Monday, January 26, 2009

Exotic idea for Raid benchmark

Being thrust into a brand new spec, with a variety of new attacks available, I am at a heightened level of need for performance measurement data.

When we first started raiding in WotLK, Patchwerk was the perfect benchmark to measure changes to your gear, spec, rotations, DPS, healing through-put, damage mitigation/avoidance, threat output, you name it.

Long (ish) fight, zero maneuvering, easy to predict where incoming damage will land (if not 100% easy to know when).

One would not want every single fight to be like this, but I'm very happy they put one stand-and-deliver fight.

Patchy served as an excellent benchmark.

But then we got better.

What used to take 7 minutes now takes 3 minutes and 3 seconds (yeah, damn hunter nerf prevented us from getting Quickwerk achievement, /pout)

Mana is no longer an issue, so that's kinda removed from the benchmark equation.

Most abilities and trinkets are able to be used only one or two times during the fight. Maximizing output became more a matter of hyper-tuning the timing of special abilities to the known duration of the fight and confident knowledge of when heroism would be activated.

Over time, Patchwerk has become a less valid yard stick for me.

An alternative, flawed benchmark

Reading over at BRK, he often describes his own personal DPS tests that he performs on the dummies.

He generally uses the basis of "choose a rotation, fire 1000 rounds, assess results. change one variable, fire 1000 rounds, assess results."

The problem I see with that mentality is very similar to the problem I now have with Patchwerk....the time is variable.

Using the BRK method, if you stack haste, you'll come to the conclusion that mana is no problem what-so-ever. Autoshot blows through the 1000 rounds lickety split and you're off happily, but incorrectly, buying up haste gems and only considering Intellect as a means to increase your Attack Power (if you spec appropriately).

Benchmarks need to keep time consistent.

Exotic Idea

Using a boss as a benchmark is flawed because the time will vary as your teams DPS varies. Your team composition may vary from week to week also, but given a robust Raid Comp, this is less of a factor, at least for 25's.

Using a "fire X rounds at a training dummy" is flawed because (A) time is variable by your haste, and (B) no measure of cross-player interactions and buffs. To mitigate (B), raid leaders could choose to always start raids in Exodar and begin each raid with a 6 minute dance with a dummy, but I'm doubtful that would happen.

So the new and innovative idea I have is for Blizzard to introduce a twist on Patchwerk.

You know how Sartharion has a difficulty setting, whereby you kill or don't kill drakes prior to engaging the boss.

Give us a similar mechanic for Patchwerk.

Three trash blobs stand off in corners of the room, with tiny aggro radius, so its easy to leave them alone if you want.

Kill all three, and you get Patchy as we currently know him.

Leave one blob alive before engaging Patch, and at the start of the fight he will gain an Immortal buff with a duration of four minutes. If his health reaches zero while buffed Immortal, he stays alive until the buff wears off, at which point he dies immediately. If his health is above zero when the buff wears off at the four minute mark, its just as if you had killed all three blobs, keep up the pew-pew until he's dead.

Likewise, two blobs gives a 5 minute timer, and three blobs gives a 6 minute timer. Or some other options instead of 4,5,6, but you get the point.

One special case to consider, for example in the "one blob" (4 minute) option, if you bring him to 0 health at 3 minutes, but then wipe while the Immortal buff is active, don't release, he'll die the moment the buff wears off. They need to program in this special case so he doesnt reset or evade bug. Maybe have him do funny emotes or some other strange easter egg.


I wonder if raids would accept adding 3 additional minutes, and perhaps additional consumable usage (I need to drink zero potions on Patchwerk now, might need to drink a mana pot if fight lasted 6 minutes), to an encounter in the pursuit of better measurement of performance?

Most raids would want to actually ding the Quickwerk achievement before entertaining this type of scheme, but personally, as a player who is very interested in learning the capabilities of a brand new spec and selection of shots, this would be very valuable to me.


Anonymous said...

Its an interesting idea - I'm not sure it would be that as useful as you think.

Lets assume that you hit 0 health afer 3 mins and 3 sec - a sensible raid leader is going to have all healing capable classes move to the tanks (since dps no longer matters) and so your damage buff / proc issues remain.

Perhaps you could alter the mechanic so that it gives him a shield worth X hit points (or X times 2 or 3) that you have to burst through to take him from 1 health to 0.

It is tough to get that level of detail without going into the maths behind the various buffs (which is why the theory crafters are so valuable). Even tougher for a guild without a robust raid comp.

Just one other thing - when you say you use no potions during Patchwerk do you use potions other than mana potions?

Although hunters may not get the same benefit from haste as say a mage surely 500 haste rating for 15 seconds is better than nothing when your 3 seconds short of the achievement?


Amava said...


You make an interesting point, that raids might switch focus to tank survival once reaching 0 health. I might be making an incorrect assumption here, but I'm thinking that a raid that is knocking on 3 minute kill is neither struggling with keeping the tanks alive, nor reaching healer OOM points. If they choose to benchmark the fight for 4 or 6 or whatever minutes, again its an assumption, but do not think it would be necessary to change the routine for survival reasons.

However, if they were benchmarking out to longer and longer times, OOM may become an issue, and so might bring things back to what you propose, silk.

Mana is funny lately. Our raid has one regular shadow priest (sometimes a second one) and one regular ret pally for mana battery purposes. With the new SV build, we've got two hunters who go one point into Hunting Party each. I know these things dont really stack, two batteries basically cover the 20 or so mana users, but....

Mana barely budges now, even when continuously firing. Makes 4-piece T7 a non-issue (increased shot speed while in viper). I'm looking forward to fights where I will need to consider putting all 4 of those babies on.