to achieve 5.4% haste from gear.
c) During RFs, only 12.93% haste from gear is required for your SSs to be capped. This haste level results in a 1.4s SS cast time when RF is not active.

Thus, for early MoP, I suggest somewhere between 5.4% and 12.93% haste from gear trying to get closer to 12.93% as possible without sacrificing too much crit or mastery. This means we should shot for between 2295 and 5495 haste rating. The 5111 in the Sim falls in this range. With the stats available in the T14H gear set, it appears that getting to the 5495 amount is reasonable and a DPS increase, even with giving up some crit to achieve it. Making one reforge change to get to 5484 haste rating increased DPS by an additional 111 DPS with the lower CS wait time. However, reforging more crit to haste after that is a DPS loss.

#3 Rivkah

Rivkah

    Great Tiger

  • Members
  • 838 posts

Posted 19 August 2012 - 12:37 AM

I finally had a chance to look at your MM simulation. The initial DPS was 96992.

Here are some comments I have for increasing the MM numbers or possible bugs that need to be fixed.

1) PS not being calculated correctly:


This was definitely a bug- it was using the old single talent point value for piercing shots which was bringing it down a lot. I fixed it and that should be far more correct (right now it just tracks the number of crits of the appropriate shots and takes 30% of the crit damage and adds it up, but because of the bug it was only taking 10%).

2) Rapid Recuperation not incorporated correctly


This was the same bug as piercing shots and it's been fixed. It seems I missed a few corrections to these talents that had multiple talent points when I converted to the new talent system. I did a look through to make sure I hadn't missed any others and I couldn't find them, but if you notice anything else strange let me know.

3) Aspect of the Iron Hawk not implemented

Both WoWhead and the WoW talent calculators show Aspect of the Iron Hawk also providing 10% AP. The ability is red on the FD talent calculator showing no DPS effect, and when selected or unselected, the DPS does not change. Obviously, this affects all of the hunter specs.


Iron Hawk just replaces Hawk, which has the exact same 10% AP bonus. So it isn't really a damage talent, it just gives you damage reduction while in hawk. I've verified this in game- are you seeing something different?

4)Should not stack RF and BL

The Sim starts not only with the back to back RFs but also Heroism. This is not ideal and ends up being a lot of wasted haste for MM since SS cast time is at 0.713s and AI cast time is at 1.034s. Just like in Cata, ideally MMs do not want to stack RFs with BL. If the raid is starting the fight with BL, then start the RFs after.


I do have 2 options in the settings which somewhat give you control over this- there is a minimum focus shot speed to activate haste cooldowns, which prevents additional haste cooldowns from being cast until the speed passes that limit. If you have ideas about better ways to give control over this without making the options too complicated that'd be helpful. When I use that setting, it seems to reduce dps, but that may be due to less than ideal shot choices (or due to delays in readiness use).

There is also the option to set what time into the fight to use heroism. There isn't really a good way to control right now what priority to use for haste cooldowns- I'm not sure if this is worth putting in but I've been considering it.

5) Unideal CA phase shot priority

When looking at the CA phase only of 30s duration, a few items are not ideal with the starting rotation, these include:

a) Starting with 2 SSs to get the Steady Focus buff. This wastes all of the focus generated by the 2 SSs - which is roughly 37 focus. Furthermore, after the 2 SSs all the sim uses is all instant cast abilities that do not take advantage of the SF, RF-Stampede-MoC-GT-DB-Readiness-GT-DB. This requires a new SS pair to be cast to refresh SF once all these instant abilities are done. In fact, SF actually falls off briefly during the second SS cast to refresh it. The only items that take advantage of the additional haste from the SF are your autoshots and Stampede. The first AI cast is not until 13s into the CA phase. Only 8 AIs are used the whole time, with 2 of them being instant cast MMM AIs.

Personally, if it is best to blow all those instant cast CDs at the start, I would not bother casting the first SS pair and Stampede until after since Stampede is not affected by Readiness and since the full 20s will still be affected by RF haste. However, I could not figure out anyway to delay the casts of the SSs initially in FD without changing ISS behavior to Standard Priority, which is a bad option since it does not guarantee SS pairs to get SF.


I'll try to think of a way to give more control over the steady focus behavior without overwhelming the settings. With regards to some of the other long CD abilities, one issue is that things like stampede you will want to cast early if you're prepotting so you can get the benefit from the prepot on it, so I expect often hunters will want to cast it very early on. We also will want stampede to be during RF to benefit from the extra pet regen which may help us get an extra wild hunt or basic attack. I'm not sure how best to provide control over the use of those abilities.

You can move MoC above Stampede and it will cast first (although there is still the SF issue). Right now the default shot priority is somewhat loose with the talent shots due to the way I set it up, but I'll work on improving that once we're clearer on our real priorities.

c) Currently, after casting RF, 7 instant abilities are cast that are not affected by RF, except for Stampede. These were listed previously in a). To better take advantage of both RF and Readiness, if it is worthwhile to use all of these abilities, then I suggest moving the MoC, DB, and GT initial casts before the RF cast.


I'm assuming this is primarily a suggestion for the profile and not the site. If you want to optimize the MM profile setting and give me a link to the new version, I'd be happy to use it- I haven't spent much time on optimization of it. Make sure to grab the latest one I'm using though (Zeherah's Hunter DPS Analyzer) as it's been tweaked a bit.

d) Just like we defer casting SrS and CS during the CA phase, we need to evaluate whether it is beneficial to cast these other abilities during the CA phase. Below is an analysis of each of the abilities currently being used in the CA phase in the sim. Also note that each ability used other than AIs and SSs also slightly reduce the chance for MMM AI Procs.


I'll take a look at what abilities might make sense to disable during CA and see how to add options for it without overwhelming the user.

7) CS wait time is too large


I'm happy to change the default value for this if we can figure out what's a reasonably appropriate number for all the specs. I haven't spent a lot of time analyzing it. People can always change the value but I'd like the default to be good.

Thanks for all the feedback and helping me find the bugs. There's a lot of tweaks I'd like to do to the code and settings yet but it's been challenging finding time to work on it, so it's extremely helpful to have detailed feedback.

#4 Whitefyst

Whitefyst

    Great Tiger

  • Members
  • 771 posts

Posted 19 August 2012 - 02:15 AM

Thanks for the quick fixes. I will check things out later.

1) and 2). Glad to hearthese were easy fixes.

3) Nope, you have it right. That was just my misunderstanding since I am not on the PTR. I was just reading the talent out of the replaces Hawk context.

4) Yeah, I use the BL setting all the time. The BL timing comment was only in regards that we do not currently have the optimal setting. I have not tried the other setting for a while and kind of forgot about it. I will check it out again.

5a) Good point on the prepot.

5c) Yes that was a suggestion for the profile setting.

7) The suggestion on the wait time was for the Max MM DPS profile only. Since we are reporting numbers comparing the specs, It helps to make sure that we have as close to max as possible for each spec. The previous numbers had MM way lower, which was very depressing for me. So I am hoping to find some changes to the profile settings to improve the numbers.

Once again, thanks for the great work on this wonderful tool.

#5 Whitefyst

Whitefyst

    Great Tiger

  • Members
  • 771 posts

Posted 19 August 2012 - 05:29 AM

Both PS and RR look much better.

I also see that the T14 4P is implemented too. That threw me off at first since I do not think it was showing up earlier today. This obviously affects the required haste amounts. For SS to be GCD capped during RF only requires 3.896% haste (or 1656 haste rating) from gear.

I also see that there is more difference between the ISS behavior options for the <= 4s and <=3 cases. The 3s setting starts the fight with a SS pair, but the 4s setting does not, which is more desirable and a 1052 DPS gain. However, this option does allow for the first AIs to be cast before SSs. I would think that we would want the SSs first since the 15% to 25% haste buff really speeds up the AI cast times (by about .335s each of both AIs in this case).

I am also feeling more and more confident that we will want to disable GT during the CA phase (unless there are multiple targets).

#6 Rivkah

Rivkah

    Great Tiger

  • Members
  • 838 posts

Posted 19 August 2012 - 05:49 AM

I also see that the T14 4P is implemented too. That threw me off at first since I do not think it was showing up earlier today. This obviously affects the required haste amounts. For SS to be GCD capped during RF only requires 3.896% haste (or 1656 haste rating) from gear.


I don't believe I changed anything in the 4pc modeling- did you maybe test with a gear set that wasn't wearing the 4pc?

I did add the 4pc T14 to my beta haste calculator today, which btw should be up to date if you find it useful to figure out specific haste points.

I also see that there is more difference between the ISS behavior options for the <= 4s and <=3 cases. The 3s setting starts the fight with a SS pair, but the 4s setting does not, which is more desirable and a 1052 DPS gain. However, this option does allow for the first AIs to be cast before SSs. I would think that we would want the SSs first since the 15% to 25% haste buff really speeds up the AI cast times (by about .335s each of both AIs in this case).


Can you think of a good logical way to give control over the ISS rule? Do we want to disable the rule entirely during careful aim or would that be too much? Maybe disable it if certain haste buffs are active, but that could encompass a lot. Or I could maybe have an additional option that applies the rule only if focus is below a certain amount? Or I could just delay applying that rule in the first x seconds of the fight. Any ideas that don't greatly overcomplicate things? MM already has more options than any other spec and I suspect the list will get longer since it's so special.

#7 Nooska

Nooska

    King Hippo

  • Members
  • 540 posts

Posted 19 August 2012 - 07:57 AM

On the wait time for the big shots, for BM I've had increses (of varying degrees) down to 0.3 seconds (at the standard 150 ms latency) and 0.2 seconds with 25 ms latency (the latency I have real world) - 0.2 has shifted a bit up and down with hastevalues though (I assume because of losing or gaining 1 shot somewhere from the combination of 0.1 second wait differences)
0.3 is the ideal I'll be recommending for BM, haven't seen that be anything but a gain with the same variation of haste values.

#8 Whitefyst

Whitefyst

    Great Tiger

  • Members
  • 771 posts

Posted 19 August 2012 - 12:59 PM

Can you think of a good logical way to give control over the ISS rule? Do we want to disable the rule entirely during careful aim or would that be too much? Maybe disable it if certain haste buffs are active, but that could encompass a lot. Or I could maybe have an additional option that applies the rule only if focus is below a certain amount? Or I could just delay applying that rule in the first x seconds of the fight. Any ideas that don't greatly overcomplicate things? MM already has more options than any other spec and I suspect the list will get longer since it's so special.


I cannot think of a good set of ISS rules to start the fight since there are so many variables in play, especially with being able to chose from so many abilities now to use or skip during CA phase. The Sim I did started with:

MoC
DB
GT (would want to disable this in CA phase)
RF
Stampede
Readiness
DB
GT (would want to disable this in CA phase)

After this there was 97.56 focus. So the decision to perform at least one AI first is a good one and what I would do in game since we are hasted with RF.

Now that I analyze the situation more closely with taking into account the much higher focus regen with DB going than what I was used to in Cata, doing the 2 AIs before the first SS is the best thing to do since we do have the AIs still hasted by RF and we would waste about 63 focus if cast the 2 SSs first and would still waste about 14 focus if only cast 1 AI first. We definitely should cast 1 AI first, but it is debatable whether casting the second or not is most ideal. As far as modeling goes, I would not worry about it.

If the 2 GTs are disabled, it is about a wash focus-wise since the 30 focus that is no longer is spent is roughly counteracted by the focus regens over their GCDs.

#9 Rivkah

Rivkah

    Great Tiger

  • Members
  • 838 posts

Posted 20 August 2012 - 05:24 AM

I added a few new options to help with the simulation:
- You can now turn off focus savings for murder. This isn't recommended unless you also don't have focus saving for your money shot enabled and you have a min focus set for arcane shot, as you might never end up casting murder.
- You can disable glaive toss during careful aim (I also moved the arcane shot setting for this to the MM only section)
- You can now set the number of stacks to use focus fire at. I doubt it'll become profitable to use it below the max, but I was cleaning up some focus fire code anyhow and it was easy to add this option

I also added in estimated damage reductions from parry to Lynx Rush (and to other abilities your pet might use during Lynx Rush). For now this also applies to Kill Command (currently it just applies a flat reduction to all pet abilities assuming 50% chance of parries for the uptime of lynx rush), although it occurs to me that you probably would try to time lynx rush between kill commands to avoid the risk of a parry, so I may take that out later, or switch to actually tracking attack counts during lynx rush. I still have a faint hope that maybe they'll realize it's silly to have a pet ability which forces the pet to attack from the front randomly and change it, so I don't want to go all out in the implementation yet, but I doubt they'll change it.

#10 Whitefyst

Whitefyst

    Great Tiger

  • Members
  • 771 posts

Posted 20 August 2012 - 03:39 PM

I added a few new options to help with the simulation:
- You can disable glaive toss during careful aim (I also moved the arcane shot setting for this to the MM only section)


Thanks for adding that option. As anticipated, disabling it during the CA phase is a DPS benefit by about 250 DPS in the case I did.

However, I thought that after disabling GT that using SrS/CS instead would be even a greater DPS benefit. This is because CS itself has damage consistent with GT, but with adding in the SrS damage and the PS damage is much higher. Although it is slightly better than using GT during CA phase, it is considerably worse than not using either. So I would only recommend using CS during the CA phase on fights where the self-healing over that first 10% of boss health is needed.

Due to DB's focus gains and high deamage, it is definitely still useful to use during the CA phase since it gives more focus for more AIs. Also a single cast gains about twice as much focus than an SS cast does and still does much more damage than an SS. Even with 100% crit and PS damage for the SS, DB does over 4x the damage in FD. Thus, I see no need to add in an option to possibly disable DB during the CA phase as I had mentioned as a possibility previously just to have the option available for messing around. We are going to want to use DB as much as possibly since it is essentially is a direct replacement for extra SSs that do not maintain SF with doing both more damage and providing more focus than that SS.

In fact, I would almost put DB higher on the priority list than CS but still is better to have CS higher on the priority list if CS and Db are up at the same time due to the CS cast clearing focus space for the full regen from DB, while with the other order, there is a greater possibility for focus wasting.

#11 Lokrick

Lokrick

    Piston Honda

  • Members
  • 121 posts

Posted 20 August 2012 - 11:31 PM

...I also added in estimated damage reductions from parry to Lynx Rush (and to other abilities your pet might use during Lynx Rush). For now this also applies to Kill Command (currently it just applies a flat reduction to all pet abilities assuming 50% chance of parries for the uptime of lynx rush), although it occurs to me that you probably would try to time lynx rush between kill commands to avoid the risk of a parry, so I may take that out later, or switch to actually tracking attack counts during lynx rush. I still have a faint hope that maybe they'll realize it's silly to have a pet ability which forces the pet to attack from the front randomly and change it, so I don't want to go all out in the implementation yet, but I doubt they'll change it.


I implemented the actual pet movement to get a more direct simulation of the dps impact of the pet moving. In this implementation, on each lynx rush bite, there's a 50/50 chance whether the pet is in the front (and thus subject to parry and block) or the back. These sims were run against T14H BM; since they only affect pet damage, they are a pretty good estimate of the impact. Frenzy uptime was 81.6% rather than 81.7%; there was no noticeable dps impact other than the pet dps below.

Character | | Total | Hit | Crit | Count | Crit% | Avoid% | Glance % | Block %
cat 52,811 dmg| claw | 4,983,386 | 24,120 | 48,820 | 130.4 | 57.10% | 0.00% | 0.00% | 0.00%
| kill_command | 7,898,669 | 87,405 | 177,711 | 63.8 | 40.30% | 0.00% | 0.00% | 0.00%
| lynx_rush | 2,621,825 | 31,613 | 66,531 | 57 | 41.10% | 0.00% | 0.00% | 0.00%
| melee | 8,276,836 | 14,975 | 30,579 | 406.2 | 40.20% | 0.00% | 24.00% | 0.00%
| | | | | | | | |
cat 50,999 dmg | claw | 4,795,326 | 24,167 | 48,874 | 130.1 | 55.00% | 3.80% | 0.00% | 1.00%
| kill_command | 7,576,868 | 87,523 | 177,324 | 63.8 | 38.50% | 3.70% | 0.00% | 1.40%
| lynx_rush | 2,517,949 | 31,587 | 66,533 | 57 | 39.40% | 3.60% | 0.00% | 1.40%
| melee | 8,025,650 | 14,963 | 30,605 | 405.4 | 40.00% | 3.70% | 24.00% | 0.70%
| | | | | | | | |
diff | claw | 188,060 | 3.8% | | | | | |
| kill_command | 321,801 | 4.1% | | | | | |
| lynx_rush | 103,876 | 4.0% | | | | | |
| melee | 251,186 | 3.0% | | | | | |


Note the reduced crit. My hypothesis is that it's because every other lynx rush correlates with heavy agility procs. Thus those misses for kill command and base attack are more significant than misses outside the lynx rush time.

EDIT: These numbers seem high, so I'll look at what's going on in a log. I suspect that the pet is not getting back behind the boss.

#12 Rivkah

Rivkah

    Great Tiger

  • Members
  • 838 posts

Posted 21 August 2012 - 03:43 AM

It was pointed out to me that focus fire wasn't properly costing a GCD. I've fixed this, and while I was in there I also noticed it wasn't properly multiplying the pet regen by the number of stacks. So it may be a net dps gain in the sim, but I figured I should bring it up so people understood where the change came from.

Because it's now on the GCD, I have changed Focus Fire so it is no longer automatically cast and instead is a selection in the shot priority. This means that any saved setting that needs to have focus fire will need to be manually updated to include it (and you may need to refresh your browser cache to get the shot to highlight). I also removed the option to disable focus fire, since now just leaving it out of the shot priority does the trick.

#13 Whitefyst

Whitefyst

    Great Tiger

  • Members
  • 771 posts

Posted 21 August 2012 - 05:09 AM

Did you change anything else? I do not see anything else in the version history.

I had a setting saved from just a couple hours ago today that I have reloaded several times all with the same result of 101397. I just reloaded it, and it now provides 102827 DPS for a 1430 DPS gain. Unfortunately I did not save any copies of the shot totals or anything from the previous loads to compare against the current value to see what results changed.

The settings look to be all the same with no values in the Custom Gear Stats.

I do see that in the Stats From Gear table that the AP value is blank. I do not recall it being blank previously.

#14 Rivkah

Rivkah

    Great Tiger

  • Members
  • 838 posts

Posted 21 August 2012 - 05:25 AM

Did you change anything else? I do not see anything else in the version history.

I had a setting saved from just a couple hours ago today that I have reloaded several times all with the same result of 101397. I just reloaded it, and it now provides 102827 DPS for a 1430 DPS gain. Unfortunately I did not save any copies of the shot totals or anything from the previous loads to compare against the current value to see what results changed.

The settings look to be all the same with no values in the Custom Gear Stats.

I do see that in the Stats From Gear table that the AP value is blank. I do not recall it being blank previously.


Actually it's normal for stats from gear AP to be blank. The only actual AP we usually get from gear is the leg patch and the new MoP leg patches don't have AP anymore. It's only showing you AP actually on the gear, not calculated AP from the agility on gear, so I don't think that's likely to have changed for you.

You may have caught me in the midst of another update- I did 2 more changes after my original update (they're in the version history but might not have been yet when you looked). The first one was the one that probably changed your dps- I retested Dire Beast today and verified that it is scaling with haste (only melee haste like pets, which means it doesn't benefit from rapid fire, focus fire or steady focus), which it wasn't when I tested it quite awhile ago. So I modified the dps calculation to reflect that, which gave it a big boost. It's overcalculating it a bit right now because I'm not accounting for dire beast lasting past the end of the fight, but I'll go back and address that later when I have more time.

The second change I made won't affect dps at all, but it will be useful for people doing a lot of tests on my site- I added a setting which allows you to turn off the stat valuation calculations (i.e. the dps per stat). This is the slowest part of the dps calculation, so by turning this off you will speed the time to get your results considerably. This setting is stored as a cookie, rather than with your saved settings, so it will persist until you turn it off or back on again.

#15 Lokrick

Lokrick

    Piston Honda

  • Members
  • 121 posts

Posted 22 August 2012 - 12:25 AM

I implemented the actual pet movement to get a more direct simulation of the dps impact of the pet moving. In this implementation, on each lynx rush bite, there's a 50/50 chance whether the pet is in the front (and thus subject to parry and block) or the back. These sims were run against T14H BM; since they only affect pet damage, they are a pretty good estimate of the impact. Frenzy uptime was 81.6% rather than 81.7%; there was no noticeable dps impact other than the pet dps below.


After fixing/working around a bug that was leaving the pet in front of the boss, the avoidance and block rates were 0.3% and 0.1% respectively for kill command and basic attack (still 3.6 for lynx rush). The actual dps effect was in the noise except for 2-300 dps difference for lynx rush.

#16 Whitefyst

Whitefyst

    Great Tiger

  • Members
  • 771 posts

Posted 22 August 2012 - 05:16 AM

With the shot nunmbers that FD provides, I just completed an updated analysis in T14 heroic gear for the tipping point where casting AI as the focus dump is better than AS. I did the analysis with the following formula:

AI + WQ = (x * AS) + (1.25 * auto) + ( (1.25 + x) * WQ)

where:
AI is AI average damage with including PS.
WQ is average WQ damage per shot taken that can proc it.
AS is AS average damage.
auto is average autoshot damage.
x is the AI cast time and fraction of ASs performed over that cast time

On the left side is 1 AI cast and the average WQ damage.
On the right side is the damage from the fractional number of ASs that can be performed during the AI cast time plus the damage from 1.25 autoshots being suppressed during the AI cast plus the average WQ damage from the ASs and autoshot.

This resulted in a 1.424s AI cast to reach the tipping point when assuming 1.25 autoshots are suppressed per AI cast.

If going with the about the best case scenario of only exactly 1 auto being suppressed during an AI cast, the result is a 1.597s AI cast time.

The haste requirements to make it a DPS gain to cast AI as the focus dump instead of AS for these two cases is shown in the following table:

|Normal|T14 4P|Normal*|T14 4P*
AI Cast Time|1.424|1.424|1.597|1.597
Bloodlust|23.84%|13.93%|10.42%|1.59%
Rapid Fire|14.99%|5.79%|2.54%|0%

* Assuming exactly 1 auto suppressed per AI cast as opposed to 1.25 autos.

The following conclusions can be drawn from this:

1) MMs definitely want to be using AS as their focus dump when unhasted.
2) The AI cast time at which AI becomes better to use as the focus dump is somewhere between 1.424s and 1.597s. I suggest using either 1.6s or 1.5s as the new default value for maximum speed to cast AI setting.
3) Without the T14 4P haste bonus, it may not be beneficial to cast AI during BLs. It requires at least 10.42% (when assuming exactly 1 autoshot is suppressed during the AI cast) and possibly up to 23.84% haste from gear (when assumming 1.25 autoshots are suppressed per AI.
4) With the T14 4P, it is advised to cast AI during BL with the recommended haste amounts.

Is anyone else seeing a similar tipping point?

What is a good estimate to use for the number of autoshots suppressed during an AI?

EDIT:

I did some tests in FD with a 1 min test case during the Standard phase with BL active through most of it at the 5442 haste rating (12.80% haste) from gear that I had in my quick optimal case. AI cast time in this condition is 1.438s. I lowered the maximum cast time for AI to 1.4s, and DPS increased by over 500 DPS. Hence, in FD, it appears that a 1.4s cutoff may be better and that AI should only be cast during RFs or BL plus some haste proc.

During the CA phase with 100% AI crit chance, then it is always better to use AI as the focus dump, even with no haste on gear and no dynamic haste effects.

Also latency affects this number some since it affects the number of ASs that can be performed during an AI cast.

#17 Nerec

Nerec

    Von Kaiser

  • Members
  • 39 posts

Posted 23 August 2012 - 10:59 AM

What is a good estimate to use for the number of autoshots suppressed during an AI?


Don't pin my down on that, but I tested the behavior of slam on an warrior alt on live servers few weeks ago and get a positive reponse from a gildmate who didn't believed it. Slam dont interrupt the swingtimer, but rather just suspend your current swing for the cast and resume to the exact % of the progressed swing. Can't check it right know, but if Aimed use the excact same behavior, to rather suspend the swing/autoshot than interrupt it, with a normalized weapon speed (2.9/3.0) you will be able to math out something like:

  • haste * weaponspeed = time between autoshots
  • casttime(Aimed) * time between autoshot = lost autoshots
  • lost autoshot * ~dmg of autoshot = lost dmg due to casted aimed
  • etc..

I know this is a big "IF", but can't test it right know, sorry!

#18 Rivkah

Rivkah

    Great Tiger

  • Members
  • 838 posts

Posted 23 August 2012 - 04:00 PM

It looks like the formula I'm currently using to simulate autoshot suppression (which I assume is based on aimed shot mechanics we discussed during Cata) is:
(Average Aimed Shot Shot Cast Time / Weapon Speed + .5) * Aimed Shot Count = Autoshots lost

I'm not sure if that's really the best formula, so I'm open to suggestions.

#19 Whitefyst

Whitefyst

    Great Tiger

  • Members
  • 771 posts

Posted 23 August 2012 - 04:31 PM

Don't pin my down on that, but I tested the behavior of slam on an warrior alt on live servers few weeks ago and get a positive reponse from a gildmate who didn't believed it. Slam dont interrupt the swingtimer, but rather just suspend your current swing for the cast and resume to the exact % of the progressed swing. Can't check it right know, but if Aimed use the excact same behavior, to rather suspend the swing/autoshot than interrupt it, with a normalized weapon speed (2.9/3.0) you will be able to math out something like:

[*] haste * weaponspeed = time between autoshots
[*] casttime(Aimed) * time between autoshot = lost autoshots
[*] lost autoshot * ~dmg of autoshot = lost dmg due to casted aimed
[*] etc..

I know this is a big "IF", but can't test it right know, sorry!


Thanks. I have been doing some analysis with assuming 1 autoshot lost but with using 1.25 to be conservative, especially since 1/4 of an autoshot and its expected WQ damage does not change the analysis too much. I have been suing both values to provide kind of a range.

#20 Rivkah

Rivkah

    Great Tiger

  • Members
  • 838 posts

Posted 26 August 2012 - 10:48 PM

Just an FYI for folks who use both my site and simulationcraft- you can now get a simc profile with most of your setting data (excluding action lists and buffs) from your dps output. If you enable the debug data option the simc profile is displayed below your simulated shot breakdown. This should hopefully make it easier to transfer settings into simc (I may consider adding the ability to parse a simc profile in a future release, but that'd be a lot harder to implement).




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users