[Mage] TC after 2.3 - Page 135

maxi · 03/07/08, 10:54 AM

Originally Posted by Roywyn

Improved Amplify/Dampen Magic in Sunwell - Yay/Nay?

Think that talent is worth it in Sunwell Plateau? Seems not to me, with so much magic damage flying around.
Would love some feedback/Opinions of others though who were on the PTR more often.

Amplify was always worthwhile on full physical fights
It also works reasonably well on fights with small amounts of HUGE magical attacks.

Dampen is only good on fights where there is a large amount of small spells (Illhoof imps, Malacrass spirit bolts etc).

I think dampen is useless in Sunwell, and Amplify might be good on Brutallius, but don't hold your breath :p

Zene · 03/07/08, 11:01 AM

Originally Posted by maxi

Dampen is only good on fights where there is a large amount of small spells (Illhoof imps, Malacrass spirit bolts etc).

I don't have a link but I'm pretty sure Dampen got nerfed for Hex Lord's spirit bolts.

maxi · 03/07/08, 11:05 AM

Originally Posted by Zene

I don't have a link but I'm pretty sure Dampen got nerfed for Hex Lord's spirit bolts.

Well, it doesn't affect spirit bolts to full extent anymore, but it still helps the raid survive a lot longer

Duodecimal · 03/07/08, 11:31 AM

Well, no, Amplify on ranged is what you want for the Hex Lord. Maybe dampen melee for consecrate, but I'd say it's a wash over leaving them unbuffed.

sambjo · 03/07/08, 1:26 PM

Amp is almost always worth putting on the tank(s).

Jarlyn · 03/07/08, 2:19 PM

I've not been able to test the Twins much, but having at least some experience at every other open fight in Sunwell, I think Imp Magics is a bit of a wash. For Kalecgos, a number of his abilities (Arcane Buffet) are physical, but the corrupter's shadow abilities are not. Brutallus it would be useful on for tanks, but Felmyst and M'uru both involve large quantities of magic damage and it isn't likely you'd use anything there.

[e] grammars

alvinrod · 03/07/08, 2:27 PM

Looks like there's a new caster neck that dropped in Sunwell.

24 Stamina
17 Intellect
15 Spell hit
32 Spell haste
39 Spell damage

About the same as the JC neck except it has spell hit instead of crit. Depending on your spell hit this might be a better alternative than the JC neck.

Sancus · 03/07/08, 3:31 PM

bout the same as the JC neck except it has spell hit instead of crit.

No socket, it's not as good.

Kavan · 03/07/08, 4:29 PM

Originally Posted by Roywyn

- FrB3
Is that cycle what it states? You 1 AB + 3 FrB, to get one 2.13s AB (pre-haste) at base mana cost?
My gut feeling from previous calculations with Vontre's is that is lower DPS and only very slightly better mana efficiency than pure FrB spam - in a setting with CoE and WC.

The ideal thing to use is AB3/FrB3, but the AB debuff delay bug kills it.

AB/FrB/AB/FrB/AB/FrB3 is a suggestion of mine. Separate the ABs with FrBs to make sure you get the cast speed debuff, and end with FrB3 to have the debuff stack run out during the following AB.
It's just a suggestion - put it into a sim/sheet and see how it works.

From what I remember, AB/FrB rotations only a tiny bit better than FrB spam and AB spam mixed.
Seeing that you will still run into the same issue of having a 0.8s cast gap for the AB debuff, it's probably not worth it.

What I'd strongly recommend is having a "FrB spam and AB spam only" pattern to fall back to if you run into debuff timing issues.
It avoids timing headache, and is very potent with CoE+WC+4T6, on par with rotations.

Actually it already takes all options into account. It only makes some small optimizations like it doesn't try using AB cycles in deep fire build for example or flame caps in arcane builds. But you can force it to take even those cases into account if you think it's relevant by disabling the Smart Optimization checkbox in options. So it already is considering the FrB spam as fallback. The fact that you're seeing AB cycle with the pause just says that even that AB cycle with relatively large pause is superior to using simple FrB spam.

All the cycles are at least a little dynamic and change depending on haste. You can check what they are at the current settings by mousing over them in the Spell Info. For example what I have open at the moment in the solution it uses ABFrB3Frb and if I look over in Spell Info it says that is AB-FrB-AB-FrB-AB-FrB-FrB-FrB-Pause. I have the ABx3+FrBxN variants in code, but they are inferior in all cases I checked so I'm not displaying them.

After testing things yesterday I found that letting AB debuff drop is always worse than the pause, but Scorch filler is surprisingly effective. Next version will have those new cycles in.

Kavan · 03/07/08, 4:57 PM

I'm adding Drums of Battle stacking to Rawr.Mage and I have one question. Does spell haste reduce global cooldown on drums in 2.4?

Roywyn · 03/07/08, 6:38 PM

Originally Posted by Kavan

Actually it already takes all options into account. It only makes some small optimizations like it doesn't try using AB cycles in deep fire build for example or flame caps in arcane builds. But you can force it to take even those cases into account if you think it's relevant by disabling the Smart Optimization checkbox in options. So it already is considering the FrB spam as fallback. The fact that you're seeing AB cycle with the pause just says that even that AB cycle with relatively large pause is superior to using simple FrB spam.

All the cycles are at least a little dynamic and change depending on haste. You can check what they are at the current settings by mousing over them in the Spell Info. For example what I have open at the moment in the solution it uses ABFrB3Frb and if I look over in Spell Info it says that is AB-FrB-AB-FrB-AB-FrB-FrB-FrB-Pause. I have the ABx3+FrBxN variants in code, but they are inferior in all cases I checked so I'm not displaying them.

Aha!

I checked the mouseover of the spells, but not of the spell sequences. Now it's all crystal clear.
ABFrB3FrB actually means (ABFrB)*3+FrB and not AB+FrB*3+FrB, I just wasn't used to that notation.

DL'ed the new Rawr, added the Twins' necklace and Sunflare and checked some numbers.
Non-JCers now get even more default +hit, giving even more favour to T6 hat.
For arcane specs, I used T5 hat/shoulders, T6 gloves.

Same setup as before - 360s fight, 350mp5 SP, 50ms lag, draenei elemental shaman.
The above gear with 2T5, +40 dmg weapon/+70 dmg flask and the usual buffs.
DPS values are with molten/mage armour.

DPS in a 6 minute fight with 2T5

2/48/11 - 2460/2439
10/0/51 - 2219/2174
40/18/3 - 2391/2395

48/0/13 - 2328/2351 (without WC: 2231/2262)
43/0/18 - 2450/2457 (without WC: 2344/2361)
40/0/21 - 2498/2502 (without WC: 2388/2401)

So, given equal gear, Arcane/Frost can beat full fire even on a long 6 minute fight if Winter's Chill is up. By ~2%. It lags ~2% behind without WC. And that does not include ramp-up time (I think).

Reducing the fight to 4 minutes, it will get even closer.

DPS in a 4 minute fight with 2T5

2/48/11 - 2582/2553
40/0/21 - 2653/2654 (without WC: 2548/2560)

Note that this is with slightly less than optimal gems, some Reckless Pyrestones could be swapped for Potent ones for Arcane. Also, Fire spec is 0.5% under the cap, so this could even out, and a few changed Pyrestones don't affect DPS a lot.

Now, this was with 2T5.
Drop T5 for Sunwell gear, and use Sunfire enchant + Flask of Pure Death and Fire will make a huge leap.
Even if we swap back to +40 dmg weapon/+70 dmg flask, Arc/Frost will drop with the loss of 2T5.

DPS in a 6 minute fight without T5

2/48/11 - 2606/2586, that's ~150 DPS from better gear mostly (enchant/flask are the minor part).
40/0/21 - 2408/2387 (without WC: 2281/2271)

Seems that gear progression finally killed AB rotations for good.

Rawr cycle issue?

Kavan, I've had one oddity in the calculations.

With the T5 setup, WC, Arc/Frost, the spells were as follows:
Frostbolt - 2039 DPS, -17mps (i.e. gain mana while casting)
ABFrB3FrB - 2065 DPS. 11mps
AB spam - 2784 DPS, 336 mps

So, going from FrB to rotations is +26 DPS at the cost of +28mps spent (0.93 DPS/mps = 0.93 DPM).
Going from FrB to AB spam is +745 DPS at the cost of +353mps spent (2.11 DPS/mp2 = 2.11 DPM).

That means it's better to just use FrB and AB spam, than trying to do rotations and AB spam.
Rawr however uses ABFrB3FrB (188s) and some AB spam time (43s).
I should get more damage from using just FrB and AB though, unless I read the info wrong?

Rawr correctly uses IV/Hex/AP with AB spam. It also uses Hex+AB spam.
It uses all other haste procs with ABFrB3FrB, and another 188s of just ABFrB3FrB, and 43s of AB spam.
Heroism is paired with IV/Hex/Skull and ABFrB3FrB, but not AP-AB, which seems not optimised.

If I understand the spell tooltips correctly, it is better to skip ABFrB3FrB fully.
Stacking cooldowns with AB spam as much as possible and using FrB on the rest should be more damage, unless I got something wrong.

◊ manly · 03/07/08, 6:44 PM

I wonder, is the rotation [AB, AM] supported in rawr? Yes, I mean intentionally using 1.5s AB all the time and enjoying a FSR mana tick every rotation. The numbers should be close to frostbolt numbers, but not rely on WC or COE. That rotation has many advantages over frostbolts, namely uninterruptibility and FSR mana tick, in addition to what I mentioned previously.

Roywyn · 03/07/08, 7:04 PM

Originally Posted by manly

I wonder, is the rotation [AB, AM] supported in rawr? Yes, I mean intentionally using 1.5s AB all the time and enjoying a FSR mana tick every rotation. The numbers should be close to frostbolt numbers, but not rely on WC or COE. That rotation has many advantages over frostbolts, namely uninterruptibility and FSR mana tick, in addition to what I mentioned previously.

Not sure which cycle you mean exactly, but Rawr does have an "AB-AM-Pause" cycle.

With 48/0/13, mage armour, no WC/CoE, no JoW, it's 1349 DPS while regaining 31 mps (you stand AFK for a few seconds every ~7 seconds).
AM spam is 1729 DPS while costing 37 mps.
"AB-AM-AB-AM-AB-AM-FrB*" is 1807 DPS while costing 36 DPS.

(There are some more options with another filler * than FrB, that are less damage and not worth the loss. Options being Sc-Sc and Sc-Pause. It also adds FrB-Pause and FrB-FrB if you have a lot of haste and need more spells to fill the pause.)

[Edit]:
Ah, that one. No, rawr, doesn't have it.

But you can stitch it together from AM spam (1729 DPS, 37 mps regen) and AB spam (2784 DPS, 337 mps spent).
Adds up to 1973 DPS, 106 mps spent. Cycle length is 6.5s/120% = 5.4 seconds (20% haste on gear).
You have 462 mp5 from spirit, or 185 mana/tick. One OO5sR tick adds 185*40% = 74 mana over 5.4s, or 13.7 mps.

So, you get 1973 DPS, 92 mps spent in the best case.

To get DPS data from gear, you could try adding the 13.7*5 = 69 mp5 to your SP regen, as a bandaid.

Also, mind you that at 5.4s cast time per cycle, your chance to actually get that oo5sr tick is really low.

[Edit]:

Originally Posted by Kavan

The reason why it doesn't stack IV/Hex/Skull with AP-AB is GCD clipping I think. What haste levels were you testing at?

Whoops! Ran 323 passive haste, 20.5%. Clips at +25% haste from other sources. Totally forgot about that.

◊ manly · 03/07/08, 7:09 PM

No no, I mean literally what you see. [AB, AM] (no pause).
This is how it would work on a boss:
AB, AB, AB, (yay ramp-up bug -- yes this intentionally lists 3 AB as ramp-up time even though it should be 2)
AB, AM AB, AM AB, AM AB, AM ...
You do not do any pause or anything. Just spam both, giving you permanently 1.5s AB and full AM. And you don't cancelcast to use clearcasting either.

Kavan · 03/07/08, 7:22 PM

The reason why it doesn't stack IV/Hex/Skull with AP-AB is GCD clipping I think. What haste levels were you testing at?

In my T5 setup with Arcane/Frost I'm at 277 haste and it uses:

Heroism-AP=>AB spam
Heroism-IV=>AB cycle
IV-AP=>AB spam
IV=>AB spam

That thing with FrB+AB spam vs AB cycle+AB spam mix looks suspicious. Are you sure you looked at the right AB cycle? Can you open a new issue on Rawr (Rawr - Issue Tracker) and attach xml of the character setup you used (File->Save in Rawr)?

Kavan · 03/07/08, 7:25 PM

Originally Posted by manly

I wonder, is the rotation [AB, AM] supported in rawr? Yes, I mean intentionally using 1.5s AB all the time and enjoying a FSR mana tick every rotation. The numbers should be close to frostbolt numbers, but not rely on WC or COE. That rotation has many advantages over frostbolts, namely uninterruptibility and FSR mana tick, in addition to what I mentioned previously.

Good idea. Initially I thought this would be equivalent to combination of AB spam and AM spam but mps characteristic is different. I'll add it in.

◊ manly · 03/07/08, 7:32 PM

Well, I am asking for it because my data so far matches almost exactly the data from Rawr. And yes, the TC was shifting away from rotations, which is good given the ramp-up issues of AB. However, I did come up with interesting numbers with [AB, AM]. Thing is, not everyone will agree on what is 'common place' buffs in a raid. By my own definition, WC is not commonplace. I always assume it is not available. 13% COS is a given, 10% COE might be there. In any case, assuming the case where you're comparing

13% COS, [AB, AM]
10% COE, deep fire

the numbers got quite interesting. (Since I don't assume both COS and COE present at the same time with arcane/frost mages, that would mean 3+ warlocks and 2+arcane/frost mage with 1 deep frost mage, a composition I don't believe is commonplace at all)

Kavan · 03/07/08, 7:37 PM

Ok I added it in and it turns out it's not a that good. Compared to the ABAMx3+FrB cycle 1803 dps/-19.2 mps you get 1945 dps/48 mps. So you're trading at 2.1 dpm. Trading for AB spam gives you 2.9 dpm. I can see it might be worth using it when you're just a bit short of full AB spam. Actually I do that in practice to fine-tune getting OOM at the end.

EDIT: Might be worth looking at AB-AM with AM when AM procs CC.

Kavan · 03/07/08, 8:03 PM

Now this is intersting. I added a new dynamic cycle for AB-AM, when AM procs AM chain AM until you don't proc AM, then ramp up AB and return to AB-AM.

This cycle outperforms the FrB variants in no COE situations.

I'll try another dynamic cycle that ramps up with AB-AM instead of just AB and again goes to AM when AM procs CC.

EDIT: I take that back, I made a mistake in this new cycle. It's not as good as I thought it would be. Even under no COE situations the (AB-AM)x3+FrB is better.

Here's some sample values with T5:

AB-AM: 1945.13 dps/47.99 mps
AB-AM with AM on CC, straignt AB ramp: 1910.62 dps/35.74 mps
AB-AM with AM on CC, AB-AM ramp: 1890.29 dps/22.34 mps
(AB-AM)x3+FrB with AM on CC: 1813.69 dps/-20.77 mps

trading with AB spam at: 2691.35 dps/283.31 mps

To estimate how valuable a cycle is in combination with AB spam let's say we have total time T, t spent on cycle, T-t on AB spam, M total mana. Then we have:

M = t * cyclemps + (T-t) * abmps
total damage = t * cycledps + (T-t) * abdps

Solving for t we get:

M = t * (cyclemps - abmps) + T * abmps
t = (M - T * abmps) / (cyclemps - abmps)
total damage = (M - T * abmps) * (cycledps - abdps) / (cyclemps - abmps) + T * abdps

Let's call (cycledps - abdps) / (cyclemps - abmps) tradeoff coefficient. We're assuming that M < T * abmps, otherwise we could AB spam all time. So the lower tradeoff coefficient is the higher total damage is.

tradeoff coefficients for the above cycles:

AB-AM: 3.17 dpm
AB-AM with AM on CC, straignt AB ramp: 3.15 dpm
AB-AM with AM on CC, AB-AM ramp: 3.07 dpm
(AB-AM)x3+FrB with AM on CC: 2.89 dpm

• Astrylian · 03/07/08, 10:09 PM

Rawr Beta 12.1 posted!

DOWNLOAD: https://www.codeplex.com/Release/Pro...eleaseId=11483

Beta 12.1:
- Fixed a bug that would cause "Input string not in correct format" errors when Rawr was run on copies of Windows for a region where a period is not the decimal separator (ie european countries where it would be "1,23" instead of "1.23").
- For proxies that filter by user-agent, the user-agent used by Rawr is now customizable in the proxy options.
- Config settings (like recent files) are now user specific.
- Fix for occasionally not loading icons immediately even though a network connection is available. Also fixed an error when there was no network connection.
- Rawr.Mage: Added new AB-AM cycle and new AB cycles with Frostbolt/Scorch filler, added a fix for the item budget comparison, added scrolling on the options screen, added Drums of Battle stacking,
- Rawr.Warlock: Fixed issue when altering spell cycle

Sinless · 03/08/08, 12:00 AM

Originally Posted by Kavan

Now this is intersting. I added a new dynamic cycle for AB-AM, when AM procs AM chain AM until you don't proc AM, then ramp up AB and return to AB-AM.

This cycle outperforms the FrB variants in no COE situations.

I'll try another dynamic cycle that ramps up with AB-AM instead of just AB and again goes to AM when AM procs CC.

EDIT: I take that back, I made a mistake in this new cycle. It's not as good as I thought it would be. Even under no COE situations the (AB-AM)x3+FrB is better.

Here's some sample values with T5:

AB-AM: 1945.13 dps/47.99 mps
AB-AM with AM on CC, straignt AB ramp: 1910.62 dps/35.74 mps
AB-AM with AM on CC, AB-AM ramp: 1890.29 dps/22.34 mps
(AB-AM)x3+FrB with AM on CC: 1813.69 dps/-20.77 mps

trading with AB spam at: 2691.35 dps/283.31 mps

To estimate how valuable a cycle is in combination with AB spam let's say we have total time T, t spent on cycle, T-t on AB spam, M total mana. Then we have:

M = t * cyclemps + (T-t) * abmps
total damage = t * cycledps + (T-t) * abdps

Solving for t we get:

M = t * (cyclemps - abmps) + T * abmps
t = (M - T * abmps) / (cyclemps - abmps)
total damage = (M - T * abmps) * (cycledps - abdps) / (cyclemps - abmps) + T * abdps

Let's call (cycledps - abdps) / (cyclemps - abmps) tradeoff coefficient. We're assuming that M < T * abmps, otherwise we could AB spam all time. So the lower tradeoff coefficient is the higher total damage is.

tradeoff coefficients for the above cycles:

AB-AM: 3.17 dpm
AB-AM with AM on CC, straignt AB ramp: 3.15 dpm
AB-AM with AM on CC, AB-AM ramp: 3.07 dpm
(AB-AM)x3+FrB with AM on CC: 2.89 dpm

Kavan, what is the trade-off coefficient for (AB-FrB)x3 -> FrBx3 cycle?

Kavan · 03/08/08, 12:21 AM

Originally Posted by Sinless

Kavan, what is the trade-off coefficient for (AB-FrB)x3 -> FrBx3 cycle?

I added the AB spam tradeoff coefficient to tooltip in Rawr.Mage for AB cycles (I got tired doing it by hand :P)

For slightly different gear for Arcane/Frost spec I get 2.92 dpm for ABFrBx3+FrB. Same gear with Arcane spec is:

ABFrBx3+FrB: 3.37 dpm
ABAMx3+FrB: 3.15 dpm

So it depends a lot on specific gear and spec.

Sinless · 03/08/08, 12:51 AM

So, I am not gonna go into individual calculations, but with the following stats (raid buffed):

* +1500 spell damage
* 27% crit with spells
* 200mp5 while casting
* Hit-capped for both arcane and frost
* No spell haste
* CoS (no CoE)
* No JoW
* No WC

These are very close to my stats raid buffed /w mage armor on, and we always have CoS up.

1069 DPS for AM (61/0/0) (including -6% for partial resists)
1037 DPS for FrB (40/0/21)

AM will have, 784 - 100 - 150 = 534 mp5
FrB will have (272 - 100) x 2 = 344 mp5

Hence, the question:

Is a DPS gain of about ~32 DPS worth an mp5 loss of about 190 mp5?

Edit: I am so torn between AM and FrB as rotation fillers, so just trying to make up my mind

Kavan · 03/08/08, 12:54 AM

Originally Posted by Roywyn

Rawr cycle issue?

Kavan, I've had one oddity in the calculations.

With the T5 setup, WC, Arc/Frost, the spells were as follows:
Frostbolt - 2039 DPS, -17mps (i.e. gain mana while casting)
ABFrB3FrB - 2065 DPS. 11mps
AB spam - 2784 DPS, 336 mps

So, going from FrB to rotations is +26 DPS at the cost of +28mps spent (0.93 DPS/mps = 0.93 DPM).
Going from FrB to AB spam is +745 DPS at the cost of +353mps spent (2.11 DPS/mp2 = 2.11 DPM).

That means it's better to just use FrB and AB spam, than trying to do rotations and AB spam.
Rawr however uses ABFrB3FrB (188s) and some AB spam time (43s).
I should get more damage from using just FrB and AB though, unless I read the info wrong?

Rawr correctly uses IV/Hex/AP with AB spam. It also uses Hex+AB spam.
It uses all other haste procs with ABFrB3FrB, and another 188s of just ABFrB3FrB, and 43s of AB spam.
Heroism is paired with IV/Hex/Skull and ABFrB3FrB, but not AP-AB, which seems not optimised.

If I understand the spell tooltips correctly, it is better to skip ABFrB3FrB fully.
Stacking cooldowns with AB spam as much as possible and using FrB on the rest should be more damage, unless I got something wrong.

I just realized what is going on with FrB. I had it disabled with smart optimizations if you didn't have Emp. Frostbolt. I'll turn it on in all arcane builds.

But it's not clear that FrB is better in all cases. For example in my T5 setup I get 2.52 dpm tradeoff for FrB and 2.46 dpm for ABFrB3FrB.

Icicles · 03/08/08, 2:31 AM

I was trying out a setup in rawr pretty close to chardev.org - A World of Warcraft character planner v.2.beta, basically spamming +10 spirit gems everywhere.

With reasonable buffs and a good shadow priest, it looked like Arcane Blast could last almost the entire fight (toss an Innervate in there and gg), but the dps was a good bit (200ish) lower than what I was getting with an optimal fire setup (best gear options, 4pc, correct gems, etc.).

I was getting a +dmg number about 100 short of fully buffed fire, (thanks MM), a crit rate of roughly 30% for AB, a little too much hit (mainly due to the sunwell staff), and 12ish% haste.

In any event, bringing this up as a possible point of discussion, despite the loss of virtually all socket bonuses (which is pretty bad, granted), could this ever potentially be a viable setup, in order to allow yourself to spam AB for an entire fight, regardless of how long the fight was? First time using the software, so I might have missed a thing or two.

Side note: with the level of Int in that setup (buffed, about 875), each point of spirit is worth roughly 1.3 mp5 noncasting, and at just under 700 buffed spirit, that's about 600 mp5 while casting, before factoring in a shadow priest.

Summary + Question: Would the use of all Spirit gems/enchants/gear and (ideally) the ability to spam AB an entire fight's duration make up for the loss of damage from those slots/bonuses if optimized better than how I did it or under other circumstances?