A DPS Simulator

[pandabearcat]

I must retract my previous statement.

Checking Bards, even though they run out of TP at around 5 minutes sustained, blow dragoons out of the water.

Previous bug was consuming GCD's prematurely, instead of only being consumed on player action.

They are at least 20% more potency.

Nerf is necessitated!

PS give me a dragoon rotation that I can code in that performs better please D=

EDIT: OH GOD BARD IS OUT OF CONTROLLLL

EDIT2: oops!

Crit: 1200
Duration: 60000
GCD: 245
Number of trials: 10000
Priority avg potency: 60676.6138
Rotation avg potency: 53138.5257
Bard avg potency: 107088.6278

[pandabearcat]

OH GOD I DON'T KNOW WHAT I'M DOING I MUST RETRACT STATEMENTS LEFT AND RIGHT AHHHHHHHH.

Oops, same bug that plagued Bardsim plagued Dragoonsim as well.

All is right with the world:

Crit: 1200
Duration: 60000
GCD: 245
Number of trials: 10000
Priority avg potency: 110569.2626 PPS: 184.28210433333334
Rotation avg potency: 97613.4686 PPS: 162.68911433333332
Bard avg potency: 107090.9382 PPS: 178.48489700000002

EDIT: I hit my post cap T.T

EDIT2: thought about priority a bit more, added a VERY conservative FR filler to priority rotation, upped DPS by a consistent, but low, 0.15%

Thinking about putting it into the "hard" rotation too.

EDIT3: Conservative work and fixing a few TP issues with the "hard" rotation paid off! Here are the new patchwerk numbers:

Crit: 1200
Duration: 60000
GCD: 245
Number of trials: 20000
Priority avg potency: 110728.80215 PPS: 184.54800358333333
Rotation avg potency: 111254.9575 PPS: 185.42492916666666
Bard avg potency: 107097.2132 PPS: 178.49535533333335

Hard rotation pays off! Woo! By about 0.45% difference...hah.

Bards are falling by the wayside, by about 3.8% worse off than dragoons in pure potency, closer to 10-15% worse because of weapon damage penalty.

Almost entirely because of TP usage, btw. Bards are just too inefficient.

[pandabearcat]

Just a small update for today.

Tweaked BRDsim so that it has a toggle for allowing disembowel.

When allowing disembowel, BRD is simming about 4% higher than DRG, without, about 4% lower.

Here is the current numbers:

Crit: 1200
Duration: 60000
GCD: 245
Number of trials: 20000
DRG Priority avg potency: 110060.34405 PPS: 183.43390674999998
DRG Rotation avg potency: 111253.8622 PPS: 185.4231036666667
Bard avg potency (w/ DE): 115540.0205 PPS: 192.56670083333336

I have a lab to complete over the weekend, so likely won't start monk work until monday or tuesday. Sorry!

Heres what you can do to help:

• Need to know if IR affects DoT clipping. If it doesn't, I need to know whether it is snapshotted or not.
• Need to know exact formula for GL stacks contributions to GCD speed.
• Need to have some decent damage profiles for "loss of uptime" for more realistic melee sims. Currently going to be normal distributed 20 seconds, 2 second STD, with a duration of 2 seconds, 0.5 seconds STD. No warning.
• Would also like to have "warning" simmed downtime, aka when you can anticipate downtime. These should be different profiles for different fights. For example, when you know bombs are coming out or plumes for titan.
• In that vein, would like to know people's general thought process when getting back on a boss, in order to make a smarter priority rotation. For example, what DoTs do you check? When do you start reapplying? Do you keep your normal priority? Do you refresh DoTs beforehand if you know a downtime is coming?

Thanks, and have a good weekend!

[EasymodeX]

Need to know if IR affects DoT clipping. If it doesn't, I need to know whether it is snapshotted or not.

Did a 1-time casual test last night while doing other things and pretty sure IR does affect DOT clipping.

Need to know exact formula for GL stacks contributions to GCD speed.

Pretty sure it's close to: GCD = (2.5-0.01*(total_SS-341)/10) * (1 - .05*GL_stacks)

In other words the stacks of GL are additive; GL as a whole modifies the resulting GCD directly after SS is applied.

Need to have some decent damage profiles for "loss of uptime" for more realistic melee sims.

Depends too much on boss. For ADS1 I would guesstimate 1s drop every minute (or a 50% chance of a 1s drop every 30s). For Cad, I would guesstimate 5s every 3 minutes due to oddball mechanics. For t2, same as ADS1 for miniADS, but add in a bonus 5s every two minutes for some ADS2 mechanics.

In that vein, would like to know people's general thought process when getting back on a boss, in order to make a smarter priority rotation. For example, what DoTs do you check? When do you start reapplying? Do you keep your normal priority? Do you refresh DoTs beforehand if you know a downtime is coming?

This is more of a strategic question. Guessing individual players will do things rather differently.

Generally I re-engage a rotation using the normal logic that I suspended with unless it was a long disconnect (in which case I skew towards refreshing HT and re-starting the rotation). I use the same checks for all DOTs that I use in the normal rotation (my checks in-game are similar to the priority ruleset discussed above, but less tuned for the sim).

For MNK, I "do a short cycle / no fillers to refresh GL asap", followed by checking Demolish timing, because TOD has a long duration and isn't likely to have dropped from a short disconnect. That said I haven't played my Monk much so I kind of suck at it and have bad reflexes for it. I suspect a more experienced Monk would be able to check DOT timing more quickly and intuitively as well as maintain a higher % of bootshines.

[pandabearcat]

Thanks for the info, I might go start some work on it.

I was bored and wanted to test what would happen if a DRG couldn't flank or backstab target.

Losing flank = ~7.85% dps loss.
Losing backstab = ~6.8% dps loss.
Losing both = ~13.2% dps loss.

Don't ask me why. I don't know. I figured the losses would be higher.

EDIT: this is with priority, didn't know or want to do hard rotations.

EDIT: I guess this means if you are super lazy, you can still do >90% damage if you just sit on flank and ht, ph, TTT, esp if there is another DRG for DE.

[EasymodeX]

That doesn't make sense because the DE debuff alone is 10% of physical damage. Physical damage should contribute [checks model] over 90% of DRG DPS, probably in the 95% range. So that's at least a 9% DPS drop for not being able to use DE. Is DE not-applying appropriately when you fail a rear position on ID? And/or the ID combo should not be registered when the position fails, etc ~

[pandabearcat]

I found the problem to be AA was not being counted for DE.

Going to run it around again at some point.

Here's some interesting info:

Damage breakdown by percent (DRG Hard Rotation)

Autoattack	   : 24.80994403804434
Full Thrust	   : 13.427450699667673
Vorpal Thrust	   : 8.74680073415357
Heavy Thrust	   : 5.461454471204516
Phlebotomize	   : 5.696775002473597
True Thrust	   : 6.627933455030803
Chaos Thrust	   : 3.6600681026464557
Leg Sweep	   : 4.641830295724796
Phlebotomize (DoT) : 3.5361663693006506
Impulse Drive	   : 3.961845609934256
Chaos Thrust (DoT) : 3.8505354205518953
Disembowel	   : 4.714614754248133
Jump	           : 4.443250545214875
Fracture	   : 1.437419857378893
Dragonfire Dive	   : 1.30678245663098
Fracture (DoT)	   : 1.5256598485918367
Spineshatter Dive  : 1.952095825620424

As for why the DoTs do less damage than their physical attacks, seems like something's up with my sorting algorithm.

[EasymodeX]

Phlebotomize : 5.696775002473597
Phlebotomize (DoT) : 3.5361663693006506

PH's direct component is 170p (187 w/ DE), with the DOT at 120p. This is not surprising.

Chaos Thrust : 3.6600681026464557
Chaos Thrust (DoT) : 3.8505354205518953

CT's is 160 (176 w/ DE), with the DOT at 200p. This also seems reasonable.

But yeah the sorting is funny.

[pandabearcat]

Yea I fixed the sorting. Apparently integer truncation are bad when sorting percentages >.> Who knew!

Here is data!

DRG Hard Rotation:

Autoattack          : 24.81%
Full Thrust         : 13.42%
Vorpal Thrust       : 8.75%
True Thrust         : 6.63%
Phlebotomize        : 5.7%
Heavy Thrust        : 5.46%
Disembowel          : 4.71%
Leg Sweep           : 4.64%
Jump                : 4.45%
Impulse Drive       : 3.96%
Chaos Thrust (DoT)  : 3.85%
Chaos Thrust        : 3.66%
Phlebotomize (DoT)  : 3.54%
Spineshatter Dive   : 1.95%
Fracture (DoT)      : 1.53%
Fracture            : 1.44%
Dragonfire Dive     : 1.31%

Bard (normal priority, cooldowns on cooldown)

Autoattack          : 30.26%
Heavy Shot          : 20.89%
Bloodletter         : 12.73%
Windbite (DoT)      : 8.31%
Straight Shot       : 7.67%
Venomous Bite (DoT) : 7.1%
Flaming Arrow (DoT) : 4.2%
Venomous Bite       : 4%
Windbite            : 2.42%
Misery's End        : 2.25%

[EasymodeX]

Your sim simulates random rng for stuff right? Can you temporarily remove the rng crit% chance (e.g. just set crit = 0%, no IR buff, straight shot = 0%) and simulate 0 SS, 100 SS, and 200 SS Bard sims (but leave in rng for tick timing), with the results for absolute damage of each attack? (or every 50 SS if you want to get detailed)

I'm curious to see how SS affects the actual capture and delay of DOT ticks over a full fight. Even though SS can reduce the gap in coverage ... due to the DOTs being exact multipliers of 3s, capturing an earlier tick also loses a later tick. The acceleration must be repeated every time / multiple times to capture a net increase in ticks over a full fight duration.

Sidenote: I realize this will break ROB procs. Should not be significantly relevant / the error can be estimated easily enough I think.