Simulationcraft for Sage/Sorcerer

[lostdummy]

(01-20-2012 12:04 AM)Caltiom Wrote:  4. Negative Alacrity Values for the Balance Report: Have a look at the error number compared to the scale factor, you clearly can't say the true scale factor is negative. But it should be slighty higher because Telekinetic Throw/Force Lightning benefits from it, I'll investigate.

Not sure how to interpret error values;p

Simulation data for Madness show -0.0129 haste value, or -0.0366 normalized , with error just 0.0045.

Even if that would be absolute error, and assuming max error value, it would offset those two numbers to -0.0084 , or -0.0321 - ie still negative.

(01-20-2012 12:04 AM)Caltiom Wrote:  6. I can't find any problem with Disturb Mind/Exanguinate, the simulationcraft.org shows 2 points and affliction is cast around every 24s. If there are problems with your own profiles, please share the talent link you used.

This is where I found just 1 point in Exanguinate:
-go to: http://www.simulationcraft.org/swtor/101...101-2.html
- expand for Sith Sorcerer : Madness
- expand for Talents
- there under 'Lightning' it is liste just "1" for "Disturb Mind"

And as you mentioned, that same Madness build use 21sec Affliction.

[Caltiom]

The negative error values dissappeared in the latest reports. I couldn't find any special reason why they had been negative before, maybe some interactions with FL, resource problems or a unoptimal action list for higher haste levels.

I also updated the talent links, basically copied the talents from the excel spreadsheet, which gave all specs a dps gain. Thanks for that.

Adding the extra tick to CT, and the extra talent point for affliction duration increased Madness DPS quite a bit and it is now leading again. I'll look at the numbers in more detail next week.

[Khadroth]

(01-19-2012 04:21 AM)Shifft Wrote:  Are chain lightning / telekinetic wave implemented yet? I can't get them to cast although everything else seems to work. Great job in general with the sim.

Not sure if this was over looked or not, but I encountered this too.

When plugging in my own specs from Torhead, I noticed that Full Balance kept coming out inexplicably on top. Not that it's bad, just that it had quite a margin of lead even over the Hybrid specs. The commonality seemed fairly obvious, TKwave. So sure enough when I looked at the ability break downs on all the simulations, TKwave wasn't even listed.

Other than that I was pretty shocked at how competetive the specs were in a patchwerk style dps race. My concern is that most of the bosses require a fair amount of movement though which seems to be a nail in the coffin for specs like Full TK.

[lostdummy]

(01-20-2012 04:43 AM)Caltiom Wrote:  The negative error values dissappeared in the latest reports. I couldn't find any special reason why they had been negative before, maybe some interactions with FL, resource problems or a unoptimal action list for higher haste levels.

I also updated the talent links, basically copied the talents from the excel spreadsheet, which gave all specs a dps gain. Thanks for that.

Adding the extra tick to CT, and the extra talent point for affliction duration increased Madness DPS quite a bit and it is now leading again. I'll look at the numbers in more detail next week.

Well, reliable stat values was always main problem for simulation, or main advantage for formula based solutions like excel. Basically, if your average DPS/damage vary between two full simulator runs in similar amount as variation from +30 or whatever is increment of stat you use when finding stat value... then you get unreliable stat values.

One of first things I added to simcraft code back in time when I was contributing was 'Normalized Random generator' - to try and reduce just that problem.

As for Madness DPS being above others, that is another difference between sim and Excel, since excel show Hybrid (0/13/28) as best DPS build, not Madness. But I didnt check sim Hybrid numbers since there was no such build on sample results page.

[CaseyTheRetard]

Simulationcraft is still using the WoW method of applying haste to cast times [ base / (1 + haste) ] instead of the currently accepted [ base * (1 - haste) ] that we are using for SWTOR theorycraft.

[lostdummy]

Related to that comparison of Madness vs Hybrid, since only Hybrid build posted on sim sample page is one for Republic, I added option to select Republic or Empire names in my excel, so I can easier compare things.

And one obvious difference is : you are not using Telekinetic Wave (Chain Lightning) in Hybrid Sim.

In Excel using that spell increase DPS - although I doubt that spell alone will explain difference in Hybrid DPS on excel and sim.

BTW, I tried comparing Sorc Madness build talents again, and even with Republic talent names, I still see some differences:
- you use 2pt in Focused Insight , and you dont have at all Critical Kinesis as talent listed (probably those two are swapped?)
- you have 2pt in Psychokinesis, and that is Shadow/Assasin skill?
- you also have number of points in healing talents, but I guess those are just fillers

Also, your talent link on torhead is not working: http://www.torhead.com/skill-calc#sith_s...1201122321

When rechecking madness numbers, I noticed that your Affliction now has much higher DPE than excel (~17%), and FL has also ~10%, while other spells have some small (~2%) higher DPE than excel. Trying to find reason for that 2%, I found that your 'Spell crit %' (one that can be seen when I expand 'Stats' under Madness build) is higher than in Excel:
- for 'Unbuffed' you list 36.28%, while excel has 32.6% (with same willpower/power stats)
- for 'Raid buffed' you have 40.05%, while I have in excel 38.7% (with 'almost' same stats: same power, but you have 1904 willpower while excel has 1899)
One explanation is that 'Stats' numbers include 3% from talent 'Seeping Darkness' (Penetrating Light) - which would get those crit% close.

But related to difference in DPE for Force Lightning, it is now ~10% higher on Sim than on Excel, even if when I compared those for data from your previous posted version, FL DPE was comparable between Sim and Excel..
That DPE should basically be easiest to compare since no proc effects like Deathmarks etc increase its DPE ... in other words, its DPE should not be affected by other spells, timings and rotations. And yet simulation has FL DPS at 2574, while Excel has it at 2326. Actual FL DPE on default Madness build in excel is 2462, since on that build Critical Kinesis is used while on your build it is not, since FL has exact same (41.3%) crit chance as Affliction and other spells. But even that DPE is lower than your FL DPE without crit talent.

[lostdummy]

I think I found what cause such difference in Sim DPE for FL:

Your crit bonus damage for FL appears to be 108%, while excel has it around 78% (with 201 surge in both cases).

It seems like you use effect of Creeping Death talent (Mental Scarring) on FL too, although I think it was posted on forums that it does not affect FL, only DoT ticks.

If above would be fixed in sim, it would get FL DPE close to excel.

That leaves Affliction with highly different DPE, and one thing I noticed is that you have 8 ticks in Affliction, while there should be only 7 with exanguinate. Also, your affliction lasts 24+sec is probably explained with that.

BTW, I assume you made both of above changes in new posted version, since in last version i compared both Affliction and FL had similar DPE between excel and sim.

It would be probably good approach if you could check numbers from new builds with previous builds, or with excel - and try to catch significant changes in DPE that were not intended in new version.

[DPSFTW]

i get an error when trying to add the proc Wrath to Chain Lightning and Crushing Darkness

Code:

#!./simc

sith_sorcerer=Sith_Sorcerer_Madness
origin="create_Sith_Sorcerer.simc"
level=50
race=human
position=front
role=spell
use_pre_potion=0
talents=http://www.torhead.com/skill-calc#201ZfcM0dZcrcRsMkrc.1
actions=stim,type=rakata_resolve
actions+=/power_potion,if=time>10
actions+=/mark_of_power
actions+=/snapshot_stats
actions+=/affliction,if=!ticking
actions+=/death_field,if=!dot.crushing_darkness.ticking
actions+=/force_lightning
actions+=/chain_lightning,if=wrath
actions+=/crushing_darkness,if=wrath
shirt=foo,stats=1575willpower_536crit_306haste_107power_1217forcepower_201surge
# Gear Summary
# gear_crit_rating=274
# gear_haste_rating=271
# gear_power=304
# gear_forcepower=1210
# gear_willpower=1442
# gear_surge_rating=164

[Kaedis]

Quote:(with 'almost' same stats: same power, but you have 1904 willpower while excel has 1899)

This is an issue with how Will of the Sith is combined with Mark of Power. 1904 is the value you get if you apply them in a multiplicative fashion. They are additive in-game, however, which would give the 1899 figure.

[pingpong]

@DPSFTW
I think the line would be:

Code:

actions+=/chain_lightning,if=buff.wrath.up

However there's a couple of issues. Firstly Chain Lightning (or the Sage equivalent, Telekinetic Wave) are not yet implemented so this can't be modelled at all yet. Secondly, Wrath isn't in yet but the Sage equivalent, "Presence of Mind", is so you can model your Crushing Darkness line:

Code:

actions+=/crushing_darkness,if=buff.presence_of_mind.up

It's okay to mix and match Sage and Sorc names so it will still check for the same buff.

I find that ensuring Crushing Darkness is buffed by Wrath gives a 2DPS increase but I think you're right that it should be specified and once we can model Chain Lightning using up the same buff it will become more important, since at the minute there's very rarely a time when Wrath isn't up.