However it would be nice if the 23_5_43 profile could include poison_doses in it's priority list. I'm in no way certain poison_doses is the correct parameter to check, but you'd be monitoring the amount of doses on the target when playing the spec. The priority list should reflect this and not only check if the Envenom buff is already up.
Not actually sure this is needed. At least for Mutilate, Envenom > Eviscerate even with only 1 dose of DP on the target, and that's after accounting for the loss of DP damage. I don't see why the same wouldn't hold true for NHaT as well, since it too takes full imp and vile poisons.
Good point, remember going over that math once actually...
From what I can see 23_5_43 is currently at some 80% uptime on the Envenom buff. Using more Envenoms (even low poison dose Envenoms) will naturally change this uptime. I guess we just have to check if overlapping the Envenom buff (which was only a risk for mutilate during the beta days of pve deadly brew) with just one or 2 seconds (the opportunity cost) decreases the value of succesive Envenoms enough for Eviscerate to come out on top?
If not we only need to add a check the use of Eviscerate in the priority list, so Eviscerate is only used if there's one or more Deadly Poison doses on the target.
Last edited by bural : 03/14/09 at 9:00 AM.
Reason: additional comment about the priority list
Good point, remember going over that math once actually...
From what I can see 23_5_43 is currently at some 80% uptime on the Envenom buff. Using more Envenoms (even low poison dose Envenoms) will naturally change this uptime. The question is if overlapping the Envenom buff (which was only a risk for mutilate during the beta days of pve deadly brew) with just one or 2 seconds (the opportunity cost) decreases the value of succesive Envenoms enough for Eviscerate to come out on top?
You can use "env>=" and "env<=" conditionals to make decisions based upon how much time is left on the Envenom buff......
Right got hold on a windows machine to run it on.
First of all sorry for the clumsy way of linking it, it's not my pc and had to do a fix or two to get the results below.
Lining up nHAT:
I tried with other priority queues including low combo point Envenom finishers depending on the remaining time of the Envenom buff and some queues with Shiv over Hemo. They all resulted in lower dps.
I realised the nvHAT build (in lack of a better shorthand) didn't include Imp. Evis, so created another profile with the following build Talent Calculator - World of Warcraft
Envenom seems to be taxed too much by low poison doses to allow it to effectively be spammed.
The build does seem to spend a somewhat high amount of time at the energy cap (4%) so i played around with the overlap time on successive Envenoms. Increasing this from 1 to 3 seconds decreases the output by some 50-80dps while bringing up the Envenom buff uptime to some 93% (nearly as much as nHAT) and completely removing the time spend at the energy cap. It might be worth checking if swapping a talent point from Lethaligy to Vigor is worth it for a build with "env<=1". If this turns out to work on live servers it'll be TBC Find Weakness management all over again. Maintaining a buff with a max duration of 6 seconds could prove to be hard (considering snd duration is down by 50% compared to vHAT) with varying HAT procs, but even dropping envenom entirely from the rotation is a dps upgrade over 8 20 43. Thus doing a castsequence macro with as much as 4 (depending on your group, use the simulator to determin a better ratio - prolly something close to 1:2 ratio) eviscerates and then envenom could prove to maintain decent uptime on the envenom buff while being fairly easy to play. Anyway this aspect of the discussion belongs in the "Free HAT: Deep Sub"-thread really.
I didn't change gear setup for comparison purposes, but it should be noted that swapping SR for WD increased the dps of the above nvHAT profile to 7705 dps. The effect of of "env<=1" over "env<=3" was furthermore diminished alot when changing to the faster offhand.
So, as for SimulationCraft we should prolly check to see if Envenom damage is calculated correctly. Secondly the 23_05_43 build should be changed to include Imp. Eviscerate by taking points from Lethaligy and the action queue should thirdly be changed to prioritise keeping the Envenom buff going while Eviscerating in between. Last but not least should the offhand weapon be changed to a 156.4dps, 1.4 speed dagger.
Hope I didnt hack around enough to screw something up, so please check the results accordingly.
Last edited by bural : 03/16/09 at 9:36 AM.
Reason: edit1 final comment, edit2 syntax, edit3 comments about the energy cap, edit4 Webbed Death>SR
Did you take into account the variability in the rate of HaT procs? Also do you have a way of measuring the number of HaT procs per second?
When simulating a Rogue outside of a party environment (Rogue_T8.simcraft or Rogue_23_05_43.simcraft), one must specify the average interval between HAT procs. This is just an average. The gap between two procs is a random number between 0.5*interval and 1.5*interval.
When simulating a full raid (see Raid_T8.simcraft) turning the interval option off will allow HAT procs to come from party members. The frequency of this occurrence is part of the "Proc" report at the end of simulation.
There are a number of interesting reports other than just DPS. Studying these proc-rates, up-times, energy-sources, etc is usually our best way of finding modeling inconsistencies.
Difference for the mutilate rogue a whopping 900 dps.
This cant be right because they have the same buffs (optimal_raid=1).
Looking at the breakdown envenom is missing 1100 DPE and the others are aboute 17-20% lower in the singel Mut rogue setup.
P.S: For comparing rogue DPS either have no rogue be target of Trick of the trade or give them an equal share in the default Rogue/Raid.simcraft.
Last edited by Karmon : 03/15/09 at 9:17 AM.
Reason: Removed premature reasoning
There is a strange interaction going on between simulations of Mutilate (51/13/7) rogues alone and a mix with Mutilate and other rogues.
<snip>
For me it looks like the poison stacks are not done properly.
1) Each rogue gets its own stack on the boss (deadly, wound)
2) Envenom operates with this private stack
3) Mutilate needs to check if any raidmember has a posion on the boss (including hunter stings e.g)
P.S: For comparing rogue DPS either have no rogue be target of Trick of the trade or give them an equal share in the default Rogue/Raid.simcraft.
Hmmmm..... Those three items are implemented.... but obviously there is a bug somewhere. I'll dig in later tonight.
Regarding TotT: If no target is specified, I allow them to give the buff to THEMSELVES. I know this isn't correct, but it was a convenient method for testing a Rogue in isolation. In multi-player sims, I supply an actual target. If you look in Raid_T8.simcraft or Rogue_T8.simcraft you will see that TotT is applied in a circular fashion among the Rogues.
The reason it happens is that if one rogue uses 'rupture' in its cycle the global buffs list 'mangle' in addition to the others.
Now it seems that this 'mangle' adds about 17% damge across all damage sources.
If 'mangle' is in fact the cat/bear mangle effect on the boss it should only increase the DOT's from rupture and similar bleed effects of other classes.
Yesterday I was playing with comparing mutilate specs including rupture or just envenom spamming. It appears with this 1832 version that including rupture back into 51/13/7 is a dps gain over envenom spamming. Which means that Blizz's change to the envenom glyph and MP is working well. Fast OH is better than slow OH. MH is still best slow. Poisons are still best IP/DP.
But I notice that in the output of Simcraft rupture is not included in the list of Up-Times? At the moment you can figure out the uptime by looking at the number of rupture ticks instead, but it would be nice to see a straight up % too.
Difference for the mutilate rogue a whopping 900 dps.
This cant be right because they have the same buffs (optimal_raid=1).
Looking at the breakdown envenom is missing 1100 DPE and the others are aboute 17-20% lower in the singel Mut rogue setup.
P.S: For comparing rogue DPS either have no rogue be target of Trick of the trade or give them an equal share in the default Rogue/Raid.simcraft.
Somehow, in the single setup the mutilate rogue is not using hunger for blood at all.
Edit:
Found it, optimal_raid did not supply the target with a bleeding effect, together with the mutilate profile not using rupture this resulted in no HfB use at all
Fixed in r1839
Last edited by Starfox : 03/15/09 at 1:42 PM.
Hello.
Light the fuse.
For all my homies.
Do not run, we are your friends.
SimulationCraft Druid Guy
Regarding Mutilate build and nHaT. I was actually playing with the sim after I fixed some bugs before the weekends but had to leave town and did not have time to upload something or post here. I just came back, so I will be able to mess with the sim a little bit only tommorow, but I want to post some thoughts (the may be in response to some posts here, but I wont quote).
Couple points regarding Mutilate:
1. With my limited testing it is actually worth using Rupture in a rotation (with blood spatter and a glyph). Even with some very strict rules on rupture use the profile (51/13/7) gains ~50 dps.
2. 51/18/2 falls behind 51/13/7 with the default profile no matter how you tweak the cycle. Weaving ruptures is basically night impossible due to lack of RS. I believe that with some a point shifted to impSnD it may be possible, but I don't think it will catch up to 51/13/7.
3. TtT is actually a very slight dps loss now compared to MP (especially for 51/13/7).
4. I did not play with the offhand speed, but it may be more optimal to use faster oh (though one major point about it in the end).
About NHaT:
When I removed Envenom glyph I simply changed NHaT to weave in Eviscerates because my very short test showed an increase in DPS. I did not really tweak it much as I was more interested in digging into the sim mechanics.
Regarding changing weapons:
The sim currently uses actual gear profiles (for Mutilate and HaT profiles there are chardev links, for Combat there isn't because I forgot to save them :/). Now, what it means is that the stats on the profiles include the stats on the weapons. While it is reasonable to tweak the speed and dps on weapons to see how the specs react to them in general the actual dps numbers will be skewed because the gear stats do not change.
So, if you are trying to see if its better to use WD or SR in OH you have to change the gear stats accordingly.
Somehow, in the single setup the mutilate rogue is not using hunger for blood at all.
Edit:
Found it, optimal_raid did not supply the target with a bleeding effect, together with the mutilate profile not using rupture this resulted in no HfB use at all
Fixed in r1839
This would have been easier to find if we reported all actions instead of just those that directly contributed damage.
I have two major work projects with end dates of early-April.
After they complete, my plan is to port the code to the Qt environment........ which opens up many, many opportunities......
Originally Posted by Thargood
But I notice that in the output of Simcraft rupture is not included in the list of Up-Times? At the moment you can figure out the uptime by looking at the number of rupture ticks instead, but it would be nice to see a straight up % too.
Added in r1840.
Last edited by dedmonwakeen : 03/15/09 at 4:08 PM.
Well, I was actually quite aware of that. I was going to post a suggestion for it (treating target always bleeding if optimal_raid=1) but as I said I had to go.
Also, what's your guys stance on letting rogues start the fight stealthed? I understand we are not always allowed to (though we may trick with vanish). But with implementing garrrote (which I've done locally for the time being) it has some positive effect on wind-up time of mutilate profiles (and some dps too).
Last edited by nuoHep : 03/15/09 at 4:00 PM.
Reason: spelling, meh :/
Also, what's your guys stance on letting rogues start the fight stealthed? I understand we are not always allowed to (though we may trick with vanish). But with implementing garrrote (which I've done locally for the time being) it has some positive effect on wind-up time of mutilate profiles (and some dps too).
I instrumented most of the functionality for starting stealthed..... but my lack of experience with the class meant I was uncertain how to test it.
Feel free to open with the "stealth" action and proceed accordingly.
We might need to be more thorough about setting the "harmful" bit in the actions to make sure some of the setup actions do not erroneously break stealth....
I understand that the whole discussion is about 3.1
But wouldn't it also be a dps increase for the current live build to use 23/5/43 build and doing a mix of Envenoms and Eviscerate over what the vast majority of the rogues are using (13/15/43 if you have a semi-decent group composition with Calamity's Grasp and Webbed Death, DP/IP)?
Or are there any particular changes in 3.1 which makes the envenom build worth using?
Am I right understanding tthat this line:
actions+=/envenom,min_combo_points=5,env<=1
means that you are using Envenom only at 5 combo points and whenever you got a 1+ dose of Deadly Poison?
By the way, what does MP stand for? :P
means that you are using Envenom only at 5 combo points and whenever you got a 1+ dose of Deadly Poison?
By the way, what does MP stand for? :P
env<=1 means the time left on the envenom buff, for poison doses you can use min_doses=Y, no_buff=1 is another option so envenom is only used when you don't have the buff
Hello.
Light the fuse.
For all my homies.
Do not run, we are your friends.
SimulationCraft Druid Guy
Would love a quick guide on how to input your own stats and talent specs in this utility. Keep up the good work guys!
Just open one of the Rogue_xxx.simcraft in a texteditor.
The talents are input through a link to an online talentcalc, be it mmo-champion, wowhead or the wow page itself.
How you define your gear can be seen in Melee_T7_Base_Gear.simcraft, which you happen to see in the rogue-files. When the sim sees a valid filename in a config file, it just treats that as if the contents of this file were there, so it's easier to have multiple profiles with the same equip. You'll notice fast how SC is told that some special proc items are equiped, it's mostly the name of the item in lowercase and spaces replaces with an underscore [Fury of the Five Flights] equipped?
DMC?
fury_of_the_five_flights=1
darkmoon_greatness=1
But be aware that those only active the special behaviour of the items, e.g. the DMC option does not include the 90agi/str/etc. that trinkets provides
The gear_xxx values represent stats from gear without talents!
Setbonuses are just:
tierX_Ypc=1
Activate the Y piece bonus of Tier X set.
The action strings define the logic behind what skill is used next, I don't think there is a detailed wiki page about any other class then druids :/
On a sidenode, if you want to simulate under PTR conditions:
patch=3.1.0
Hello.
Light the fuse.
For all my homies.
Do not run, we are your friends.
SimulationCraft Druid Guy