Register

Experimental simulationcraft script

Face-rippin fun.
Exalted
User avatar
Posts: 1186
Joined: Fri Nov 25, 2011 7:49 pm

Re: Experimental simulationcraft script

Postby aggixx » Tue Oct 23, 2012 12:06 pm

Leafkiller wrote:I ran your new script with HotW + DoC and the dps is down 8k (113738)

That's... odd. I'm not sure why the incarnation change would affect any of the other talents. Furthermore, I don't think I saw a DPS change when I was comparing Inc/NV vs SotF profiles.

PS: Why does this need to be a different thread than the SimC discussion thread?
Image

Exalted
User avatar
Posts: 1145
Joined: Fri Jun 04, 2010 4:28 am

Re: Experimental simulationcraft script

Postby Leafkiller » Tue Oct 23, 2012 12:52 pm

I did not intend to move the discussion. Mostly I wanted to post the most current version of the script in an easy place to find it until we actually check it in. The rest was accidental...

I did not check to see what caused the dps drop.

Exalted
User avatar
Posts: 1145
Joined: Fri Jun 04, 2010 4:28 am

Re: Experimental simulationcraft script

Postby Leafkiller » Tue Oct 23, 2012 2:56 pm

Posted a new version of the script. The only change is this:

actions+=/shred,if=(buff.tigers_fury.up|buff.berserk.up)

becomes

actions+=/shred,if=(buff.tigers_fury.up|buff.berserk.up|buff.natures_vigil.up)

(Repeated for the Ravage line also).

It seems pretty obvious that if we are going to ignore pooling during TF, we should also ignore it during NV. Extrapolating this further, we probably should ignore pooling with many of the procs we get. Something to think about and work toward in the future. What is interesting about this is that it is very close dps-wise but increases the number of HTs that go out (by energy starving us so the HT conditionals get used). I am going to implement this in the Ovale script.

Exalted
User avatar
Posts: 1145
Joined: Fri Jun 04, 2010 4:28 am

Re: Experimental simulationcraft script

Postby Leafkiller » Tue Oct 23, 2012 3:39 pm

I forgot to repeat the "|buff.natures_vigil.up" on the Thrash line. That is fixed now.

Exalted
User avatar
Posts: 835
Joined: Tue Oct 23, 2012 7:15 am

Re: Experimental simulationcraft script

Postby raffy » Tue Oct 23, 2012 7:48 pm

Bunch of random questions wrt DoC:

Have you experimented with forcing an instant HT if PS will fall off before another ability can be cast? What if it results in energy capping?

What about HT during Berserk?

Is HT modeled as a spell-cast GCD not a cat GCD?

Should Berserk and TF be aligned with the GCD expiration?

What's worth casting other than SR if SR is down? What if you have 0 combos as well? (In practice, I find an unpooled 0-combo SR very dangerous as it occasionally expires before the next finisher)

Before BitW, is FB ever worth casting? The guards around non-BitW FB seem pretty tight, making me wonder if it ever gets cast?

If you knew the exact damage a current DoT and then expected damage of a new DoT (if clipped) and any other information necessary, what is the optimal point to clip?
Image

Exalted
User avatar
Posts: 1186
Joined: Fri Nov 25, 2011 7:49 pm

Re: Experimental simulationcraft script

Postby aggixx » Wed Oct 24, 2012 3:18 pm

raffy wrote:Have you experimented with forcing an instant HT if PS will fall off before another ability can be cast? What if it results in energy capping?

What about HT during Berserk?

Is HT modeled as a spell-cast GCD not a cat GCD?

Should Berserk and TF be aligned with the GCD expiration?

What's worth casting other than SR if SR is down? What if you have 0 combos as well? (In practice, I find an unpooled 0-combo SR very dangerous as it occasionally expires before the next finisher)

Before BitW, is FB ever worth casting? The guards around non-BitW FB seem pretty tight, making me wonder if it ever gets cast?

If you knew the exact damage a current DoT and then expected damage of a new DoT (if clipped) and any other information necessary, what is the optimal point to clip?


1. Action list already does this, worrying about energy capping is neutral at best.
2. DPS neutral if you do or don't.
3. Yes. There was a sim discrepancy during the MoP pre-patch where all spells in cat form were being forced to be a 1 second GCD (an old band-aid) but it was fixed and Healing Touch is now a 1.5 second GCD that scales with spell haste. You'll notice that if you run a sim you can compare the HT count to the time spent on HT and the latter will be higher.
4. Pretty much irrelevant. On paper you could fit 7 abilities in the 6 second window if you activate the first ability and the TF at the exact same time but in practice that is 100% impossible. SimC inserts a very very small delay inbetween each GCD so despite the TF line only ever triggering after the GCD ends it is still uncapable of fitting an extra ability in.
5. Pretty much only Faerie Fire (if it needs to be refreshed) and Healing Touch.
6. Yes, and they're actually not tight at all. It will Ferocious Bite anytime Rip is above 6 seconds remaining and clipping/refreshing Savage Roar isn't a more appealing alternative.
7. Varies dynamically based on how much stronger your new Rake would be than the current Rake. It would take a massive amount of work to figure out the actual curve.

I will say that points 1 & 2 are kind of "assumptions" at this point, as I haven't retested those things for a while and the action list has evolved a lot since I last did. I'd like to reevaluate them sometime soon when I have the time.
Image

User avatar
Posts: 1
Joined: Thu Nov 01, 2012 1:32 am

Re: Experimental simulationcraft script

Postby Apawcalypse » Thu Nov 01, 2012 5:10 am

I have a sort of a silly question about this script. If I copy the actions portion and replace the stock simc actions with it while using my toon/gear from the import will it still provide accurate results? I only ask because of the glove use portion of the script and my char is not a eng.

Exalted
User avatar
Posts: 1186
Joined: Fri Nov 25, 2011 7:49 pm

Re: Experimental simulationcraft script

Postby aggixx » Thu Nov 01, 2012 5:27 pm

Just remove the glove lines and it should, yes.
Image

Posts: 11
Joined: Mon Oct 08, 2012 9:44 am

Re: Experimental simulationcraft script

Postby Gurrshael » Sat Nov 03, 2012 5:16 pm

I have a question about tick_multiplier and dot.rake.multiplier / dot.rip.multiplier - how are those variables calculated?

1. Is it just mastery & TF & SR & AP scaling? Or the total tick damage including the AP non-scaling part ?
2. Is DoC included in dot.bleed.multiplier?
3. Is DoC included in tick_damage if the DoC damage buff is up and the bleed would consume it?

I tried looking through the simc code but after 5 years of not using C/C++ it looks like a complete gibberish to me.

Exalted
User avatar
Posts: 1145
Joined: Fri Jun 04, 2010 4:28 am

Re: Experimental simulationcraft script

Postby Leafkiller » Sat Nov 03, 2012 5:59 pm

Gurrshael wrote:I have a question about tick_multiplier and dot.rake.multiplier / dot.rip.multiplier - how are those variables calculated?

1. Is it just mastery & TF & SR & AP scaling? Or the total tick damage including the AP non-scaling part ?
2. Is DoC included in dot.bleed.multiplier?
3. Is DoC included in tick_damage if the DoC damage buff is up and the bleed would consume it?

I tried looking through the simc code but after 5 years of not using C/C++ it looks like a complete gibberish to me.


Ap and AP scaling is currently missing so it is not as precise as it should be. In Ovale, everything is being taken into account, so it is more precise. I am making the assumption that the ratios in simc are close enough to work for us although it would be useful if AP and AP scaling were taken into account.

Posts: 11
Joined: Mon Oct 08, 2012 9:44 am

Re: Experimental simulationcraft script

Postby Gurrshael » Sat Nov 03, 2012 7:28 pm

Leafkiller wrote:Ap and AP scaling is currently missing so it is not as precise as it should be. In Ovale, everything is being taken into account, so it is more precise. I am making the assumption that the ratios in simc are close enough to work for us although it would be useful if AP and AP scaling were taken into account.


Thanks. What about DoC is it included too?

The reason why I am asking is that I've rewritten the simulation logic from your script into my addon. I am using exact tick damages instead of the multipliers and sometimes it offers really counter-intuitive suggestions like:

Rake
Healing Touch (because PS is running out / I have 4+ combo points)
Rake (because this one will get buffed by DoC and thus will be 25% stronger)

This line is responsible for that:
Code: Select all
actions+=/rake,if=target.time_to_die-dot.rake.remains>3&tick_multiplier%dot.rake.multiplier>1.12


I am just trying to verify that it behaves the same way in SimC, which really dependes on whether the multipliers include damage buffs from DoC or not.

Btw, something similar happens at the start of the fight:
Rake
[Trinkets procs]
Rake again (because this one will be much stronger)
... Shred till 5 cp
NS+HT = DoC
Rip
Rake again (trinkets are still running and we now have DoC)

I understand that this situation is limited to just the start of the fight and it will happen very rarely (if at all) in the course of the fight but it is strange nonetheless.

Exalted
User avatar
Posts: 1186
Joined: Fri Nov 25, 2011 7:49 pm

Re: Experimental simulationcraft script

Postby aggixx » Sat Nov 03, 2012 7:40 pm

Yep, that's intended. Rake doesn't hit much less harder than a Shred and you generate the same amount of CP for less energy.
Image

Exalted
User avatar
Posts: 1145
Joined: Fri Jun 04, 2010 4:28 am

Re: Experimental simulationcraft script

Postby Leafkiller » Sat Nov 03, 2012 9:25 pm

DoC is included in the calculation.

Posts: 11
Joined: Mon Oct 08, 2012 9:44 am

Re: Experimental simulationcraft script

Postby Gurrshael » Sat Nov 03, 2012 9:34 pm

Thanks for explanation.

Posts: 13
Joined: Mon Sep 10, 2012 11:42 am

Re: Experimental simulationcraft script

Postby averter » Wed Nov 07, 2012 5:47 am

Any thoughts on optimizing AoE rotation especially with DoC?
Currently most guides tell to keep Trash up, Swipe for Combopoints and spend them on SR upkeep, Rip or FB.
These work with short burst AoE parts, but how does it work over a few minutes? Like on Wind Lord Mel'Jarak boss with quite tight enrage timer.

Exalted
User avatar
Posts: 1186
Joined: Fri Nov 25, 2011 7:49 pm

Re: Experimental simulationcraft script

Postby aggixx » Wed Nov 07, 2012 7:34 pm

I'm sure there's room for optimization regarding maximizing your PS and OOC procs other than just Thrashing at <3 seconds, but it's not something you could easily determine without simming it. It is possible to do such a thing in SimulationCraft (you could see some examples of multi-target or AoE action lists on the SimC website), it's just not something I've got around to doing.
Image

Posts: 22
Joined: Thu Sep 20, 2012 2:54 pm

Re: Experimental simulationcraft script

Postby Abranor » Fri Nov 09, 2012 10:59 am

Leafkiller wrote:
Here is the most current version of my simc script (using a DoC rotation):


Leafkiller, when you wrote this, did you mean using a DoC rotation or using a DoC build?

I'm new to SimC but looking in the script I see what look to be actions that involve using HotW and wrath (the portion that says "#actions+=/heart_of_the_wild,if=enabled") though to be honest I'm not sure what the "#" symbol in front of "actions" is there for. I don't see it on the other actions surrounding that line.

I'm just trying to educate myself and not sure other than asking what are probably just noob questions.

Thanks.

Honored
Posts: 73
Joined: Wed Oct 03, 2012 1:19 pm

Re: Experimental simulationcraft script

Postby kaiadam » Fri Nov 09, 2012 1:54 pm

averter wrote:Any thoughts on optimizing AoE rotation especially with DoC?
Currently most guides tell to keep Trash up, Swipe for Combopoints and spend them on SR upkeep, Rip or FB.
These work with short burst AoE parts, but how does it work over a few minutes? Like on Wind Lord Mel'Jarak boss with quite tight enrage timer.


Napkin math I did after trash:

My rake averages about 49k damage on application and bleed (damage is applied 6 times per rake - 1 on application and 5 times during the bleed)
My swipe averages about 38k damage.

To make swipe overcome rake damage, you'll need 6*49/38 = 7.7 or around 8 targets, ignoring combo points.

Personally though, I'd swipe at 6 targets because of the extra combos you get for better energy management for savage roar and the occasional rip.

Keep thrash up in either case, obviously.

Exalted
User avatar
Posts: 1145
Joined: Fri Jun 04, 2010 4:28 am

Re: Experimental simulationcraft script

Postby Leafkiller » Fri Nov 09, 2012 3:49 pm

Abranor wrote:
Leafkiller wrote:
Here is the most current version of my simc script (using a DoC rotation):


Leafkiller, when you wrote this, did you mean using a DoC rotation or using a DoC build?

I'm new to SimC but looking in the script I see what look to be actions that involve using HotW and wrath (the portion that says "#actions+=/heart_of_the_wild,if=enabled") though to be honest I'm not sure what the "#" symbol in front of "actions" is there for. I don't see it on the other actions surrounding that line.

I'm just trying to educate myself and not sure other than asking what are probably just noob questions.

Thanks.


When I started this thread I was focusing on the DoC rotation. Later I spent time on NV and HotW so the currently posted script has had optimizations for all three specs. You have to change the talents line near the top of the script to select what you want to test. I notice that it is currently set to NV: "talents=010103". If you want to test DoC you would simply change it to "talents=010102" and for HotW you would use "talents=010101"

The hash mark "#" simply means the line is a comment and Simulationcraft will ignore it. The reason you see the hash mark on that line is because I disabled it in order to test HotW without doing Wrath spam (what we refer to as passive-only HotW since we are testing the passive stat bonus). If you want to test HotW with Wrath spam you will need to do 2 things to the script. First you would want to remove the hash mark from that line. You also would want to add in some additional spell power to emulate doing a weapon swap. This is easily done by adding ",addon=7932sp" to the end of one of the gear lines in the script. For example you can change:
main_hand=gaorei_staff_of_the_legendary_protector,id=87156,gems=500agi,enchant=dancing_steel,reforge=exp_hit

to:
main_hand=gaorei_staff_of_the_legendary_protector,id=87156,gems=500agi,enchant=dancing_steel,reforge=exp_hit,addon=7932sp

If you are testing your own profile, then set the amount of spell power to whatever matches the weapon you are swapping to.

Exalted
User avatar
Posts: 1145
Joined: Fri Jun 04, 2010 4:28 am

Re: Experimental simulationcraft script

Postby Leafkiller » Fri Nov 09, 2012 4:08 pm

kaiadam wrote:
averter wrote:Any thoughts on optimizing AoE rotation especially with DoC?
Currently most guides tell to keep Trash up, Swipe for Combopoints and spend them on SR upkeep, Rip or FB.
These work with short burst AoE parts, but how does it work over a few minutes? Like on Wind Lord Mel'Jarak boss with quite tight enrage timer.


Napkin math I did after trash:

My rake averages about 49k damage on application and bleed (damage is applied 6 times per rake - 1 on application and 5 times during the bleed)
My swipe averages about 38k damage.

To make swipe overcome rake damage, you'll need 6*49/38 = 7.7 or around 8 targets, ignoring combo points.

Personally though, I'd swipe at 6 targets because of the extra combos you get for better energy management for savage roar and the occasional rip.

Keep thrash up in either case, obviously.


I am dubious about the napkin math since it is not accounting for the elapsed time it takes to get Rake on every target, nor the different energy costs. There is also a question of target time to die which works in concert with the speed at which you can apply Rake to multiple targets compared to just hitting all of them with Swipe. I have always been pretty dubious of napkin math because I have seen many times that simulations produce results that initially seem to be counter-intuitive until you dig into the data to see what the actual interactions are. I am not disputing your premise that there are times where multi-dotting Rake will provide superior dps, rather the thresholds you came up with. There may also be the opportunity for a hybrid approach where you Thrash, multi-dot Rake and use Swipes as fillers. Then there is the question of energy pooling for Thrash and SR (not to mention praying for omen procs :P).

One last thing that is worth considering is how the adds are spawning. Elegon is a good example where I Rake a lot of the adds as they are spawning, and then use Thrash and Swipe once they are grouped up (not to mention sometimes single targeting the first couple of adds that spawn). It is hard to capture all of that in a simulation script...

Posts: 22
Joined: Thu Sep 20, 2012 2:54 pm

Re: Experimental simulationcraft script

Postby Abranor » Fri Nov 09, 2012 4:50 pm

Thanks Leafkiller for the reply. That all makes sense now. I currently play HotW passive only for DPS saving the HotW CD for times when stuff gets a little crazy. Thanks to a HotW buffed tranquility the other day I brought us back from the edge just enough to get a boss first kill. I like having that kind of value to our group. I remember playing in TBC and quick shifting to Bear more than once bringing us back from what appeared to be an inevitable wipe.

This forum and its community is such a valuable tool, I'm glad I found it.

Honored
Posts: 73
Joined: Wed Oct 03, 2012 1:19 pm

Re: Experimental simulationcraft script

Postby kaiadam » Fri Nov 09, 2012 9:09 pm

Hence, napkin math, heh.

I also ignored the fact that swipe costs 10 more energy than rake, which balances out a lot of the rake drawbacks.

On a fight like wind lord though where rake is definitely going to have time to tick, I think it's going to be better dps as long as you can track which adds have rake and how long is left.

I didn't get to dps our wind lord normal kill last week and this week heroic vizier is kicking our asses, I can probably give some better experimental results in a few days.

Exalted
User avatar
Posts: 1533
Joined: Mon May 24, 2010 10:21 pm

Re: Experimental simulationcraft script

Postby Tinderhoof » Sat Nov 10, 2012 3:04 pm

I am thinking of messing with this next week on Windlord. However are you also taking into account that doing the tab Rake on the adds/boss will leave you with a max of 2 combo points to keep SR up? Usually it will only be 1. That is a lot of energy to use and you have to be lucky with a crit and the timing to get better then using 70 energy per minute.

I usually use HotW for the hurricane and then for Tranq at the end of the fight. Are you planning on using DoC for the Rake doting?

Exalted
User avatar
Posts: 1145
Joined: Fri Jun 04, 2010 4:28 am

Re: Experimental simulationcraft script

Postby Leafkiller » Sat Nov 10, 2012 3:35 pm

Using DoC for Rake dotting is problematic since PS procs are harder to come by. You also minimize the value of SotF during those periods with all the "wasted" combo points, so you might want to use FoN on that fight. I would consider going with the combination of HotW and FoN since those talents don't "demand" that you use your combo points efficiently. FoN is less DPS than SotF if you are finishing your combo points but combo point generation with AOE or multi-dotting likely removes most if not all of that advantage and in a similar vein I would expect the passive bonus of HotW to be worth more than DoC except at the point of the fight where the adds are all dead. It kind of depends on which part of the fight is the most difficult, the adds, or the burn down afterwards. There is of course the added benefit of HotW that you have tranq available, not to mention Hurricane is still a very strong AOE.

Honored
Posts: 73
Joined: Wed Oct 03, 2012 1:19 pm

Re: Experimental simulationcraft script

Postby kaiadam » Sat Nov 10, 2012 10:21 pm

Ok, so heroic wind lord basically has 7 targets up the entire time you're aoeing.

I use HoTW for the fight as hurricane is still ridiculous, and incarnation for this fight specifically (because of the 600% burst) but SoTF normally over FoN.

Swipe vs rake:

With 7 targets, it was really a tossup as to which method was more damaging. Rake really put up a good fight vs shred, but ultimately I felt the massively easier rotation with swipe and the fact that you can keep rip up on the boss for the burst phase after a set of adds die is a pretty good increase.

After burst phase was over with only 4 targets up for 15 seconds, I definitely found rake beat swipe handily and was keeping me at the limits of my energy regen while keeping thrash up.

For the moment, I'm going to maintain my earlier claim of 6 or higher = swipe, 5 or lower (and will survive for the length of the dot) = rake.

PreviousNext

Return to Kitty DPS

Who is online

Users browsing this forum: No registered users and 3 guests