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.
Doing that rotation, would it be possible to limit your haste to still catch full regen ticks? Considering the amount of haste on almost all the gear in Sunwell, it's nearly impossible to avoid a rather decent amount, but at what point with haste does AM -> 1.5 AB finish in under 5 seconds, thus nullifying the advantage of that rotation?
Are you sure you got the haste calculation for arcane correct in Rawr. It seems you have used the ratio: 1 haste = 1,1 spell dmg (Fire ratio). The value of haste seems too high. Im quite sure the value of haste will be lower for arcane. I would guess: 1 haste = 0.65-0.85 spell dmg.
Awesome software btw. Well done
Last edited by Andersnordic : 03/08/08 at 5:58 AM.
Are you sure you got the haste calculation for arcane correct in Rawr. It seems you have used the ratio: 1 haste = 1,1 spell dmg (Fire ratio). The value of haste seems too high. Im quite sure the value of haste will be lower for arcane. I would guess: 1 haste = 0.65-0.85 spell dmg.
Awesome software btw. Well done
I'm not sure I understand. Haste gives the same casting increase to all spells. You can mouse over the spells in Spell Info and it should tell you their casting time (without latency). Let me know if you see some spell that is getting the wrong effect, but I kindof doubt it since they all share the same code path.
As time decreases from 300s to 280s Arcane DPS is at 2696 (CoS 13%, CoE 10%).
As time increases from 300s to 320s Arcane DPS is at 2615.
At 360s, 2552 DPS.
At 420s, 2572 DPS.
At 480s, 2505 DPS.
Fire:
As time decreases from 300s to 280s Fire DPS is at 2670 (CoE 10%).
As time increases from 300s to 320s Fire DPS is at 2605.
At 360s, 2568 DPS.
At 420s, 2564 DPS.
At 480s, 2494 DPS.
It looks like just under 5min is the sweet spot for fire but arcane appears to be a viable spec come 2.4. I personally think I'll be going arcane for the extra ice block, trash DPS, and AoE perks.
Thanks again for Rawr, it's simply amazing.
Edit: Both specs were using optimal gear/enchants/consumables for each but with the same raid buffs.
Doing that rotation, would it be possible to limit your haste to still catch full regen ticks? Considering the amount of haste on almost all the gear in Sunwell, it's nearly impossible to avoid a rather decent amount, but at what point with haste does AM -> 1.5 AB finish in under 5 seconds, thus nullifying the advantage of that rotation?
OO5SR isn't easy to work with: exactly 5sec after spending mana you're classified as "not casting" and out of the 5SR, but it'll only show up in the next tic. That means: in order to 100% guarantee you'll get an OO5SR while rotating AM/x the x needs to be 2.0sec. Follow the time-line:
0.00 AM
5.00 Toggled as OO5SR
5.00 AM end
5.01 AB start channeling
The only thing you know for sure is that by 7.01 you'll have had at least one OO5SR tic. However, it could be that the tic was at 5.02, or it could be the tic was at 6.99. When energy gain tics are processed is not related to casting but rather to an internal "clock" in the game.
Technically you're not looking for "how much haste will make 1.5sec AB + AM = 5sec" because that will make the chance you OO5SR next to 0.00%. With 0 haste as is there's a chance you don't OO5SR.
I'm not sure on the statistics, but given that a 2.0 time interval OO5SR will guaranteed give you 1 and not more tics, then the highest two debuff ABs have only a 75% and a 91.5% to give you an OO5SR, while the default has 125% (ie, 100% to give you one tic and 25% to give you two) and the first debuff one 108% likewise.
Perhaps we should integrate the exact timing on Rawr, however I'm slightly weary of making the simulation behave erratically, along the lines of +1haste = 4% shift in mana.
Edit: Interestingly, given you cycle once of each debuff AB in your AM/AB, it seems the average chance you tic a OO5SR is technically [1.25+1.08+0.92+0.75]/4. Which is 100%. Not surprising, given the average cast time of AB is 2.0. So given a cycle of the ABAM3x type, it's rather safe to assume that with 0 haste you'll proc an average of exactly 1 OO5SR each AB/AM.
Perhaps we should integrate the exact timing on Rawr, however I'm slightly weary of making the simulation behave erratically, along the lines of +1haste = 4% shift in mana.
Actually Rawr does this, including the effect of clearcasting extending the OO5SR time (on cast time spells only, channeling always forces you in 5SR).
Actually Rawr does this, including the effect of clearcasting extending the OO5SR time (on cast time spells only, channeling always forces you in 5SR).
Channeling always forces FSR? even when clearcasted? this is news to me.
Apparently, Blink has been changed to 21% base mana, according to the latest PTR notes. This would effectively make it circa 470mana.
I'm getting rather annoyed with this ping-pong. It went down from 630 to a shocking 230, then was amended to 345, now it's back at 470. Now taking bets on where you think it'll be by the next version of PTR.
On a fruity side-note, Essence of Immortals has been removed from the game in a bid to give all realms equal access to bosses. The stages of the instance will instead open over time. Good news for those of us in desire of progress but who suffer from under-performing realms.
Kavan: Could you please elaborate on the channeling->5SR rule you pointed out?
Andorian, would it be possible to put up a link on chardev or something similar with the setup you made that got you those numbers? I'm curious to see what you put together, cause I'm like you, personally I favor arcane and would love to see it go toe-to-toe with fire in 2.4.
Andorian, would it be possible to put up a link on chardev or something similar with the setup you made that got you those numbers? I'm curious to see what you put together, cause I'm like you, personally I favor arcane and would love to see it go toe-to-toe with fire in 2.4.
Kavan: Could you please elaborate on the channeling->5SR rule you pointed out?
Mana Regen:
if channeling or time since last mana spent < 5 => 5SR regen
else => OO5SR regen
It's easy to test if you have a mod that shows you mana regen ticks. Go to Dr. Boom and chain cast AM. Clearcasted AM will give you small/no ticks. Also ask any priest, when they mind control for example, even though the channel is longer than 5 sec they never drop out of 5SR until they get out of the channel.
(1) While channeling, one cannot have non-casting regen active, even if it was active before the channeled spell and the channeled spell cost no mana; non-casting regen immediately resumes when channeling completes
(2) While channeling, one cannot go from while-casting regen to non-casting regen, but the 5-second timer continues, so that one might immediately go from while-casting regen to non-casting regen when channeling completes
(3) While channeling, one cannot go from while-casting regen to non-casting regen, and the 5-second timer is on hold, resuming when channeling completes and allowing non-casting regen to begin some time after
And what evidence do we have that distinguishes these?
(1) While channeling, one cannot have non-casting regen active, even if it was active before the channeled spell and the channeled spell cost no mana; non-casting regen immediately resumes when channeling completes
(2) While channeling, one cannot go from while-casting regen to non-casting regen, but the 5-second timer continues, so that one might immediately go from while-casting regen to non-casting regen when channeling completes
(3) While channeling, one cannot go from while-casting regen to non-casting regen, and the 5-second timer is on hold, resuming when channeling completes and allowing non-casting regen to begin some time after
And what evidence do we have that distinguishes these?
It's (1). Pretty easy to verify... just chain cast AM until you get a clearcast proc. Wait until you regen a tick of mana from being OO5SR then consume your clearcast with another AM. Watch your regen ticks... they will not be full OO5SR ticks.
2657 DPS (CoS 13%, CoE 10%)
is extremely unlikely too. You need 3 warlocks and 3+ 40/0/21 mages before its worth it, or 3 warlocks + 1+ fire/frost mage + 1+ 40/0/21 mage. And even 1/1 mage might not cut it depending on the performance of your mages.
2762 DPS (CoS 13%, CoE 13%, WC, Never Going to Happen)
is probably never going to happen
Likewise, 13% COE is like so remote I wouldn't even dare hope it to happen.
I think here are the more realistic scenarios:
13% COS
13% COS + JOW
10% COE
And add possibly moonkin and ret pally (+3% crit) to the mix.
<Eej> YOU"RE GONNA PULL
<Eej> IF YOU SQUEEZE OFF ANOTHER ARCANE BLAST
<Spectear> You've obviously never played with Manly.
<Spectear> That's hardly a reason to stop DPS. Very Manly Staff
Kavan would it be possible to add some Fireblast rotations for less progressed mages.
Maybe:
(FBx2, Fireblast)x2 FBx2, Scorch
with enough spell haste (2.4 style) it goes to:
(FBx2, Fireblast)x3 FBx2, Scorch
If these are too complicated just
FBx2, Fireblast
I haven't heard a concrete number, only a general observation that spamming one spell is optimum, while alternating spells does not work perfectly with the current spell casting system. Does anyone know how much this hurts a rotation like this?
2657 DPS (CoS 13%, CoE 10%)
is extremely unlikely too. You need 3 warlocks and 3+ 40/0/21 mages before its worth it, or 3 warlocks + 1+ fire/frost mage + 1+ 40/0/21 mage. And even 1/1 mage might not cut it depending on the performance of your mages.
Likewise, 13% COE is like so remote I wouldn't even dare hope it to happen.
I think here are the more realistic scenarios:
13% COS
13% COS + JOW
10% COE
I agree, those are the most likely situations. I just thought we might be seeing more CoS 13% + CoE 10% since more locks will be specing fire/destro. I'll recalculate the arc/frost for CoS 13% only. I just added the other totals out of curiosity.
It's (1). Pretty easy to verify... just chain cast AM until you get a clearcast proc. Wait until you regen a tick of mana from being OO5SR then consume your clearcast with another AM. Watch your regen ticks... they will not be full OO5SR ticks.
This is mostly correct. Unless clearcasting mechanics have changed within the last few patches, there's one exception, however: If you clearcast an AM, and that AM itself procs clearcasting, the AM won't restart the 5SR timer, and it's possible to receive full mana regen ticks during the AM. This requires some luck, since you need back-to-back clearcasting procs with AM, but it's known to happen.
A quick question on rawr; with the elemental shaman option checked and 13% hit (along with EP) I am being told gemming other items with hit/dmg gems, as opposed to crit/dmg or pure dmg gems is superior dps. Any reason you can think of why this might be happening?
I think, I finally decided to go with 40/0/21 for multiple reasons but most importantly for Frostbolt being a very mana efficient cast with moderately high dps. Now, time to decide on the trinkets.
Two of the most potent trinket candidates for a deep arcane or 40/0/21 build are the lightning capacitor (TLC) and Darkmoon Card: Crusade (DCC). The below comparison is strictly for 40/0/21 arcane/frost build but trivial conclusions can be drawn at the end of this analysis for other common builds such as deep fire and frost.
A 40/0/21 will only cast 2 spells throughout the fight: Arcane blast (AB) and Frostbolt (FB). In order to understand which trinket is better for this talent build, we need to analyze the individual effects on both AB and FB.
I am doing the analysis for my particular raid setup (an elemental shaman in my group) and will also assume the following:
* 34 % raid buffed crit (including 3% from Arcane Potency talent)
* 11.32% spell hit (94.32% to hit with any spell, 99% to hit with frost and arcane spells with Elemental Precision and Arcane Subtlety)
* 6% partial resist on boss for all spells except frost spells
* Chaotic Skyfire Diamond is equipped in head meta-gem slot (see: Lhiv’s post WoW Forums -> CSD: I've been wrong! for the latest info on CSD’s crit modifier)
* Misery, Curse of Shadows, Curse of Elements are always up on the boss (Misery affects TLC)
* Damage and crit modifiers from talents:
- Arcane Instability: + 3% spell damage (affects TLC)
- Arcane Impact: +6% critical strike chance for AB
- Spell Power: +50% spell critical damage (does not affect TLC)
- Ice Shards: +100% critical strike bonus for frost spells (does not affect TLC)
- Piercing Ice: +6% frost damage (does not affect TLC)
Known facts about TLC suggest that, it is affected by the debuffs on the boss that affect nature spells, it is affected by your nature spell crit chance and nature spell hit chance (for a mage, this would be base spell crit and hit chances).
Now, let’s get to the calculations. We start with TLC.
Number of AB’s required to fire 1 lightning bolt: (remember 40% crit rate for AB)
100 / 40 * 3 = 7.5
Average damage contribution to each AB when TLC is used:
841.4 / 7.5 = 112.2
Number of FB’s required to fire 1 lightning bolt: (remember 34% crit rate for FB)
100 / 34 * 3 = 8.8
Average damage contribution to each FB when TLC is used:
841.4 / 8.8 = 95.6
Now, let’s take a look at DCC. We will assume that +80 spell damage is stacked and never needs to be restacked (not very realistic, but for the sake of simplicity)
For AB:
AB has 71.43% spell damage coefficient. So the damage contribution to each blast for 80 spell damage is:
in terms of damage increase per bolt due to using respective trinket. So, what do these results show?
If you use equal numbers of FB and AB in a fight, the average contribution from the trinkets will be:
TLC = 103.9
DCC = 102.6
However, a 40/0/21 build (especially with 2 piece T5) will always use many more AB’s than FB’s in any given fight, which makes TLC much superior to DCC.
What about deep fire, frost or even 40/0/21 just spamming frostbolts? DCC is significantly better than TLC. Remember, even for a 40/0/21 spamming only frostbolts TLC = 95.6 and DCC = 119. Now consider, all the + damage modifiers coming from deep fire and deep frost talents, DCC is MUCH superior to TLC in any way.
CONCLUSIONS:
Deep arcane – No analysis needed, hands down TLC.
40/0/21 – TLC is much better.
40+ fire builds – DCC is much better.
40+ frost builds – DCC is much better
Kavan would it be possible to add some Fireblast rotations for less progressed mages.
Maybe:
(FBx2, Fireblast)x2 FBx2, Scorch
with enough spell haste (2.4 style) it goes to:
(FBx2, Fireblast)x3 FBx2, Scorch
If these are too complicated just
FBx2, Fireblast
I haven't heard a concrete number, only a general observation that spamming one spell is optimum, while alternating spells does not work perfectly with the current spell casting system. Does anyone know how much this hurts a rotation like this?
I added the cycle in. It'll display in Spell Info, but it won't be considered in final solution unless you disable smart optimizations option.
Originally Posted by Duravi
A quick question on rawr; with the elemental shaman option checked and 13% hit (along with EP) I am being told gemming other items with hit/dmg gems, as opposed to crit/dmg or pure dmg gems is superior dps. Any reason you can think of why this might be happening?
What exactly do you mean by you are being told to pick hit/dmg gems? If you mouse over Frostbolt for example what is the calculated Hit Rate? If it still doesn't make sense you'll probably have to tell me exactly what gear and settings you're using.
Originally Posted by Andorian
Rawr may not be calculating TLC correctly then. I'm still showing TLC well below DCC in a 40/0/21 spec.
Synless is computing the contribution of TLC in a wrong way. You have to consider the 2.5 sec internal cooldown on proc aquiring after a discharge. I don't see what was the spell haste used in that example, but let's assume it's 0. Even if you have enough haste to get AB to 1 sec the value is still lower than DMC.
Kavan is right, I forgot to include the effect of 2.5 second cooldown in my calculations BUT my results do not change a lot even when including the internal cooldown (which only triggers not after every charge but after every bolt).
First of all, I am not calculating the dps values, rather the +damage contribution to each bolt/blast from each trinket.
So, for AB, we said it takes us 7.5 ABs to fire one lightning bolt. The only way trinket's internal cooldown can affect our calculations is if the very first AB cast after a lightning bolt is a crit. In that case we need 8.5 ABs instead of 7.5. The probability of this happening is 0.4 (AB crit chance).
Thus instead of 7.5 ABs, we need 7.5 * 0.6 + 8.5 * 0.4 = 7.9 ABs => 841.4 / 7.9 = 106.5
Similarly for FB, instead of 8.8 FBs, we need 8.8 * 0.66 + 9.8 * 0.34 = 9.14 FBs => 841.4 / 9.14 = 89.51
Compare them to, 112.2 and 95.6 and yeah there definitely is a decrease but it still does not make it worse than DMC.
Kavan as long as I know, the TLC CD is only for the Discharge, not the gathering of charges. Hence the only possible scenaria it becomes a limiting factor is (a) enough haste to make 3 AM within 2.5sec, which all three crit or (b) AoE.
Unless I'm wrong and the CD in fact does impede the gathering of charges.
There must be some error in his work however, I'm seeing TLC as optimal sine-qua-non only for 50.0.11, while for 40.0.21 it's only superior for higher AB cycles. The more FrB you mix in the more relative it's value is compared to DMC. Even so, the diff between DMC and TLC is small, in the tens.
Also gentlemen, bear in mind: When we're talking about a 5m sim which includes 6 iterations of BL, IV, AP, Trink, Destro, AB, Cycles, it starts to get distinctly likely that pitfalls are more of an issue than the 7dps difference in sim. Pairing trinkets will always be worse than self-activated affairs are simulated to be, unless you have some 100% sure-fire way to ensure they are scripted in to the optimal scenario (read: macrod, and you know exactly how and when to use them). DMC also suffers from needing stacking and the ubiquitous falling off. Granted, not very often, but if you're comparing TLC and DMC and seeing a 12dps difference, then the assumed "all CDs at start" scenario becomes a compromize in it's self as you either delay by 20sec or blow everything while 80dmg down.
You're right, I expressed it in terms of dps contribution. The cooldown is between discharge and first charge, not between charges.
If you have 0 haste then yes only the first AB after discharge can be affected. I've used a continuous model in Rawr and that is an underestimate since I assume first 1 sec of second AB also can't proc it, but that obviously doesn't matter. I'll change to discrete model and see how things change.
What Pint said is also true, the more frostbolts you introduce to the mix, the more DMC will pull forward. I was just talking about a typical scenario where the number of ABs cast is more than the number of FBs cast (which I understand may not always be the case). Assuming the DMC buff might fall off a couple times during a fight, you are looking at a much better situation for TLC, too.
03/08/08, 2:35 AM
Similar Threads
