[5.2] PTR Standardized DPS Testing - Come Help!

[zomgDPS]

A Call for Standardized Testing

Ok nublets, if you are quite done getting your knickers in a twist I suggest we all calm our tits so that we can all start being constructive.
If you want to REALLY help the mage class outside of b!tching and moaning, please join me and a few others here in some standardized testing on the PTR.
Seriously, right now we need some hard data and if each one of you contributed to it, I'm sure we can get somewhere.

The problem right now is that we have very little in terms of theorycrafting ability. SimC is nowhere near up to date as far as being able to crunch numbers for us, and neither is some of the tools we use in our guild (at least until Logix gets them up and running, which, given his temperament of late, may not be for a while).
So really, all we have now is hard testing. (oh how we miss thee Zal!)


The Setup
Please create your standardized template.

0. Download and install the latest version of recount

1. Create a template toon.

2. Get the Tier gear from the massive cow vendor at the vp/jp vendors in pandaland.

3. For the non-set slots - get the tyrannical PvP gear from the big cow vendor in vale.

4. Gear/Gem/Chant yourself (the stat weight priority is up to you - the more variety we have the better, it will allow for better comparisons, just make sure your hit is hitcapped for bosses). Make note of your gearing strategy (went all crit, all haste, all mastery, etc etc, whatever it is, note it down).

5. Find a boss dummy (no cleaves please) n.b. the dummy in SW works well (for ally), so do the dummies in silvermoon (for horde)

6. ONLY SELF BUFFS. No flasks, no elixers, no pots. Just your armor and AI


The Tests
Please carry out the following tests:
(note: you can use the stopwatch found in the clock in game for precise timing)

1. "Cruise" speed
Duration: doesn't matter, its purely sustained. But you should be able to hold the DPS you do in this test in a stable way somewhat indefinitely.
Method: This is the test of the 'sustained' DPS. I.e. DPS outside of cooldowns using nothing but the spec's regular rotation and your bombs. (For frost, this does NOT include Orb).


2. "A Stormwind Minute"
Duration: 2 mins.
Method: Use any cooldown you wish except for warp.


(optional) 3. "Patchwerk"
Duration: 5 mins.
Method: Use everything you got.

The Report
Please report as follows:

Build: [Spec] | [level90talent] | [stat]
Test: which test you did (1 for "cruise", 2 for "Stormwind Min", 3 for "Patchwerk")
Result: DPS
Strategy: A short description of what you actually did


Example:

Build: Arcane | Invocation | Mastery build
Test: 1
Result: 56-59k
Strategy: An experimental strategy.
I work in 1 min time-frames (Invoc buff).
For the first 15-20 seconds of the buff, I try to stay at high mana levels (using AB only twice and using any AM charges to get to 4 stack, and then use ABr). This way I can get 1 or 2 fully stacked ABrs deployed at high mana.
When I have around 20 seconds left on the buff, I start burning harder. Getting ABs to full 4 stacks and cycling in AMs (not using ABrs at this stage). I keep burning AB/AMs till low mana and try to be at 10%ish mana by the time the invoc buff is wearing off, then I evo to full, reset buff, and repeat.


Get testing nublets. Chop chop.

[Swizzle]

I'll try doing some tomorrow. Today is Valentine's Day, which means I'm not touching any video game things. I'm on the PvP server IIRC, so I'll be testing there.

[Shangalar]

I'll see about it over the weekend. Was ill a few days so have to catch up with training and uni stuff plus got some b-day events to pay for...

[zomgDPS]

Build: Arcane | RoP | Mastery build
Test: 1
Result: 60-62k
Strategy: A bog standard strategy.
Good 'ol ABx4 AM ABr. Nothing special here. Build up to 4 stacks with AB, use your AMs, slap your ABr at the end. You will usually have at least your 2nd AM and the ABr deployed at full mana due to RoP regen (note, this will prob not be the case next PTR build where AB mana cost goes back up to 150.. we'll see)


A note on my bomb usage.

I'm using Frost bomb (but either should work). I'm timing my bomb to be available to cast when I have a full 4 stack.

If I'm at 4 stack with 2 AM charges, I use 1 AM charge, then cast my bomb, in the hope that the bomb cast gets me an extra AM proc. I use the same logic to actually cast fireblasts at max stack, hoping for the fireblast to cause an AM proc. This can lead to some interesting chains, e.g. say I have 2 AM procs already

ABx4->AM->FrBomb(*proc)->AM->Fireblast (*proc)->AM->AM->ABr

Allowing up to 4 AMs per cycle.

Not sure if its optimal or not (I know weaving in your bombs to 'fish' for AM procs is optimal, just not sure about using Fireblasts too, not yet at least), but having a cycle with that many AM procs is notbad.jpg

Today is Valentine's Day, which means I'm not touching any video game things.

Sh!t.. I knew I was forgetting something...

[Shangalar]

I have reservations about using fire blast in a standard rotation, not coz it doesn't fit, but... let me just scribble random stuff.

AB does more damage than FBlast at all levels. Barrage does more damage than FB as well. A Bomb does more damage than FB so its use to proc Missiles is warranted and often helpful, but wasting a gcd for a low damage spell with a 40% chance to proc our top damage spell... maybe in some special circumstances like when you know you won't be able to dps for some time due to reposotioning or immunities so you try to pump as much juice as possible before the event.

[Deleted]

you might want to add that people should pick a dps-neutral race. (kind of obvious but better play it safe)

[Shangalar]

Dps neutral race isn't such a big deal here since we're looking for tests without cooldowns and on the tests with cooldowns the dps provided by special abilities like berserking is also valuable information.

[Deleted]

1% haste is not important? O_o

[Shangalar]

Dps neutral race isn't such a big deal here since we're looking for tests without cooldowns and on the tests with cooldowns the dps provided by special abilities like berserking is also valuable information.

I will go so far as to make a post quoting my own post. It's not my desire to "troll" you, but I have at no point indicated that it is definitively irrelevant.

[zomgDPS]

I have reservations about using fire blast in a standard rotation, not coz it doesn't fit, but... let me just scribble random stuff.

You could end up being right, but the way I'm rationalizing it for now (again, I can't crunch the full numbers yet), is that Fireblast actually does more damage than Scorch (and scales better), and has the same cast cost (a global).

Scorch worked out being spammed at max stack, purely for the AM proc gain.
The idea is to capture some of that play and bring it into the spec fulltime.
Now sure, in some situations it doesn't work, especially when I'm just burning away my mana to get it low so that I can evocate for the buff, but it helps out somewhat in this experimental strategy of mine where I am actually trying to maintain a high level of mana for at least a few cycles before I start to burn it away for the buff.


The thing is, I actually tested out quite a few other strategies, neither performed as well as this random "conserve for 2-3 cycles then burn" thing that I have going on in my first report.

I tried a simple, cyclical ABx4 AM ABr, just burning away my mana till I get low and evocating it up and hoping the buff timer lines up to when I get low. I even tried a full burn with AM cycles (I ended up needed to evocate half way through the Invoc buff) and neither strategies beat the one I have up there right now.


What I'm hoping is that we get a LOT of data to work with, so we can iron out the inherent RNG-yness of what is going on here. Just having a cycle where you get unlucky with AM procs makes a huuuge difference.
I'm hoping to beat this RNG with pure numbers, which is why we need many people to test.



What I am worried about right now to be honest, is the fact that this patch will go live and Arcane will not be in a well tested spot (just because its being iterated on for so much). It looks like a lot of the rest of the patch is ready to go and I'm quite certain they will not hold back the patch just because 1 spec is not ready.

All that means is that at least for the first part of when this patch will be live, Arcane as a spec will be broken/non-functional. This is not a good place to be, especially after having so much dev effort being put in during the PTR phase.

[Deleted]

I will go so far as to make a post quoting my own post. It's not my desire to "troll" you, but I have at no point indicated that it is definitively irrelevant.

Nor did I. the hole point (and title) of this thread is to have standardized dps testing. the results will differ greatly (latency/luck/skill you name it) even with the exact same gear. there should be an arrangement for these little details to improve the results. I know that the impact of racials are not that big but there is more. Every template toon is jc and enchanter. Are we supposed to use the jc-gems and ring enchants?
A worgen mage with jc gems and ring enchants has 600 crit rating and 640 int more. to me thats a big deal

[zomgDPS]

- snip -

Racials are irrelevant for the intent and purpose of this thread and the tests we are doing herein.

For the kinds of things we are trying to discern (overall casting strategy), racials are not going to make as much of an impact as what strategy you use. That is what I am interested in.

As far as everything else is concerned. Yes, go ahead and gem/chant yourself fully.


That is the context of this thread, to go out there and do tests. It is not to come in here and spam the thread with stuff that is not relevant right now.

So yea.. either get out there and help, or don't. This thread isn't your personal podium to attempt to educate us on the theoretical value of 1% haste. Trust me, everyone here knows that value, perhaps even better than you.

But we are not trying to discern that value right now. IN time, the tools we need to perform the optimization step will take care of racials. Right now we are not optimizing. Right now we are discerning strategy.


Take it for what it is, or leave the thread. Your choice.


Either way. Discussion of altering the test setup is closed and not up for debate. If that is your problem with this thread, move on.

[Deleted]

I never ment to upset you (or anyone else). I think the thread's title must have misled me. I am sorry. But there is no reason to be such a diva :P

Build: Arcane | evo | haste
Test: 1
Result: 59-61k
Strategy: ab(4stacks) --> am --> abarr

tried to use abarr before evo (shorter burnphase/more evo) but did not notice a difference...

Build: Arcane | evo | haste
Test: 1
Result: 60-62k
Strategy: burn --> evo --> burn... (no abarr)



refreshing NT right before evocating makes it quite easy to fish for procs between the next 2 am procs

[Llewllyn]

I'm on my phone so I can't look right now. But from what I remember fire last costs over 10k mana which is a huge increase from the 300 that I think scorch costs. Also scorch was used rototionally not b/c it would proc AM but because it was a regen phase that also did dmg. While the AM proc was beneficial it wasn't the primary reason for using Scorch. It was the mana regen gained from using an almost mana free spell.

[Elegiac]

Build: Arcane | Invocation | Haste
Test: 2
Result: 58 - 65k
Strategy: I did two tests using this build; first, I did a straight burn style of ABx4 then spamming down to Evocate range, using procs of AM when they popped. Then, I did the standard ABx4 AM ABarr, and found that the DPS was actually markedly higher; what it would seem to indicate is that leaving ABarr out of the rotation is a DPS loss.

---------- Post added 2013-02-14 at 12:28 PM ----------

Build: Frost | Invocation | Haste
Test: 2
Result: 62 - 68k
Strategy: Using a standard rotation, the first test with Nether Tempest and the second with Frost Bomb. FB seems to ultimately result in more damage due to it making Brain Freeze procs predictable - the difference in FoF acquisition is very noticeable, and using Nether Tempest there are periods where you'll just be spamming Frostbolt constantly.

[zomgDPS]

I'm sure I don't need to mention this, but just in case, I would ask you guys to obviously do multiple runs of a specific test before averaging/ball-parking the results, and reporting back here.
Just so you don't get caught with bad/good RNG.

That being said:


Build: Frost | IW(passive) | Mastery build
Test: 1
Result: 62-64k
Strategy: Frost bomb for FFB procs (personally, I really don't see myself ever giving up the reliable FFB proc with frost). Use procs as they pop. Very simple. Good results.
n.b. No orb. Though using the orb doesn't change things THAT much, since I took extra care when doing my orb runs to ensure they don't over-generate procs due to hitting other dummies or those silly rabbits that flutter about in SW.

[nimryas]

Brilliant idea, I would help out but don't have the time right now (or in the next 2 weeks). I do however, think that this is a little too late :< 5.2 will hit in about two weeks and I doubt you would have gathered enough data to support any of your hypothesis. I also doubt that blizzard will change numbers so soon before patch release, and if they would ever consider to even use it :/

Of course this data can be used to after patch 5.2 to motivate blizzard to put some extra effort in the mage class for once, rather then randomly assigning new spell coëfficiënts.. If i can find the time, I will definitely help out! But I would have to fix a PTR client first and all that ^^;

[Mastamage]

I will be testing all builds of Arcane tomorrow and will have an extensive report of my findings. From what I've done so far with the haste build, it's not looking promising as the mana gets eaten up fairly quickly. Stay tuned until tomorrow for my results.

[Joshythemage]

Build: Frost | Invocation | Hit>Haste>Crit (to cap)>Mastery (Using T15 gear with Tyrannical PvP off pieces and blue premade trinkets since I didn't like any of the PvP ones)
Test: 1
Result: 67-70k
Strategy: "Cruise" Speed/No Frozen Orb, Alter Time, Icy Veins, or Time Warp

Build: Frost | Invocation | Hit>Haste>Crit (to cap)>Mastery
Test: 1
Result: 80-82k
Strategy: Patchwerk Style, all out/all CD's (including Time Warp) used but still only self buffed.

Rotation for both tests: Frost Bomb on CD > Frostbolt to maintain 3 stacks of Debuff > Frozen Orb on CD (2nd test only) > FFB on Brain Freeze Proc > Ice Lance on FoF Proc > Frostbolt. Both tests approx 5 minutes each.

[jtmzac]

@joshythemage That seems very high. Did you enchant your gear? Anyway I suggest people use skada instead of recount because recount has been providing wrong numbers when using invocation.

My testing:
Build: Frost | Invocation | Hit>Haste>Crit (to cap)>Mastery (Using T15 gear with Tyrannical PvP off pieces and blue premade trinkets since I didn't like any of the PvP ones) using full haste gemming without jc gems or any enchants.
Test: 1
Result: 58k (I just dps until I've done 30mil damage to try and remove some of the rng. I've noticed frost seems to keep getting lower over time but arcane is much more stable)
Strategy: "Cruise" Speed/No Frozen Orb, Alter Time, Icy Veins, or Time Warp

Build: Frost | Invocation | Hit>Haste>Crit (to cap)>Mastery, same gear as above
Test: 3
Result: 68k (again I just dps'd until 30 mil damage. I think frost really needs a longer period to test it properly)
Strategy: Patchwerk Style, all out/all CD's (excluding Time Warp and trinket) used but still only self buffed.

Standard frost rotation. The only notable thing is when using all cd's I've been delaying frozen orb until after alter time to try and minimise wasting FoF procs.

I'll redo the test's more exactly in a little bit.