The Case for Haste

Tue Jan 13, 2015 2:08 pm

Re: The Case for Haste

I found this interesting as well; the only time haste should be bad for HA+Sera is during BL when we go over the 50% cap. I would think that being able to squeeze in 15 SotR's during Seraphim (1 per GCD/every second) would be optimal in terms of survivability. Then again, it could also just be the downtime during the HA CD that devalues haste...

Quietmode

Tue Jan 13, 2015 2:14 pm

Re: The Case for Haste

And also in his pre-hotfix graph, Haste is Amazing for Sera+HA. I'm just assuming its hitting some breakpoint, or the bloodlust time hurts it significantly.

Edit: Also how does Theck get the tables that show multiple Specs, i usually change the spec and do a new simulation each time. I'm assuming there is some code you can type in the simulate area to rerun with another spec.

Donatist

Tue Jan 13, 2015 5:30 pm

Re: The Case for Haste

I believe Theck addressed this with the graph versus pure numbers. We see that through the plots haste is stronger when CDs like HA and Sera are not active which overall makes it a better stat. What you see in Seraphim sims are wonky because of how seraphim itself sims. Could be wrong though

Thaumatrope

Tue Jan 13, 2015 5:50 pm

Re: The Case for Haste

The graphs were really useful for me – but this one, Sera_HA, is hella confusing.

As Theck said: "It looks like haste is great on one half of the plot and terrible on the other."

Still... reading the rest of that blog post, it sounds as if haste is still great even for Sera_HA, and that the graphs are simply too wonky and erratic to show it in this case. I think?

As someone who loves Sera + HA, I'd love to know for sure what's going on here.

theckhd

Tue Jan 13, 2015 7:53 pm

Re: The Case for Haste

This is one case where Scale Factors are not very reliable. Scale Factors use the delta between the data points for +0 and +350 on that plot to determine the scale factor. As you can see from the graph, the +350 data point shows haste as the weakest stat, hence the low scale factor value.

However, that's not really representative of how haste is actually performing. If we compared, e.g. the -350 point to the 0 data point, we'd find that haste is way better than the other (non-BA) secondaries. Doing the calculation that way would give us a haste value of -7 or -8.

So, what's the real value? Probably some average of those, putting it around -5 or -6. Still better than everything except Bonus Armor. Like I said in the blog post, a more accurate way of calculating the scale factors would be to average the differences between every point on that line, which is what I plan on implementing in SimC eventually. I'd say "soon (tm)," but I may not get to it for a few weeks at this rate.

If you don't understand why we'd want to perform that average, consider that this is for one very specific gear configuration and boss. Adding 500 haste rating here gives a very weak improvement. But if the gear set had e.g. 500 less haste to start with, adding 500 haste would give you an AMAZING improvement - almost as good as adding 500 bonus armor! So the value of haste will vary from gear set to gear set.

How valuable is it to you, or any other player? It all depends, you could be anywhere on that curve, so it could be great or it could be awful. But if you stack e.g. 1000 haste, you'll go through some strong sections and some weak sections, so on average it'll work out to be better than the rest of the stats you could be stacking.

Perhaps more importantly, keep in mind that I think a lot of the features on the Seraphim haste plots are unrealistic. A real player (who has variable reaction time, lag, makes mistakes, has to break rotation to do things, doesn't perfectly sync Seraphim with the GCD, and so on) plays differently than a robot. Those plateaus are smoothed out by all of those factors, such that I think the gains from haste are far more linear than the plots suggest.

theckhd

Wed Jan 14, 2015 11:22 am

Re: The Case for Haste

Some updates:

Regarding questions about glyphs, enchants, trinkets, etc. Note that I'm using a heroic highmaul gear set for these sims rather than T17H, so they'll be a little more relevant to current Mythic progression (i.e. if you aren't already farming it). Unless otherwise specified, talents should be DP+LH+Seraphim.

Glyphs:
Glyphs - Single-Target
Glyphs - Three Targets

Weapon Enchants:
Weapon Enchants - Single-Target

Trinkets:
Trinkets - Single-Target

Talents (this uses T17H again):
Talents - All L100+L75 combos

Highlights:
Glyphs
- Harsh Words still the single-target DPS king.
- Focused Shield is the next best single-target glyph.
- Final Wrath and Alabaster Shield are roughly tied for third against a single target, with FW probably slightly ahead.
- On 2+ targets, Double Jeopardy is your new second-place glyph, provided you can constantly alternate Judgment between two different targets.
- Alabaster Shield is the clear third-place winner for 3 targets. Based on how close the two were for a single target, I think it's safe to say that Alabaster > Final Wrath for 2 targets as well.
- I have no idea why Battle Healer shows a 10-40 DPS gain. My guess is statistical error, even though SimC claims these should be good to about 11 DPS. Could also be some really unlikely interaction (ex: The sim is healing the tank to 90% every 10 seconds, and BH removes the SoI healing that occasionally prevents you from dropping below the 60% threshold for Mark of Blackrock, leading to slightly higher uptime on that buff).

Summary: Harsh Words > Double Jeopardy (2T+) > Focused Shield (1T) > Alabaster (2T+) > Final Wrath = Alabaster (1T)

Weapon Enchants:
- Haste buff made Mark of Warsong the new DPS winner. Blackrock isn't that far behind though (~75 DPS).
- Blackrock has a small edge for survival, but Warsong and the others aren't at all far behind.

All of the usual caveats we discussed upthread about Blackrock and multi-tank situations still apply. In short, either enchant is justifiable one way or another, depending on what you like. Warsong probably the better choice for DPS (+100 or more once tank swaps are considered), Blackrock probably the better choice for survival (shouldn't proc while you're not tanking).

Trinkets:
- Not a lot has changed here. Bonus Armor and STR trinkets still win for DPS, stamina trinkets are still better for raw survivability. Since we care about DPS, Bonus Armor is probably the better overall choice (but obviously, use what drops).
- Worth mentioning: the Knights' Badge here is a 3/3 upgraded one, and the rest of the trinkets are Mythic. So 3/3 KB is a great starter trinket, and probably won't be replaced until you get something like Blast Furnace Door from BRF.
- Also worth noting that the on-use trinkets like Pol's Blinded Eye may be slightly preferred if you use HA instead of DP due to overlap. This sim is actually pretty fast, so I may re-run it later with Sera+HA (and maybe add Scabbard, which somehow isn't on the list of trinkets I used?).

Talents:
- Sera still king for DPS and Survival. Remember this is using T17H, so even once we get set bonuses, Sera is at an advantage provided you play it properly. If you don't play it properly, HS is the safer bet. We may want to discuss a rule of thumb like "if you can't maintain 40%+ Seraphim uptime, you're probably actually underperforming compared to taking HS."
- ES is still in last place post-haste-buff. Worse for survival than HS, and not a very large DPS gain over HS either (obv a DPS loss compared to Sera). And again, this is played perfectly juggling all 3 buffs. The penalty for error with ES is probably just as large as Seraphim since all three of the buffs are pretty significant effects. If your uptime on any of the buffs is <75%, you're probably better off with HS.