Summary of a post I made regarding four DPS test runs discussing Marksmanship damage on the PTR. The thread can found here:
The TLDR being this:
2PC negatively impacts 4PC. The more haste you have the less benefit 4PC provides.
Arcane is bloated. There is no reason to use Aimed outside of CA.
Regardless of focus talent selected and rotation adaptation therefore, the average throughput was 170k. If you haven't read Arth's post about the skill gap yet, read it here:
The TLDR being this:
So long as you're pressing buttons you're going to do similar damage. Standard deviation is nil. Apply forehead to keyboard. That last comment was me being bitter about Survival.
Marksman T16 2PC - 15 second reduction for MM (10 for bm, 12 for sv)
Here is the summary of the post with the several tests:
I took four videos of 7.5 minute test sessions on the target dummies.
Run one was the "normal" MM rotation with Dire Beast.
Run two was the "normal" MM rotation with Fervor.
Run three was the "normal" MM rotation with TotH.
Run four was the "spam" MM rotation with TotH.
I haven't done this with BM yet. This was also only one session and I kept my RPPM/trinkets/etc etc equipped. I also allowed some margin of error with play. In a real situation I feel these are all natural elements of the game and totally different from theoretical mathematics. 7.5M is long enough to grant averages.
Used the audio as a time cue. Sorry about that again.
Run 1
Dire Beast. Ratio of 92/60 to arcane/aimed with 7.3% damage incoming from steady shot. Affected the most by Careful Aim bloat - therefore lowest damage of the four.
Top end burst: 205k.
5 minutes: 175k.
end: 170k.
Ability breakdown:
Video:
Run 2
Fervor. Ratio of 125/53 with 6.8% damage incoming from steady shot. Affected second most by Careful Aim bloat, similarly to run 1 - highest damage of the three, however, though more or less in line by 5-10k with the rest.
Top end burst: 245k.
5 minutes: 180k.
end: 180k.
Ability breakdown:
Video:
Run 3
TotH. Ratio of 143/46 with 6.4% damage incoming from steady shot. Slightly less damage bloat than the first two, also second highest damage of the three ... Nonetheless, everything still floating around a maintained 170k.
Top end burst: 250k.
5 minutes: 170k.
end: 175k.
Ability breakdown:
Video:
Run 4
TotH arcane spam. Ratio of 295/15 with 4.4% damage incoming from steady shot. Affected the least by damage bloat.
Top end burst: 235k.
5 minutes: 175k.
end: 170k.
Ability breakdown:
Video:
Ok, I generally agree with everything you've written here, except that 7.5 minutes is statistically not very significant. Or at least, I am not sure what the statistical significance of such a length of time is. In my own in-game tests I often get quite a bit of variance between tests, even tests of 5 minutes or more, due to RNG.
ReplyDeleteThat said, obviously all these options are 'roughly' the same, but my point is that the reason for doing theorycrafting instead of in-game tests is to remove the effect of the RNG.
Assuming you are correct here, I don't so much mind that the focus generation talent has very little effect on overall dps, but it is sad that spamming arcane seems so effective.
Question - since a target dummy is always at max health, doesn't Careful Aim always apply to it? I've always been concerned about this throwing off my numbers with marksmanship tests. Also, what addon did you have in the bottom centre of your screen to keep track of the damage ratio from different shots?
Yes, RNG is a fact of the game. The lower the test is, the more likely it will to have variance. For example, grabbing a 5 minute test and having a string of scope procs in the first opening minute is going to easily make a big difference. Which, of course, unless you have several such tests, might not be entirely accurate (but that's where a sim would probably be better).
DeleteIn a way, in-game tests share both large similarities and large differences to theorycrafting. Theorycrafting, of course, can negate the effect of RNG - but in game tests will help you decide how best to react to those instances of RNG. In the case of the PTR, there's larger chances for variance in play style that a simulation might not let you tinker with as accurately. Once you've gone in and know what you want to sim, though, it becomes easier to supplement that theorycrafting. Generally, then, I prefer to do a bit of both!
I used 7.5 minutes because that's primarily what people do use for theorycrafting sim tests, so they'd transfer over well and be cross comparable (especially for the RNG elements). It's also a good estimate to how long a rough progression boss will last, since they're generally around this time stamp. It also ensures you're not in the middle of a "burst" phase, so instead of seeing the higher end spectrum of your throughput you're seeing the sustained downtime. It's a long enough time that you've already seen at least one or two of these burst periods, and that the sustained downtime is relatively accurate ... RNG not withstanding, though generally accountable for.
And yeah - only tinkering with the focus regeneration talent was primarily to see the effect of arcane shot. It's the only real talent change we can make to be able to evaluate the value of those shots.
Careful Aim will always apply to test dummies, yes. All your steady shots and all your aimed shots will indefinitely crit on the target dummies. These have to be evaluated down and accounted for, but luckily since it's not theorycrafting / simming it's fair enough to just eyeball it based on personal judgement and the values in your combat logger.
I'm not sure what addon you're referring to. The bottom centre of my screen is bartender, and above that are my ability timers in Event Horizon.
I made a post in the PTR hunter thread that explored this a little bit more, including some more 7.5 turret tests - this time with BM included:
http://us.battle.net/wow/en/forum/topic/9377368613?page=23#453
Was this what you're referring to? Because I had clips up from that using Skada to keep track of damage ratios. The ability breakdown was just Skada.
EventHorizon was the addon I was wondering about! I'd never seen that particular addon.
Delete