Hey y’all - I’m doing some simming with 2p of the tier set and seeing some odd behavior in the results. This is the sim I’m looking at: 7d20903802884ae79431a1ae1cd3e465.
In that 5m sim, I’m only seeing about 45% uptime on Elemental, which felt low (given Surge of Power and Night Fae Unity leggo). I’m expecting that between the 20% increase of Lava Burst damage during Elemental from 2p and the 20% increase from the WLR legendary, we’d want to use Surge of Power procs on Lava Burst every time, both for DPS and for Elemental uptime. However, it appears that in the APL, Surge of Power Lightning Bolts are above Lava Bursts. In fact, there’s no logic for Lava Burst in relation to Surge of Power at all, that I can see. This doesn’t feel right.
This is probably what’s resulting in the DPET for Lightning Bolt to be higher than Lava Burst, which I would really doubt is the case in a normal situation:
Digging into the event log, I’m seeing some strange things. Here, it looks like we have an Earth Shock (triggers SoP and WLR) and then a Lava Burst, but the LvB does not consume the SoP buff. Strange! Moments later, we see a Lightning Bolt cast that does consume the SoP buff:
This behavior happens a few times throughout the log. Is it possible that the sim is missing the interaction between Surge of Power and Lava Burst CDR on Fire Elemental altogether?
I felt like I was going crazy parsing through this stuff. And I might be wrong! Would appreciate some guidance. Thanks for taking a look!
Actually, I think there might be problems with Elemental Shaman simming more generally. The recent SimCraft results were posted here and I was very surprised to see Ele Sham waaaaay up top. Out of curiosity, I wanted to see what AMR would report for the same character and the same(ish) kind of encounter. So, I went ahead and created this simulation (f7499c6f06e44ec99b2d713d9e9e9fa2).
SimCraft reports about 17,500 DPS and AMR reports about 13,000 DPS. This is a big enough discrepancy that it’s worth diving in. Need to be said - I really like y’all’s presentation of ability breakdowns and cast logs more than SimCraft’s. Way easier to digest!
Here are a couple of things that stick out to me:
Fire Elemental uptime is 100% in SimCraft and 73% in AMR - this is biggest indicator because the Tier set directly impacts uptime here, and uptime = more Lava Burst Damage. Tier set possibly not implemented correctly?
AMR overcaps on Maelstrom very often whereas SimCraft doesn’t - this is the part that makes me double-take the APL. I notice long stretches (especially during Ascendance, where LvB has no cooldown) where it chain casts Lava Burst and Elemental Blast, resulting in wild Mael gen with no Earth Shocks to spend.
Cast chain after Ascendance on pull. This is certainly overcapped on Maelstrom by 10 seconds in:
Thanks for the posts - I looked into it. I’m not necessarily going to try to match dps with simc… to do that I’d have to verify everything that simc is doing, which I don’t have time to do. But I can go through and verify that our simulation is working.
I found a bug with the Essential Extraction conduit that was not reducing the cooldown on Fae Transfusion, so I’ll post an update to that. I also found a bug with Lava Surge - I was trying to fix a problem with it and must have caused a new problem and didn’t notice.
You wouldn’t want to do that if stormkeeper was up!
I checked and surge of power is working with lava burst casts
Tier set is ok, this was due to the other bugs, although I’m not able to reach 100% uptime with that set of gear/talents. I’m only able to reach 100% uptime with surge of power as well.
Wasting some maelstrom has surprisingly little impact on total output because earth shock isn’t much of a spender. I put in some rotation logic to use earth shock during ascendance with windspeaker, because that give enough value to earth shock to make it worthwhile. I also moved elemental blast down in the priority in general, as it seems to be slightly lower priority now with how everything has scaled.
You’ll still see the simc simulation doing a lot more damage than our simulation… but I cannot find any further issues with our simulation. I stripped down a test rotation and set it up specifically to maximize fire elemental uptime - I was not able to reproduce 100% fire elemental uptime with the given build, but I was able to get close - above 95% in some cases. Also keep in mind that our “single target” script involves some movement/interruptions due to mechanics. You have to use the “training dummy” to do direct comparisons. The simulators are also built differently, so simc is usually able to do more total damage, but the gear that simulates best is almost always the same.
Thanks for digging in! I know that SimC and AMR won’t perfectly match (their APL might be a little overengineered for my taste, anyways), but I figured the results given the same inputs should be within 10% - hence my poking around. I believe you that both would return nearly the same recommendations on gear. I went down this rabbit hole because I was interested in possibly making custom encounter scripts for Sepulcher bosses. It’s becoming really difficult to parse “best” setups in WarcraftLogs when you don’t have the full tier set - so I wanted to possibly sim a few of them here!
Good point, and very true!
It didn’t appear to consume the buff in the sims I was looking at, but I’ll take another look. Maybe related to Lava Surge bug you mentioned - like SoP isn’t consumed on Lava Surged LvB casts?
This is the part where I start to have trouble determining priorities. EB’s value is mostly about Mael gen and the marginal gains from secondaries. ES’s value is damage per cast, but also in the reduced cast time and increased damage from LvB with Windspeakers. With tier, that LvB is even high value, but also has value in increasing Elemental uptime.
Do we try and spread out the “value” of a cast along that entire chain of events? Is part of EB’s value the slightly increased Elemental uptime, given the Mael that drives more ESs and the increased number of LvBs as a result?
A question for you - how do you all determine the APL? Is it an upfront analytical exercise, or more like simming permutations of the APL to maximize DPS? Do you all use any specific tools to help with determining priority, like a spell damage calculator that you can add buffs/modifiers to?
I’ve been a sim junky for many years - I make the APLs and edit them as necessary. I have the ability to test the rotations using our optimizer to check rotation changes against a range of gear sets, when necessary. At this point I can make a rotation fairly easily for any spec, though. The hard part is combining the rotation logic for all specs/builds into one rotation. Sometimes it gets confusing…
