Ancestral Swiftness vs. Hailstorm

[Hinfa]

Hi everyone,

Long-time reader here, first time poster.

I wanted to get your thoughts on why Hailstorm is considered to be better than Ancestral Swiftness by this community. Pretty much everyone I see is saying that Hailstorm is a cookie-cutter, while I don't find this to be true for following reasons:

• DPS: for my current gear (873 ilvl, 57% Mastery, 17% Haste from gear, 27% Critical Strike, 5% Versatility) sims (I use askmrrobot) with AS actually show a ~5K dps increase compared to HS on Ursoc (H). This can change depending on my gear, obviously.
• Ramp-up time: 1 less GCD to ramp-up can mean a lot especially in Mythic+, and on raid encounters with lots of movement.
• Ease of play: HS is an extra button you have to press while AS a passive ability. While it's definitely not a problem for top-notch raiders, for mid- and low-tier raiders it can definitely be a major benefit. Also, more attacks means more maelstorm, and, potentially, more lava lashes, which can advocate for AS even more in case you have Akainu's bracers.
• Fun aspect - it's not the most robust argument, but I personally enjoy having more SS to throw at the boss vs. having fewer, stronger attacks.

Is there anything I may be missing? Why is HS considered to be miles ahead of AS?

[Huanak]

It is because blizzard hot fix doomwind/wimdfurry.

Now we are pretty much GCD cap and cause some opportunity cost to spent 1GCD on frostnrand.

If you change simc apl to cause some imperfect uptime on frostbrand, you will lost huge amount of dps. For example, move the refresh frostbrand line after stormstrike (which is most common mistake to make when our stormstrike pro no stop and I fail to refresh frostbrand with 20s duration).

Another example is refresh with 1s left also cause dps drop.

Overall it is harder to reach simc dps with hailstorm with our GCD cap half of time

[Arachnofiend]

I've tried switching to Hailstorm on both a target dummy and during raid progression. Every attempt has led to a pretty significant DPS loss compared to Ancestral Swiftness. It feels like I waste a lot more stormstrikes trying to keep up three buffs than two.

[Rahkhun]

It is because blizzard hot fix doomwind/wimdfurry.

Now we are pretty much GCD cap and cause some opportunity cost to spent 1GCD on frostnrand.

If you change simc apl to cause some imperfect uptime on frostbrand, you will lost huge amount of dps. For example, move the refresh frostbrand line after stormstrike (which is most common mistake to make when our stormstrike pro no stop and I fail to refresh frostbrand with 20s duration).

Another example is refresh with 1s left also cause dps drop.

Overall it is harder to reach simc dps with hailstorm with our GCD cap half of time

Why is refreshing with 1 seconds remaining a dps loss and when is the actual optimal time to refresh?

[Jotaux]

Why is refreshing with 1 seconds remaining a dps loss and when is the actual optimal time to refresh?

Its fine to refresh it at 1s.

[Deleted]

I use askmrrobot

Robot has often simmed AS ahead for many people in the main discussion thread - however simcraft has always firmly had hailstorm ahead. Hailstorm is simply better, it does 15%+ on every pull and scales very well with both haste and mastery.

If you want to make a big judgment like that, I would suggest always using simulationcraft as well to make sure you have a differing opinion, although I'm sure the Mr Robot person will swoop down to tell me I'm wrong pretty soon.

Overall, AS is tempting because it isn't actually that bad and it simplifies your rotation quite a lot, but it's really up to if you want to sacrifice the optimal spec just because you couldn't be bothered to keep up another buff.

[Darleth]

Why is refreshing with 1 seconds remaining a dps loss and when is the actual optimal time to refresh?

1 second is fine. You just shouldn't rebuff above 4 seconds. Otherwise you're loosing some seconds on it because its pandemic and its max duration is capped to 21. Refreshing at 4 seconds basically rebuffs it to 20, which is the "standard" time more or less.

The other thing is weaving stormstrikes in without the risk of loosing buffs and procs.

Robot has often simmed AS ahead for many people in the main discussion thread - however simcraft has always firmly had hailstorm ahead. Hailstorm is simply better, it does 15%+ on every pull and scales very well with both haste and mastery.

If you want to make a big judgment like that, I would suggest always using simulationcraft as well to make sure you have a differing opinion, although I'm sure the Mr Robot person will swoop down to tell me I'm wrong pretty soon.

Overall, AS is tempting because it isn't actually that bad and it simplifies your rotation quite a lot, but it's really up to if you want to sacrifice the optimal spec just because you couldn't be bothered to keep up another buff.

Pssh... dont talk bad about AMR... its working as intended. Even after telling them multiple times that there is an error somewhere in there, they still decide to tell us "Its working fine!"

Sarcasm aside, AS is the better talent for AoE fights.

[Swol]

We create simulators to answer these questions for us! I have also seen the same results that the OP posted: AS is simulating to slightly higher damage than Hailstorm in AMR, but Hailstorm is simulating to higher damage in SimC. So then it comes down to: which one is right?

I'm not sure what Darleth is referring to about an error pointed out that hasn't been fixed. Last week someone pointed out that we didn't have spontaneous appendages proc'ing stormbringer. We updated that with our 7.1 stuff today, as I said I would as soon as it was brought to my attention. But, I noticed while digging into it that SimC has that trinket modeled incorrectly. It is doing too many attacks. A quick look at some logs on WCL can confirm this. I also finally got it in-game so I was able to confirm in-game as well.

Also, I have repeatedly pointed out to the SimC authors that the doom wolves are using the wrong attack speed in SimC and using their abilities more than they do in-game. So, the damage on doom wolves in SimC is wrong.

As for why SimC does significantly higher damage with hailstorm than with AS... The AMR enhance rotation gets more damage than the SimC rotation, as I've done a lot of testing on that. I think it is because I'm able to get more boulderfists and I don't use crash lightning as much. The SimC rotation has you use crash lightning as soon as you summon the feral spirits - which is good against multiple targets, but, a DPS loss against a single target if better actions are available. They are also using Crash Lightning again if crashing storm is still ticking, but has less than one GCD left. I'm waiting until crashing storm isn't up at all. This could explain part of the difference between the two talents, since I'm using less abilities that proc hailstorm, but, getting more damage as a result by using more boulderfist. Looks like SimC isn't getting enough Unleash Doom procs either - not sure if that's a missing ability that procs it or something with how the rotation is working.

So, this is a good case to show how some very detailed, specific optimizations to rotations and small bugs in our models of the game can make a large difference to theorycraft. As far as I know at this point, there are no bugs in the AMR enhancement implementation. If you find any or think there are, hit us up anytime on our forums, discord, PM me here, whatever. I think the OP's findings are perfectly valid, based on the evidence available.

[Azelas]

not only is Hailstorm just downright better DPS wise if you keep your buffs up, it also scales way better due to Mastery and not having a bazillion haste. Not only THAT, but Hailstorm is just clearly better when having the legendary ring, since you'll have to use frostbrand anyways to get the 2% buff.

[Nitwit]

They are also using Crash Lightning again if crashing storm is still ticking, but has less than one GCD left. I'm waiting until crashing storm isn't up at all.

Unless they changed something, you can have more than 1 crashing storm area DoT up at once, so why would you specifically go out of your way to ensure that the previous one expired before casting another?

[Jotaux]

We create simulators to answer these questions for us! I have also seen the same results that the OP posted: AS is simulating to slightly higher damage than Hailstorm in AMR, but Hailstorm is simulating to higher damage in SimC. So then it comes down to: which one is right?

I'm not sure what Darleth is referring to about an error pointed out that hasn't been fixed. Last week someone pointed out that we didn't have spontaneous appendages proc'ing stormbringer. We updated that with our 7.1 stuff today, as I said I would as soon as it was brought to my attention. But, I noticed while digging into it that SimC has that trinket modeled incorrectly. It is doing too many attacks. A quick look at some logs on WCL can confirm this. I also finally got it in-game so I was able to confirm in-game as well.

Also, I have repeatedly pointed out to the SimC authors that the doom wolves are using the wrong attack speed in SimC and using their abilities more than they do in-game. So, the damage on doom wolves in SimC is wrong.

As for why SimC does significantly higher damage with hailstorm than with AS... The AMR enhance rotation gets more damage than the SimC rotation, as I've done a lot of testing on that. I think it is because I'm able to get more boulderfists and I don't use crash lightning as much. The SimC rotation has you use crash lightning as soon as you summon the feral spirits - which is good against multiple targets, but, a DPS loss against a single target if better actions are available. They are also using Crash Lightning again if crashing storm is still ticking, but has less than one GCD left. I'm waiting until crashing storm isn't up at all. This could explain part of the difference between the two talents, since I'm using less abilities that proc hailstorm, but, getting more damage as a result by using more boulderfist. Looks like SimC isn't getting enough Unleash Doom procs either - not sure if that's a missing ability that procs it or something with how the rotation is working.

So, this is a good case to show how some very detailed, specific optimizations to rotations and small bugs in our models of the game can make a large difference to theorycraft. As far as I know at this point, there are no bugs in the AMR enhancement implementation. If you find any or think there are, hit us up anytime on our forums, discord, PM me here, whatever. I think the OP's findings are perfectly valid, based on the evidence available.

I've asked this every time I've seen you post, have you fixed your gear optimizer so it looks at caster trinkets for enhance yet?

[Yellowfive]

All rankings for all items have been updated for 7.1 -- it should rank any special effects that can be reasonably used by enhancement shamans.

[Swol]

I've asked this every time I've seen you post, have you fixed your gear optimizer so it looks at caster trinkets for enhance yet?

The simulator has been able to use caster trinkets for weeks now. Our pre-calculated gear rankings were updated for 7.1 and include rankings for appropriate caster trinkets. Of course, kara trinkets are still subject to tweaking, especially since we don't know how the chest piece interacts with them exactly yet.

- - - Updated - - -

Unless they changed something, you can have more than 1 crashing storm area DoT up at once, so why would you specifically go out of your way to ensure that the previous one expired before casting another?

For all practical purposes, it doesn't matter a ton. I was just trying to figure out why SimC is using crash lightning 25% more often. That might not actually be the source, I think I misread the SimC APL. I'm allowing Crash Lightning to be used, no matter how long you have left on crashing storm, just as a low priority. I have a higher priority action to use crash lightning on a single target if you don't have crashing storm up at all, or if you don't have gathering storms up. I don't prioritize crash lightning over refreshing frostbrand/flametongue unless you don't have crashing storm or gathering storms up. SimC prioritizes crash lightning over the refresh no matter what. Maybe that is leading to a different timing of abilities.

The way Crashing Storm works isn't that there are two up. If you use crash lightning again while crashing storm is up, it will reset the tick timer and make a new one, but the new crashing storm lasts 1 second longer. The net effect is the same, but the timing is different than just overlapping them. You can test this by using crash lightning and putting a crashing storm under one enemy, then run over to a different enemy (out of the original crashing storm) and crash lightning again on that one ASAP. You will see that the first target takes 6 ticks of crashing storm, and the second target takes 8 ticks. It doesn't overlap them, it just makes the second one tick an extra time.

[ehxy]

I pull 330k ST no buffs so I'll test this out later it's the only way to be sure but i get the feeling this'll be one of those better on AoE than ST situations.

[Deleted]

I'm going to use Ancestral Swiftness until I get the legendary ring... or well if I do. Hailstorm feels so clunky to me, I need that extra haste and I do beat almost every other enhancement I play with so I don't know what's up.

[RuneDK]

I must suck with hailstorm...I'm personally doing better with AS over it...will stick with that for now.

[Sanitee]

I tend to use ancestral swiftness for dungeons and am seeing similar results but in raids, hailstorm pulls way ahead. I have 33% with the talent, 74% mastery, the trinket and neck from cos and arcway and the legendary that gives you haste increase when you cc. It's a LOT of fun when these proc together.....

Raid wise, hailstorm still comes out way better for me.

[Darleth]

We create simulators to answer these questions for us! I have also seen the same results that the OP posted: AS is simulating to slightly higher damage than Hailstorm in AMR, but Hailstorm is simulating to higher damage in SimC. So then it comes down to: which one is right?

I'm not sure what Darleth is referring to about an error pointed out that hasn't been fixed. Last week someone pointed out that we didn't have spontaneous appendages proc'ing stormbringer. We updated that with our 7.1 stuff today, as I said I would as soon as it was brought to my attention. But, I noticed while digging into it that SimC has that trinket modeled incorrectly. It is doing too many attacks. A quick look at some logs on WCL can confirm this. I also finally got it in-game so I was able to confirm in-game as well.

Also, I have repeatedly pointed out to the SimC authors that the doom wolves are using the wrong attack speed in SimC and using their abilities more than they do in-game. So, the damage on doom wolves in SimC is wrong.

As for why SimC does significantly higher damage with hailstorm than with AS... The AMR enhance rotation gets more damage than the SimC rotation, as I've done a lot of testing on that. I think it is because I'm able to get more boulderfists and I don't use crash lightning as much. The SimC rotation has you use crash lightning as soon as you summon the feral spirits - which is good against multiple targets, but, a DPS loss against a single target if better actions are available. They are also using Crash Lightning again if crashing storm is still ticking, but has less than one GCD left. I'm waiting until crashing storm isn't up at all. This could explain part of the difference between the two talents, since I'm using less abilities that proc hailstorm, but, getting more damage as a result by using more boulderfist. Looks like SimC isn't getting enough Unleash Doom procs either - not sure if that's a missing ability that procs it or something with how the rotation is working.

So, this is a good case to show how some very detailed, specific optimizations to rotations and small bugs in our models of the game can make a large difference to theorycraft. As far as I know at this point, there are no bugs in the AMR enhancement implementation. If you find any or think there are, hit us up anytime on our forums, discord, PM me here, whatever. I think the OP's findings are perfectly valid, based on the evidence available.

Swol, I'm not trying to hate on your work or anyone elses work that is associated with AMR. You guys put a ton of hours into it and its a good thing that we have people like you working on this kind of stuff.

But its just simply wrong that AS pulls ahead on singletarget fights. Unless you have 5% haste or something like that, its just not possible. The damage done by Hailstorm is just SO HIGH, that it just CANT BE worse than AS on a pure singletarget fight or during small burst windows on priority adds. Hailstorm is LITERALLY the 3rd highest damage source in a fight. Every log shows this, but your simulation doesn't - so there must be an error somewhere there.
Considering another factor, where we have a patchwerk fight, it makes even less sense. If you're able to just STAND AT THE BOSS and just nuke it, Hailstorm will always come out as the alpha wolf (pun intended). You really need to have EXTREMELY low Hastelevels to not be able to play with HS and for this to happen, you have to be one of the most unluckiest person on the planet that rolled everything but haste on his gear.

Hailstorm also scales double, higher haste = more attacks = more procs and higher mastery = more damage. While haste only really increases our chance to proc SB and WF more often and use stuff more often. That factor alone just shows that there MUST be an error somewhere in there, because mastery scales way higher and even has the chance to be better than agility with certain trinkets and stats.

[Swol]

Hailstorm does show as the third highest source of damage in simulations though... Here's a typical set of heroic gear with high mastery:
http://www.askmrrobot.com/wow/simula...bb45d42e2cdcda

Hailstorm actually does more damage in AMR than SimC. We have more procs. Maybe SimC is missing an ability that procs it, it's only different by like 20 out of 770 or so total. (Edit: looks like it is just because of how the two rotations are different. Using SimC's rotation, I see the same number of procs in both simulators.)


Ok, I think I found the issue. I still had the attack speed increase on Ancestral Switftness that they took out a long time ago. Must have just forgotten about it, since I was always using Hailstorm for testing. Now I see a big difference between the two talents. Hailstorm has always been working correctly, though.

The interesting thing is that if you leave the 10% attack speed buff on, the two simulate to almost the exact same damage. Maybe blizzard was onto something before they nerfed it... heh.

[ehxy]

Whichever simulation says AS is better is simply wrong and I don't care if they have numbers to prove it because I have in game useage over a 10 minute long rotation on a test dummy.

It might play better in an AoE situations where more actions per minute are a priority over harder hitting but when it comes down to single target wordups numbers still stand and hailstorm is very clearly much better than haste no matter what they say and what was established by wordup since his retraction of 60% of your mastery as haste being desired.

HAILSTORM.

Anyone tells ya different tellem to fuck off and that includes the AMR or SimC people.

Anyone even see what the hell we were doing with 90+ % mastery with the buff from the class hall? The shit was absolutely insane I breathed on things and they died. Haste would not do that.

And I just want you to know I have an i870 spontaneous appendages so that's included in my personal testing on a raid dummy so that being a factor with haste still doesn't even come close to touching hailstorm.

I know the shaman like I know my coffee and have the hailstorm/FT rebuff practically rocking in my internal clock and I realised I haven't done a real dps crunch on dummy in a while because I was rocking 400k for a good 3-5 minutes which is huge considering the last time I was only doing 300 the last time.

Then again I didn't haver spont. appen. then the last time.