Legion Theorycrafting [7.3, All specs]

[Effinhunter]

Any simulations on how the Flanking Strike buff will affect the numbers and priority for SV hunters?

Right now, I'm trying to catch everything up for MM and the basic spec comparisons before most people start raiding tonight. You might want to check Azor's guides for SV information. Not sure if he's addressed those changes yet.

[Deleted]

Interesting simulations for 3 target and 6. Maybe I missed them before. Did something change? Sidewinders is SO low on those sims. And on top of that, non PS builds!

Is a non sidew build now raid worthy?

[Effinhunter]

Any simulations on how the Flanking Strike buff will affect the numbers and priority for SV hunters?

Here is the DPE I'm seeing while I'm doing spec comparisons. Unfortunately, I realized something about the sims I was running that I wanted to change after many of them were done. So, unfortunately, no more sim results until after raid. However, here is where Flanking Strike positions itself in DPE after the hotfix.



- - - Updated - - -

So, I've been testing out trinkets and abilities in game and on bosses to see what I want to use on the mythic bosses, and I've been noticing some pretty big discrepancies between the Simcraft models of the trinkets and what I'm seeing in game.

The most obviously wrong trinket I have found is the Six-Feather Fan. I bought this second day of expansion, long before we really even had reasonable trinket sims (I had no idea if it would be any good at the time), and my damage from the fan proc has ranged from 0.89% to just under 3%. An even lower ilvl version of Devilsaur's Bite proc (which sims under Six-Feather Fan) is usually between 2.5% to 6% of hunter dps. One of my other hunters wears this exclusively since I logged with it a week ago, and he's seeing the same thing. Naraxas does seem to be as powerful as it seems at 20y+, but I'm sure there are other trinkets. Has anyone been trying out some of the lesser used trinkets (such as Egg of Serpentrix) to see what it is actually doing on bosses with the proc? I have a feeling that we need to do more testing of the trinkets and give some feedback to the Simcraft team (assuming we want these to be accurate).

I've also been trying out Crows on single target. I tried it on both heroic Ursoc and Nythendra tonight, and Barrage on single target was pretty much wiping the floor with Crows on CD. I also tried swapping between the talents on mythics, and looking at the damage breakdowns, something seems off in what the sims are saying (lower actual crows numbers than what is showing up in sims). Barrage usually comes in at 9-10% of hunter damage. Crows seems to range 7-8.5% of hunter damage. I'm not sure what exactly might be wrong in the models right now, but to me, the numbers aren't adding up. Has anyone else been doing their own testing of Crows vs. Barrage in game, especially on raid bosses? What percentages are you seeing of your dps?

[Chickensoup23]

Serpentrix is a gamble. It can do 0.80% of the dps and it can do almost 7%. Sometimes I will get no spawns and sometimes it will spawn 3 at the same time that will pelt my target for 60k dps on their own (850 version). I don't like the RNG of it but it's better than the blue stat stick I had before. Naraxxas & Devilsaur's seem to really be where it's at for pre-raid items.

Crows seems to range 7-8.5% of hunter damage. I'm not sure what exactly might be wrong in the models right now, but to me, the numbers aren't adding up. Has anyone else been doing their own testing of Crows vs. Barrage in game, especially on raid bosses? What percentages are you seeing of your dps?

Keep in mind that there is a channel time to barrage, which means we don't get to do anything while it's being casted. Could it mean that crows is less overall % damage as we have more time to dish out things like additional aimed shots with the focus we get to keep from barrage, kind of like a trade? Then again, I could be wrong but this seems to be a logical explanation. I had decent numbers on Nythendra and Ursoc with Amoc but barrage still feels like the right way to go, especially when you need on-demand AOE.

[Effinhunter]

Keep in mind that there is a channel time to barrage, which means we don't get to do anything while it's being casted. Could it mean that crows is less overall % damage as we have more time to dish out things like additional aimed shots with the focus we get to keep from barrage, kind of like a trade? Then again, I could be wrong but this seems to be a logical explanation. I had decent numbers on Nythendra and Ursoc with Amoc but barrage still feels like the right way to go, especially when you need on-demand AOE.

Did you have other hunters in your raid running Barrage when you were running Crows? Did you have a clean kill from last week to compare it to with your own numbers?

We've had situations in the past where abilities like AMoC in simc were not doing the same number of ticks, for instance, that they were in game. When haste was a bigger factor, especially, in RPPM and what not, we had some issues with Simcraft not matching the number of ticks that were expected (usually, the engine would give an extra tick which would skew the results). I remember this specifically happening with Dire Beast in SoO, for instance (DB was getting an extra attack in sims). Someone earlier in the thread mentioned Crows did not seem to be as good as the sims were saying they were, especially in comparison to Barrage. I tested this out tonight in raid, and Crows seemed to be at least 1% behind Barrage's contribution to overall hunter dps. It's not that the ability was bad, but it definitely seemed behind.

Obviously, I need to do more testing on these suspicions. It's the first day of the change to Barrage. Hopefully, others are looking into this too.

[satimy]

Trinkets seem like such a crapshoot. The fulminate charge and mastery stat stick or just double mastery stack sticks are probably the best.

[Feanoro]

Trinkets seem like such a crapshoot.

It's almost like they're trying frantically to rein in item and dps inflation, but not sure how to do it.

[Thursley]

This hotfix seems to have exacerbated the Barrage cancelling bug, as I'm noticing it happening in raid content where it didn't use to occur. Previously barrage was cancelled upon entering a mana rift (from the suramar buff that feeds you ancient mana a few at a time) and binding shot triggering, plus a few other assorted less common things, but now it's happening on raid bosses where I'm not using any of that. Another thread said that it was related to the six-feather fan trinket but I haven't been able to test it yet. If you have that trinket give it a shot and see, and submit bug reports for it. It hasn't always been this apparent but barrage has had this bug in some form for certain effects for years now, when it's happening in raids maybe they'll finally have enough incentive to fix it.

[Deleted]

This hotfix seems to have exacerbated the Barrage cancelling bug, as I'm noticing it happening in raid content where it didn't use to occur. Previously barrage was cancelled upon entering a mana rift (from the suramar buff that feeds you ancient mana a few at a time) and binding shot triggering, plus a few other assorted less common things, but now it's happening on raid bosses where I'm not using any of that. Another thread said that it was related to the six-feather fan trinket but I haven't been able to test it yet. If you have that trinket give it a shot and see, and submit bug reports for it. It hasn't always been this apparent but barrage has had this bug in some form for certain effects for years now, when it's happening in raids maybe they'll finally have enough incentive to fix it.

some trinket procs are confirmed to cancel channel abilites. I dont think its barrage.

[Azortharion]

We tested Crows a while back (beta) and it always did 16 ticks regardless of anything else, and that's how it is in SimCraft.

[Thursley]

some trinket procs are confirmed to cancel channel abilites. I dont think its barrage.

Ah, okay. I wonder if all those other things I've seen cancelling barrage also cancel other channeled abilities too, but I just never noticed because barrage is the only one we have. Hopefully they can fix them all together and not just the trinkets if so.

[Youn]

Trinkets are just plain frustrating this time around.

Example: Last night, I got 850 Twisting Winds. It seemed like it should be an upgrade from Three Toed Rabbit (835) or Fan (835). I switched it out for my Three Toed trinket, my mastery dropped from 97% bonus to 92% bonus. The Twisting Winds only worked out to be about 2% of my damage. This pretty much ended up as a net loss in real dps. Looking at the Fan it's 3.5% of my dps. So, I guess the two 835 trinkets just work out better.

[Effinhunter]

What the Simcraft team really needs are some naked combat logs with dummies (the same kind of logs that we did to help them fix Twisting Wind and Horrorslime trinkets). If we have the right underlying RPPM rates, then some of the problems can be fixed.

I'm going to try to do one today (if I'm not interrupted too much) for Six-Feather Fan, and I may just use barrage over and over again for an hour or two. This kind of naked test should show if Six-Feather Fan is one of the culprits of the cancel or if it is something else, and it should also provide the underlying RPPM data (if the test is long enough). Another rumor about the barrage cancel bug has been that barrage attempts to cancel if one of its targets is being cc'd after the barrage has started. I'm not sure if that is true, but if so, it could be someone else's ability/trinket that is doing it too. If it were true though, this would be highly strange as this has been rumored to happen on boss fights with no apparent cc adds visible. So, if it were true, it would mean that the proc/usage of the cc ability was causing the cancel and not the effect (basically, it would be doing it on cast rather than on hit).

I think the other proc-based trinkets need to be similarly tested, one-by-one. I have some 825-835 trinkets in my bank, but I still haven't seen a single Naraxas drop for me. I may just try to do a trinket a day. If someone has some time and the patience to do these, it's very appreciated. We just need to share the naked logs in Warcraftlogs with the Simcraft team.

As for Crows, I think I might try a different test of crows and see if I'm seeing the same crit chance and ticks that we're assuming in Simcraft. I did try a tick test earlier in the expansion, and I didn't see any problems, so you're probably right that it is doing the right number of ticks.

- - - Updated - - -

So, I've tested Six-Feather Fan with Barrage on CD for ~24 minutes, and here is what is going on.

https://www.warcraftlogs.com/reports...casts&source=1

Nakedness (892 weapon, 840 six-feather): http://imgur.com/qztta9U

Six-Feather Fan procs on an auto-attack. If it procs on either of the 3 autoattacks before you hit Barrage, it will cancel it. Six-Feather Fan doesn't just hit once, it hits 6 times over a span of time (not immediate at all). The span can take between 4 and 6 seconds. The first wind bolt happens immediately. When the others proc is a bit random. I've seen it proc and then bunch everything together between 4-5 seconds later. I've also seen it reasonably uniformly distributed every 1s over 6s. See 00:02:24.523 in my log for an example that cancels barrage and lumps the later 5 procs together in the 4-5s post proc.

I'm also seeing some instance where if wind bolt procs a second time while the first proc is active, the first proc appears to be canceled and the new proc starts its 6 wind bolts. There appears to be an example of this at 00:03:34.385 in the log. There are 11 total wind bolts here that end at 00:03:43.339. What seems likely is that the first proc got 5 bolts off and then another proc occurred which reset the ongoing wind bolt spam at 5 and then did its own 6 bolts.

- - - Updated - - -

Six Feather Fan is implemented in the unique_gear_x7 file (https://github.com/simulationcraft/s...ue_gear_x7.cpp). Right now, it is using a BUFF_TICK_CLIP (https://github.com/simulationcraft/s...r_x7.cpp#L2419), but I couldn't get anyone to verify if this is actually implementing the behavior of clipping previous 6-bolt procs or if it is meant to be a buff clip (or if it is even implemented correctly).

[Sazda]

great work effin

very odd behavior of that trinket (semi random timings of bolts) - and the cancelling it just silly as well.

[Effinhunter]

great work effin

very odd behavior of that trinket (semi random timings of bolts) - and the cancelling it just silly as well.

We've been doing some more testing in guild. It definitely cancels the DH eye beam channel also. This trinket probably cancels all channels, so let your guildies know. It is going to really affect classes that attack and also have channels, so it may also affect monks, but none of ours are on right now to try that out.

P.S. You can induce this by waiting for a wind bolt initial proc and then channeling immediately. It will always cancel it. Every time.

- - - Updated - - -

Edit. P.P.S. I'm wondering if this happens with any trinket that does more than 1 proc like this. I would assume Devilsaur's Bite is fine because it procs one time for big damage. What other trinkets are worth trying?

[BendakWoW]

Would a dummy log help for Bloodthirsty Instinct? I have no idea how the "procs more often on targets with lower health" thing works, but I am pretty sure the proc rate is way higher than it should be in simc.

[moonrakerx]

feel like the trinkets are such a mess. I have double 850 mastery stat stick and I'm just staying with that...at least I know what I'm getting.

[Azortharion]

Statstick trinkets remain some of the best in the game so.

[Wanger]

Would a dummy log help for Bloodthirsty Instinct? I have no idea how the "procs more often on targets with lower health" thing works, but I am pretty sure the proc rate is way higher than it should be in simc.

Yeah I've been having issues with my Bloody Instinct as well... seems like up time across 5 bosses was around 3.8% ... seems very low

[Effinhunter]

Would a dummy log help for Bloodthirsty Instinct? I have no idea how the "procs more often on targets with lower health" thing works, but I am pretty sure the proc rate is way higher than it should be in simc.

That's a lot harder to test. The way it is currently implemented in Simcraft is here:

https://github.com/simulationcraft/s...r_x7.cpp#L1790

The RPPM increase is currently being modeled like this:

double mod = ( 200.0 - s -> target -> health_percentage() ) / 100.0;

Was this tested? No idea. If they did, I'm not sure how they tested it. It seems to me that we can't really test the scaling of RPPM via the dummy. We would need a controlled test of an add that's health reduces very slowly over time. You would probably need to just camp a world boss with you, a tank, and a healer and the logs would still need to be 1-2 hours probably. So, you'd just have to wait for the next spawn and then kill it again until we have enough data. That's not fun, of course, but I'm not sure how else we would get the representative data.

- - - Updated - - -

P.S. What PPM are you currently seeing in Bloodthirsty Instinct in your logs? What we see in the code should start at 1 PPM at 100% health and by 10%, you would see (200 - 10 = 190) / 100 = 1.9 PPM. Reia says he doesn't remember or know how this trinket was tested to get this formula. It's very possible that this was arbitrarily inserted. So, what PPM are you guys seeing in your logs? Assuming burn rates on bosses are fast at the beginning and the end, this should make the effective PPM ~1.5. If it's not really proccing at that rate, then I guess that would tell us something. Though, if your parses are doing lust/cds toward front of fight, the PPM would shift more toward 1.6 (you spend more time toward end of fight in less health), and if you use lust/cds at the end of the fight, the PPM should shift more toward 1.4 (you spend more of the fight in higher health since you're burning down the boss at the end faster).

However, this would only really cover the situation on single target fights. For AoE situations, the PPM would be all over the damned place, and if the trinket actually is proccing more at low health, the best way to game it would be to click targets that are lowest health so your auto attacks have the best chance of proccing.