Simulationcraft for Sage/Sorcerer

[Kantern]

Would the weapon be the only one you can do this with? Otherwise it would not include weapon damage?

[Zoopercat]

We just launched our character builder a couple days ago, so now we're working on the API. We expect to have it done in 2 weeks, possibly much sooner. Stay tuned, we're working with Simulationcraft to make it easy for users

[Feyas]

Hey, quick question.

I looked through this thread to make sure, but I didn't see any reference to the reduced cooldown on force speed in the lightning build.

I was wondering if you were factoring that in to heavy movement fights?

In preparation for the expected 1.2 nerf/fix to lightning barrage, I switched to full lightning to see if I like it better. Having force speed available virtually anytime I need to move in a fight, means I'm moving faster from point A to point B, which means less time moving, which means more time doing normal rotation.

I don't know how much of an effect it is though. I remember people in WoW working out that the run speed boot enchant was worth more than anything else for raiding (and that was only 8%) ...having 150% movement available more often seems like it'd be even more worthwhile.

Given that according to the sims post 1.2, hybrid = madness = lightning in straight up DPS. Wouldn't that put lightning on top in any fight that requires movement more often than every 30 seconds?

[Kaedis]

Well, it's true that Force Speed will have an effect on movement fights, but the likelihood of 10 seconds off Force Speed's cooldown having a very measurable effect on overall damage, even on a heavy movement fight, is rather low, particularly since most movement on boss fights (specifically, short movements wrapped by stationary dps periods) are only on the order of a dozen meters at the most, which means that Force Speed saves you maybe a half-second of downtime per pair of movements within 20 seconds of each other, which is at the most once every 40 seconds (ie. movement occurring exactly every 20 seconds), which is a savings of a GCD every ~2 minutes.

Beyond that, Force Speed triggers a global cooldown. This means that it's actually a DPS loss to use Force Speed in several easily-identified circumstances. Lets say for the sake of example that you can generate about 750 damage/GCD (500 dps) mobile, compared to say 1500/GCD (1000 dps) while stationary. You need to move 12 meters. Players move at 6 meters per second in SWTOR, so this would normally take you 2 seconds. Force Speed increases your movement speed to 15 meters per second, meaning you cover the distance in 0.8 seconds instead. Without using Force Speed, you have 1 GCD at 750, then a half-second pause (since you need to be pausing for more than half a GCD at our given damage values for it to be worth casting another mobile spell), and 2/3rds of a GCD at 1500, total 1750 damage. If instead, you cast Force Speed, you get there in 0.8 seconds, wait 0.7 for the Force Speed GCD to finish, then get a full GCD at 1500, total 1500 damage. In this case, it's actually better to NOT use Force Speed, given the constraints above.

Force Speed is most useful for longer runs. If you're needing to move 30 meters, for example, Force Speed will get you there in 2 seconds, whereas normal running will get you there in 5. Normal running would give you 5 seconds of mobile dps (500 dps), so 2500 damage (unlikely to be this high, actually, since mobile dps is CD based and thus tends to go down the longer you're running). Force Speed would give you 2 seconds Force Speed GCD, 0.5 second pause, 3 seconds full damage, 3000 total damage.

Basically, the moral of the story is that not only is a reduced CD on Force Speed not really going to be a significant amount of damage increase, but using Force Speed at all isn't necessarily a good idea in every circumstance, particularly for short movements.

[Tacoman]

(03-03-2012 06:19 AM)CaseyTheRetard Wrote:  

(03-03-2012 03:36 AM)Tacoman Wrote:  Does the sim for a 7/17/17 build take into account that you cut off your current FL as soon as wrath proc's to cast the next CD/CL/LS?

Obviously the dps would be higher than letting the cast finish.

No, the sim does not. It would "obviously" be higher if there was a reliable way to interrupt the FL channel immediately after a tick, but considering how poorly the animation typically syncs with the cast in-game I don't believe that's possible. As the game exists today I think that interrupting FL is effectively going to result in you wasting half a tick time on average. To determine the best strategy, you need to evaluate the alternatives:

1. Interrupt FL: lose half an FL tick.
2. Allow n FL ticks to complete before casting CL: lose n opportunities to proc Wrath.

An FL tick is around 630 damage, half of that is 315 damage - so that part's simple enough.

n opportunities to proc Wrath will result in 0.3n Wraths on average. Every Wrath gives you the chance to convert a GCD of FL to a GCD of CL. Wrathed CL is about 1980 damage, FL's non-LB DPET is about 880. So letting n ticks of FL complete - with Wrath up already - loses you n * 0.3 * (1980 - 880 * 1.5) ~= 200n damage. So 7/18/16 would gain average damage by interrupting non-LB FL to cast Wrath+CL when there are at least n >= 2 ticks remaining in the FL channel.

but

That implies that it's only a good idea to interrupt FL in the first 1/3 of the cast, which is impossible since the first 1/2 of the cast is inside the GCD.

Without a better way to time your interrupts to get them closer to the preceding FL tick, I'd say it's not a good idea to interrupt FL for Wrath: on average it will result in a minor DPS loss.

---

And like frmorrison said, I would never cast LS as 7/18/16. It's tempting to do so when you're moving and CL is on cooldown, but it is likely a damage loss vs. simply saving the Wrath for later considering that Wrath+LS has lower DPET than FL.

Thank you, someone had told me this game had a 1 second cool down. If its 1.5 that changes everything.

I had some trouble understanding your formula ( I stopped taking math at algebra 2 in high school ) Just to see if I am reading it right, did you mean to write:

n * 0.3 * (1980 - 880 / 1.5)?
I got n220 by dividing by 1.5

I am also having trouble with your values for CL and FL.

my spell numbers are a little different from yours:

gross damage: CL = 1874 FL = 2622
damage per 1.5 seconds: 1874 FL 1311
damage per 1 second : 1249 FL 874

so could your formula be written:

n* 0.3 * (1249 - 874)?
= n112.5

or

n* 0.3 * (1874 - 1311 / 1.5)?
= n112.6

which would mean the dps loss from letting FL tick is even less ( this makes my first post even more wrong)

This makes me question the build all together as my assumption of a 1 second cool down was wrong and at 1.5 sec CL's Damage per cast time is not near as good as I had thought. ( its a bummer, the build is a ton of fun to play!)

Thanks for straightening me out on this.

[CaseyTheRetard]

(03-04-2012 11:35 AM)Tacoman Wrote:  I had some trouble understanding your formula ( I stopped taking math at algebra 2 in high school ) Just to see if I am reading it right, did you mean to write:

n * 0.3 * (1980 - 880 / 1.5)?
I got n220 by dividing by 1.5

That formula is n * 0.3 * [additional damage we get from consuming one Wrath proc]. A Wrath proc lets us replace one GCD of FL (880 DPET * 1.5 seconds = 1320 damage) with a CL (1980 damage) so it adds (1980 - 880 * 1.5) damage vs. simply casting FL.

(03-04-2012 11:35 AM)Tacoman Wrote:  I am also having trouble with your values for CL and FL.

my spell numbers are a little different from yours:

I got my numbers by importing a 7/18/16 spec (Sith_Sorcerer_Suffusion) into Simulationcraft and replacing the talents with a spec that had no Lightning Barrage; I wanted to make sure LB wasn't throwing off the DPET of FL since there's no way it's better to interrupt an LB FL. That said, your numbers are within about 6% of mine.

For your numbers:

n*0.3*(1874 - 874 * 1.5) = n*0.3*(1874-1311)=n*168.9.

and half an FL tick is 2622 / 8 = 327.75.

n >= 2 is necessary to increase damage, so the same reasoning applies about not interrupting FL.

[Caltiom]

(03-10-2012 01:04 AM)Kor Wrote:  

Quote:On another topic, with the same input in 114-3 as 114-2 my DPS skyrocketed from 1296 in 114-2 to 1540 in the newest release. I haven't had time to see what different and sadly work beckons but is this correct and the old version was to low? or is the newest version reporting something wrong.

I've no idea. I know the profile I used was topping out at 1604, compared to the 1571 the default profile was at. I assumed I simply had some setting somewhere drastically different that y'all.

Sorry for choosing the arrogant way here: but just looking at Release Notes - SimC 114-3 (latest release) for a few seconds should give you the answer. Hint: It's the third point

[CaseyTheRetard]

(03-10-2012 02:31 AM)Caltiom Wrote:  Sorry for choosing the arrogant way here: but just looking at Release Notes - SimC 114-3 (latest release) for a few seconds should give you the answer. Hint: It's the third point

Interestingly, the armor change altered the relative performance of the specs. Was:



Hybrid: 100%
Madness: 97.3%
Lightning: 95.9%
Suffusion: 95.7%
Conduction: 86.7%

Now:



Hybrid: 100%
Suffusion: 96.2%
Madness: 94.7%
Lightning: 94.3%
Conduction: 87.3%

Madness and Lightning are less than 0.5% apart, and the FL-double-dip-exploiting specs picked up quite a bit. Projected DPS with double-dip out of the picture:



A drop of about 70 DPS for Hybrid and Suffusion, a whopping 95 DPS hit for Conduction. Hybrid/Lightning/Madness are in a very tight grouping on stationary fights. Throwing in some movement (HelterSkelter):



Lightning shows very poorly. It appears Hybrid and full Madness will be the specs to play even when/if the double-dip exploit is fixed.

---

Edit: I'm sorry the colors are horrible in the Simulationcraft charts right now. We need someone with better color sense than me to pick:

• A color for each Advanced Class: 8 or 16 depending on if you make the Empire/Repub distinction.
• A color for each damage type (4): kinetic, energy, elemental, internal.
• A color for each attack type (4): melee, ranged, force, tech.
• Ideally, a color for each combination of attack and damage type (16) that is a mix of the individual colors of the attack and damage type. It would be AMAZING to look at e.g. a damage-by-ability pie chart and be able to tell at-a-glance which abilities are force, and which are affected by armor, and which do internal damage, etc.

[Kaedis]

(03-10-2012 02:31 AM)Caltiom Wrote:  Sorry for choosing the arrogant way here: but just looking at Release Notes - SimC 114-3 (latest release) for a few seconds should give you the answer. Hint: It's the third point

Ya know, I remember seeing that in the update notes, and thinking "oh, nice they got the armor change we saw". Totally and completely failed to mentally connect that with the requisite increase in actual dps done.

On an unrelated note, Simc only takes into account 1 stack of the armor reduction, but last I checked armor reductions from different classes stacked (additively, I believe) on a target. Example, a Juggernaut and Arsenal Merc nuking a target would both have a stack of 20% armor reduction active, reducing the target's armor by 40%. That discussion sorta trailed across several threads, so I may have missed where this was disproved, but if not, will this be added to Simc?

[Caltiom]

(03-10-2012 05:18 AM)Kor Wrote:  On an unrelated note, Simc only takes into account 1 stack of the armor reduction, but last I checked armor reductions from different classes stacked (additively, I believe) on a target. Example, a Juggernaut and Arsenal Merc nuking a target would both have a stack of 20% armor reduction active, reducing the target's armor by 40%. That discussion sorta trailed across several threads, so I may have missed where this was disproved, but if not, will this be added to Simc?

Sure we'll add that to SimC, there is no reason why we should arbitrarily choose not to. Just as always, as much data as possible or confirmations from a multitude of sources is very much appreciated to have a solid empirical confirmation that something works as proposed.

A good example is the double-dip removal of 1.2. Uncritical as I am I just disabled the double-dip in simcraft for ptr mode and then Casey came along asking if there is a confirmation of the whole thing or if its just speculation.
I had to admit it is pure speculation, extrapolated from the statement of a developers in this Q&A that they will remove the double dipping for the Conveyance/Force Bending buff ( Heal spec ). To be honest, that really isn't a solid ground to forcebly disable double-diping in SimC. ( But of course we can do simulations for itand look at how dps changes without double-dip, the point is that we can't be certain that it will be that way. )

Edit: You can track http://code.google.com/p/simulationcraft...tail?id=30