The Mystery of the Missing Autoattacks

Ever since I first started looking into Tiger Strikes uptime, I noticed that the number of autoattacks in my logs were less than expected. For example, in my most recent kill, I had 171 autoattacks attempted in 5:04, or an average of 1.78 seconds between attacks, despite having a weapon with a 1.6s base attack speed! I had a number on theories on why that was happening. My first theory was simply time spent not attacking – time lost to channeling mana tea, and time lost to being fully petrified (since I largely look at Gruul logs).

This didn’t cover it, so then I had this wild theory that perhaps autoattack range is shorter than ability usage range – perhaps by trying to stay at max range to Gruul, I was cheating myself of autoattacks. In subsequent weeks I always made sure to stay closer, and it seemed like my autoattacks went up, and along with it my Tiger Strikes uptime. I had thought I had solved it, and didn’t look closely at it again until today. What I didn’t realize at the time is that I got my Kromog sword the following week, so the real culprit didn’t affect me as badly.

The problem is the numerous activities we do during combat that reset the swing timer. I finally looked closely at my autoattacks and the time between them, from this log. I exported the melee attack events, multistrikes excluded, to a csv thanks to wcl’s lovely export function (support Kihra’s patreon if you can!). I then added a column tracking the time since the last autoattack, a few columns to track any debuffs/buffs that would affect swing speed, and finally a column where I could track casts. Throw in some conditional formatting, and the result looked like this:

autoattacks

I quickly sussed out two main culprits: Expel Harm, and Surging Mists, which only used when at 5 stacks so it’s instant and has no mana cost. Still, there are clearly unaccounted-for delays in the autoattacks. The first such example happens at 13.410, where you’d expect a delay more in the ~1.0 range but we have 1.59. I went back to the casts table and finally found it – a Gift of the Serpent cast at 12.378, resetting the swing timer (at this point roughly 1.03) so that the next cast comes at 13.410. (Quick note if you’re looking in logs yourself, spellID 119031 is when they spawn, 135920 is when they expire after 30 seconds and heal somebody nearby, and 124041 is when somebody runs over it and gets healed). That’s right, the random spawning of orbs from Mistweaver mastery resets the swing timer. This actually means that mastery on gear, which increases the chance to spawn orbs, has a negative value for DPS.

So, at last we’ve solved the mystery. My preliminary theorycrafting (which, I was very close to getting a Jade Mists weight to finally finish the multistrike weight, I promise) indicates that this is about a 7.5% DPS loss just from lost autoattack damage, before you take into account lower Tiger Strikes, trinket, weapon enchant, and ring procs via less combat events. Once you add those in, it could well be over a 10% DPS loss.

Advertisements

5 thoughts on “The Mystery of the Missing Autoattacks

  1. So, you’re saying the delays are ALL caused by GotS, and SM/EH just increase the chance of spawning an orb, which is why those are delayed? Or all 3 cause it?

    Like

    1. All three cause it. I didn’t cast many expel harms last week, but I’ll give you examples of both GotS and SM.

      Since Instability falls at 0:11.260, the next autoattack after the one at 0:11.820 should have only lust and base haste in effect, for roughly a 1.03s swing timer. This would mean somewhere around 0:12.85 for the next AA. However, a GotS spawn at 0:12.378 resets the swing timer, and the next AA is actually at 0:13.410, 1.032s later.

      The next AA after that, adding another 1.032, should be 0:14.44. However, a Surging cast at 13.841 resets the swing timer, and the next actual AA is at 14.877, 1.036s later. Add another 1.03s to 14.877 to get 15.907, and since we finally get a swing with no reset in between it actually comes along at 15.897, 1.02s later – close enough to 1.03 that we can chalk it up to lag.

      My only expel harm cast in this log was at 1:34.616. At this point in combat, I don’t have any haste buffs besides what I have from gear and raid buffs, so the swing timer is roughly 1.35. The previous swing was at 1:33.233, so there should have been a swing at 1:34.583, but there was a SM at 1:33.372 which pushes the next swing back to 1:34.722. The EH pushes it back again, to 1:35.936, but then there are a bunch of orb spawns at 1:35.938. I don’t have the exact haste values I had on from gear/food so I’m just going to assume that it was either lag or a little slower swing than I thought, but those push it back again. The next actual swing is at 1:37.387, following one final orb spawn at 1:36.050. If EH didn’t reset the swing timer, we would have had a swing in between the SM cast and the batch of orb spawns.

      Liked by 1 person

      1. Okay, thanks! I’ll try to increase awareness about it so we can get it fixed (even if it means they nerf us by 10% to keep us where we are).

        Like

  2. P.S.

    “My preliminary theorycrafting (which, I was very close to getting a Jade Mists weight to finally finish the multistrike weight, I promise) indicates that this is about a 7.5% DPS loss just from lost autoattack damage, before you take into account lower Tiger Strikes, trinket, weapon enchant, and ring procs via less combat events. Once you add those in, it could well be over a 10% DPS loss.”

    Trinkets, weapon enchants, and ring procs all use the RPPM system, so the procs for that are irrelevant to the number of events that can proc it. I wouldn’t be surprised if Tiger Strikes made up another 2.5% though, since a lower uptime affects all damage. At such a high uptime, it does make up a significant portion of our DPS in Crane.

    Like

    1. You’re right about overall uptime impact of RPPM system procs. However, I wonder if more combat events means truly horrible RNG on it is less likely. Does bad luck protection mean that’s not true?

      I think with a fast weapon, the impact is lessened such that the impact on Tiger Strikes shouldn’t be a 2.5% overall DPS decrease. With a slow weapon, where you’re more likely to be hitting a reset event between swings, the overall impact could be even larger than 10%.

      Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s