TBC Mage Theorycrafting - Page 106

galzohar · 07/13/07, 2:23 PM

Remember that ignite crits do 210% damage and not 200%, so crit is slightly better than that, although not significantly. But it's a lot of little factors that add up so you shouldn't really neglect any of them ;p

• Hamlet · 07/13/07, 2:28 PM

Even if you don't like the rest of the thread, a simple computation of relative stat values for deep Fire is posted in the OP here:
Damage, Crit, Hit, and Haste: Differential evaluation for item comparisons

Zephriel · 07/15/07, 11:22 PM

I'm assuming this is the right thread for this post.

Lately I've been working on the development of a new theorycrafting model for deep arcane DPS, one able to take into account the manner in which deep arcane mages use clearcasting procs (canceling spells to channel AM). A common argument I've heard from deep arcane mages is that most theorycrafting models oversimplify Arcane Concentration and Arcane Potency by handling them as -10% mana cost and +3% crit in calculations, respectively. So I theorized that if I could break common casting cycles (such as ABx3->AM->Scorch) into segments based on the probability of a clearcasting proc at each point in the cycle, I could determine:
- the duration of each segment (taking latency and reflexes into account)
- time spent in and out of the 5SR in each segment
- mana spent and damage done within each segment
Then that info could be compiled to find extremely accurate DPS/DPM outcomes for deep arcane rotations. Well, that's my hope, anyway.

Is this probability chart for ABx3->AM->Scorch and ABx3->AM cycles broken down correctly? (Please let me know if anything is unclear so I can explain it. I can also explain why I broke the chart down in this manner, if necessary.)

NOTES:
"2x AM cycle probability" is the percentage of the arcane mage's casting cycles where two AMs are
channeled consecutively to "double-dip" on clearcasting procs.
"3x to infinite AM cycle probability" reflects unlikely (but possible) casting cycles where more
than two AMs are chain-cast due to consecutive clearcasting procs.
"(C)" listed with a spell represents a clearcasting proc on that spell.
"(all)" means the listed cycle occurs whether a clearcasting proc occurs on that cast or not. 
In the below case, the third AB is listed with "(all)" because the arcane mage will never cancel
a spell to cast a clearcasted AM after the third AB; because the cycle calls for AM after ABx3,
the arcane mage would cast AM anyway.

ABx3->AM->Scorch cycles				Cycle probability	2x AM cycle probability
									(first AM procs clearcasting)
						(n)			(.1n)

AB(C) -> AM					.1(.9) = 9%		0.9%

AB -> AB(C) -> AM				.1(.9^2) = 8.1%		0.81%

AB -> AB -> AB(all) -> AM(C) -> AM					.1(1)(.9^3) = 7.29%

AB -> AB -> AB(all) -> AM -> Sc(C) -> AM	.1(1)(.9^4) = 6.561%	0.6561%

AB -> AB -> AB(all) -> AM -> Sc			(1)(.9^4) = 65.61%

Total						 89.271%		9.6561%

3x to infinite AM cycle probability		 100%-89.271%-9.6561% = 1.0729%


ABx3->AM cycles					Probability		2x AM cycle probability
									(first AM procs clearcasting)
						(n)			(.1n)

AB(C) -> AM					.1(.9) = 9%		0.9%

AB -> AB(C) -> AM				.1(.9^2) = 8.1%		0.81%

AB -> AB -> AB(all) -> AM			(1)(.9^3) = 72.9%	7.29%

Total						90%			9%

3x to infinite AM cycle probability		100%-90%-9% = 1%

I'm pretty confident about my probability figures in the second set (the ABx3->AM set), but I'm not so sure about the first.

galzohar · 07/16/07, 2:05 AM

I wouldn't jump to assume a cast is worth breaking when you clearcasts in order to clearcast AM. Before calculating any probablility I would check when clearcasted AM is actually worth the time lost.

I'm going to use my spreadsheet's numbers for the DPS of various spells just to get an idea if it's anywhere near worth casting, clearly worth casting, or close enough to break-even to be undecided without getting more accurate/personal DPS numbers.

3XAB+AM+scorch rotation without adjusting for clearcasts is 1598 DPS (0.13ms latency. I'll assume ping as 300 but with /stopcasting you I estimate I get a 0.13ms latency with normal spellcasting). Remember while latency can be reduced with /stopcasting when you don't rely on clearcast procs, it cannot be reduced and actually gets doubled effect when you rely on clearcast procs (1Xlag to see the clearcast + 1Xlag to cancel+cast, not to mention your reaction time).

Let's start with the best-case of AM proccing clearcast:
Time lost is 6 seconds (2.5s instead of 1.5s on AB + 5 seconds for AM) and DPS dealt is 7937 with 30% added crit, or 1322 DPS over 6 seconds. Since this is ALREADY a DPS downgrade from the DPS of totally ignoring clearcasts, you're better off keeping casting your rotation. Any other reactive AM casts will only end up with more time lost on AM and in some cases AB DPS lost on top due to restarting the debuff. No case is better than just AMing again after AM procced clearcasts as far as I can see, which is already not worth it. And I didn't even take into account the 0.13 latency you should get after AM.

Unless my numbers are way off, since breaking your rotation in the best case scenario for clearcasts is already not worth it, it won't be worth it in any other case either

Of course there's the "mana efficiency" scenario, but if you run into that scenario why the hell are you AM specced? Fireballs provide much better DPM unless judgement of wisdom is up 100% in which case AM gets very close to fireball's DPM, but that's never really going to happen.

EDIT: I had an error which doesn't really change the results which is why I won't correct it, but point it out: I accidently took the DPS numbers with 0 latency and not 0.13 as I said.

Zephriel · 07/16/07, 3:57 AM

Thanks for the post. Good points, especially the one about latency and its impact on response time. It would certainly be interesting if deep arcane mages are "barking up the wrong tree", so to speak, in regards to nuke rotations. (Or maybe I just listen to the wrong bunch!) Your post makes me wonder if my pursuit is even worthwhile. =P

If the best deep arcane rotations actually involve no canceled spells, the standard theorycrafting for Arcane Concentration and Arcane Potency seems to mostly hold true (-10% mana cost and +3% crit across the board), with the exception that a clearcasting proc on AM will increase the crit rate of both that AM and the following spell by 30%. So in theory, Arcane Potency must be worth between 3% and 6% crit to AM specifically (>3% because Potency can be applied to AM from either its own proc or the preceding spell's, and <6% because of the potential for consecutive AM clearcasting procs), no?

• Hamlet · 07/16/07, 11:16 AM

Well, not to mention, not only the response time and latency when trying to react and change spells, but the global as well.

manly · 07/16/07, 12:09 PM

Originally Posted by Zephriel

If the best deep arcane rotations actually involve no canceled spells, the standard theorycrafting for Arcane Concentration and Arcane Potency seems to mostly hold true (-10% mana cost and +3% crit across the board),

I don't know why so many people make the same mistake. Your chances to proc a clearcast depend directly on your hit. As such, it can never be 10%, only worse. But I do admit it makes a very minor difference.

Zephriel · 07/16/07, 1:03 PM

Originally Posted by manly

I don't know why so many people make the same mistake. Your chances to proc a clearcast depend directly on your hit. As such, it can never be 10%, only worse. But I do admit it makes a very minor difference.

Ah. I seem to remember hearing that before, but I never applied it to any calculations because of others' assertions that the talent tooltip was "misleading" and that resisted spells could still proc clearcasting. In hindsight, I should have tested that assertion before doing anything else. =x

That does bring me to one question, though: Does AM cheat the "resists can't proc clearcasting" rule? Unlike all our non-channeled spells, clearcasting occurs at (roughly) the beginning of the AM cast, rather than at the end, and it certainly does not have one chance per missile volley to proc (that was established earlier in this thread).

(Thanks for being patient with an inexperienced theorycrafter. ^_^)

ron9 · 07/16/07, 1:29 PM

With lootzor.com - World of Warcraft search and rate items - profile your wow character what do you all think would be the best settings for it to accurately show the best gear to have to maximize your DPS with 10-48-03 spec for raiding?

I was thinking
Intelect- 1
Spell damage/fire spell damage- 2
Spell crit- 3
Spell hit- 4

galzohar · 07/16/07, 1:41 PM

Scroll back and use a spreadsheet. No way crit is wroth 1.5X spell damage.

Ibber · 07/16/07, 2:48 PM

In regards to lootzor.com - this is a spreadsheet I use which is basically an excel version of the website. You can use it to play around with the weights pretty quickly to see how the gear will stack up. There are a couple mods which do this in game as well.

http://home.cfl.rr.com/jseigle/Ibber-MEP.xls

So basically it's a dump of all viable Mage gear from the online DBs with a customizable weighting system and information about where to obtain each piece without having to click a lot of links in wowhead or the like. I'm still trying to find the correct weights to assign for the different play styles myself but have a few pre-defined sets listed as starters on the MEP tab.

Anyways, might come in handy for the gear planning theorycrafters like myself...

ron9 · 07/16/07, 2:59 PM

That spreadsheet is valueing dmg and crit to be better then +spell hit. Isn't +hit the best stat until 16%chance to hit?

Example with gems: 8Spell hit gem > 9spell damage gem > 8spell crit gem until 16%chance to hit?

Ibber · 07/16/07, 3:37 PM

Against level 73 mobs, yes. The spreadsheet is meant to be useful for a variety of playing purposes : Solo grinding (level 68-70 mobs), PVP, Instancing, Raiding, etc. Each of these styles will have a different weighting system and the spreadsheet is useful in that it allows storage of multiple weight sets and quick resorting after you change them.

The weight set that is pre loaded into it is assuming solo grinding since that was what I mostly do and it was set to my style.

It might be nice to have another column such as the AEP version of DAEP or something along those lines. But this could quickly get out of hand and I'd have a column for each play style.

Copernicus · 07/16/07, 4:32 PM

Originally Posted by ron9

With lootzor.com - World of Warcraft search and rate items - profile your wow character what do you all think would be the best settings for it to accurately show the best gear to have to maximize your DPS with 10-48-03 spec for raiding?

I was thinking
Intelect- 1
Spell damage/fire spell damage- 2
Spell crit- 3
Spell hit- 4

1 Damage = 1 Haste = 1 Hit = 1.5 Crit = 5 Int = 9 Spirit = 2 Stamina

For lootzor, the weights would be

Damage: 10
Haste Rating (not included): 10
Hit Rating: 10
Crit Rating: 6.7
Intellect: 2
Spirit: 1.1
Stamina: 5

ron9 · 07/16/07, 7:27 PM

Been hearing a lot of different things on how good spell hit is compared to other stats for 10-48-03 Raiding. But with those numbers put into Lootzor, it would make a 9spell damage gem better then a 8spell hit gem, is that actaully better for raiding?

Roywyn · 07/16/07, 7:48 PM

Originally Posted by ron9

Been hearing a lot of different things on how good spell hit is compared to other stats for 10-48-03 Raiding. But with those numbers put into Lootzor, it would make a 9spell damage gem better then a 8spell hit gem, is that actaully better for raiding?

Hm, from my sheets with a 2-roll system, 1 spell hit comes out at 1.5 dmg for 10/48/3, and at 1.9 dmg for arcane/fire specs.
But for arcane/fire specs, you'll cap arcane hit before you cap fire hit, so some of your hit will only benefit half your spells.

Also, remember that many fights involve non-73 adds, where spell hit matters. In quite a few key fights, it's the DPS on non-boss mobs that matters most, and spell hit will be wasted there.

Hence, 1 dmg = 1 hit is a good estimate for raiding overall. For particular situations, it really depends how important boss DPS is.
On Lurker, I run with high +hit, because we never lack DPS on his adds. For learning, I used lower +hit for more add DPS.
We're currently learning Kael'thas, he and his advisors are level 73, his weapons and phoenixes are not (I think), so I run with some medium +hit, taking only those pieces that would be good even without +hit on them.

Personally, I don't use any +hit gems, I just swap +hit gear in and out whenever it's needed, and don't want to have +hit on gems wasted whenever it's not needed.
Also, you cannot really avoid +hit on good gear, so with later gear you'll cap without working hard for it, so I don't really want to gem for it right now.
So I use +9 dmg and +5dmg/4crit gems and grab the +hit elsewhere.

I'm not sure whether that's a good choice or not though.

Voley · 07/17/07, 3:14 AM

Being 10\48\3 I got few new concerns. On our kael kill, at phase 4 and on, I could not spend my mana, basically staying at 90% spamming fireballs and fireblasts. I think I may try the arcane blast build, as that mana could be put into more dps instead of just staying there as just ballast.

Northerner · 07/17/07, 6:26 AM

Be extremely aware though that the conversion rate of excess mana to improved damage is abysmal. It's there and it does count but it is (based on AB spam) not terribly significant. The only real benefit is that you can with AP turn it into extremely good comparative burst (pre-20%) if the situation calls for that.

Stein · 07/17/07, 6:29 PM

Since haste now affects channeled spells, slow seems to as well.

During slow on Aran, i used to blink to the edge and cast arcane missiles (since it wasn't affected). Last week, i started thinking...are there any channeled spells that would benefit from a slow cast?

I Evocated during the slow, and it seemed to tic at the same rate as normal...but lasted MUCH longer!

Now i can't think of many situations where this would be useful (maybe pvp with vs. 41 arcane mage), but thought it was interesting none the less.

Voley · 07/19/07, 2:06 AM

Yeh, I once tried evocation with bloodlust on, oh dear, 2k mana restored >_<

Athemeus · 07/19/07, 9:09 AM

Wouldn't this mean, that spellhaste reduces the amount of Mana restored by Evocation?

MacrosTheNaked · 07/19/07, 10:06 AM

Funnily enough, this may also mean that if you somehow manage to get slow/ curse of tongues on you, you could gain 10K mana from evo!

Aldric · 07/19/07, 10:11 AM

Originally Posted by Athemeus

Wouldn't this mean, that spellhaste reduces the amount of Mana restored by Evocation?

Yes, but you would need a whole lot of it. As far as I can remember you don't receive mana from the tick on the 8th second of evocation so you would have to push it back to a 6 second channel to lose any mana.

Astrik · 07/19/07, 11:34 AM

Originally Posted by Aldric

Yes, but you would need a whole lot of it. As far as I can remember you don't receive mana from the tick on the 8th second of evocation so you would have to push it back to a 6 second channel to lose any mana.

Wouldn't that change after 2.2 where they are modifying evocation so that you normally always get the extra tick?

It would be sad if we would need the set bonus just to bring evocation back to the same level that it was pre-haste items.

Aldric · 07/19/07, 11:48 AM

Originally Posted by Astrik

Wouldn't that change after 2.2 where they are modifying evocation so that you normally always get the extra tick?

It would be sad if we would need the set bonus just to bring evocation back to the same level that it was pre-haste items.

Thats a good question, anybody have some spell haste gear that wants to test this out on the PTR? It certainly looks like the patch notes say you'll get a tick on the 8th second of evo, which would make haste gear drop that gain. Maybe I can convince my GL to give me some hearts so I can do a character copy and figure it out, I'll post if I can swing that.

2.2 Patch Notes:
Power Regeneration: Any effect which triggers a change in your rate of power regeneration (Mana, Rage, Energy, Focus) will now cause an immediate reward of some power at the old rate of increase, and then begin new “ticks” of power at the new rate approximately 2 seconds later. This was done to improve functionality of abilities such as Evocation and Innervate so that they did not have wasted “ticks”.