WotLK - Complete Mage Compendium (3.3 PTR) - Page 116

khemael · 10/06/08, 7:56 PM

2) Take a random raid boss. Shoot it with Arcane Missiles. Then cast Slow and shoot it again. Realise that you just did 12% more damage.

The point is that... is any arcane mage really going to serve as a slave in order to refresh Slow ? Look at the dps of Mage_Arcane_Slow. No one want to be at this place.

The fact is TotW actually requires One person to constantly nerf 10% of his dps in order to give +12% dps to 2 maybe max 3 person in the raid ! Whereas Balance Druid for Example doesn't nerf their dps to give +13% dps to ALL casters.

Class serving as refresh-slave for 3max player is not something Blizzard does habitually.
Balancing Dps around something like TotW is not something Anyone can agree. Remember when Evocation was a talent and Arcane Points where compulsary. I thought Blizzard didn't want that anymore :/

And for now, Only Frostfire build can bypass this TotW-around balancing. That's why I'm sad.

If at least it worked with FB/FFB...but no. It isn't the case.

Look : I have just taken Mage_Arcane_Slow away to be more realistic :

Boom !

The only thing is I don't know why FrostFire is loosing 120 dps without Mage_Arcane_Slow since it doesn't use TotW :/ [EDIT : Problem Solved]

It's even worse in realistic situation where no frost mage is here and that one of the fire/Frostfire Mage has to refresh the Scorch Debuff.

EDIT : Idea : Has anyone tryed to modify the Lag= setting from simcraft ? It's quite interesting that some spec benefits more than others from lesser lag. At 0.1 Lag down from 0.2, DestroLocks are just under Affli. The others remain quite the same. Also, at High Lag= values (e.g. 0.5) Fire and FrostFire, like all Proc Spec are destroyed. It seems any laggy man would have to go with Frost-Like specs.

solbergb · 10/06/08, 8:01 PM

Originally Posted by Vand1

With so many of the new raid buffs focusing on upping crit rate, it seems to me that fire/frostfire mages will easily exceed 50% crit in raid situations. Has anybody run any number as to how much less valuable this makes Combustion?

Combustion has negative synergy with that but hot streak has positive synergy and combustion improves the hot streak proc rate, so it's mitigated somewhat by the extra pyroblast you likely get as a result of poping the buff.

The math gets a bit hairy as you have the following inputs:

base crit rate
current crit rate via combustion
multiple effects to consider that may all have different hit rates via talents (eg scorch, fireball and pyroblast may all have different crit rates)
stacking cooldowns (combustion can cluster crits while dmg increase/haste buffs are running)

There is a decent analysis of combustion at different rates of crit on wowwiki that should give a ballpark feel for how it degrades with crit rate.

It's possibe to do the math precisely and I assume tools like RAWR and Magegraf do exactly that. For me, the result wouldn't change my behavior so not much point (if I'm doing deep fire, I need combustion for dragon breath, and it does work well with stacked cooldowns, so I figure it's worth at least as much as the 1% PWF talent if I'm applying cooldowns properly, so not a "wasted" point)

dedmonwakeen · 10/06/08, 8:27 PM

Originally Posted by khemael

The only thing is I don't know why FrostFire is loosing 150 dps without Mage_Arcane_Slow since it doesn't use TotW :/.

Was Mage_Arcane_Slow casting Focus Magic on Mage_FrostFire?

khemael · 10/06/08, 8:32 PM

Was Mage_Arcane_Slow casting Focus Magic on Mage_FrostFire?

Yes, But it was corrected by a new focus Magic circle. Dunno why Frostfire is still 120 dps under the other sim...

EDIT : Problem solved. FrostFire Was refreshing Scorch

xiaoxin21 · 10/06/08, 8:49 PM

What is the best method to track hot streak procs? By observing the crits you get on mobs and counting internally or by observing the buff bars or any other methods like an addon?

I am having some problems noticing hot streak procs to the point I am not really looking at the fight itself and just staring at my buff bar for the icon to hit pyroblast,it get even worse with a fury warrior with rampage, the buff colour looks similar and sometimes I will hit pyroblast on mistake.

arch · 10/06/08, 9:08 PM

Personally, I always keep the scrolling damage numbers in view so I'll notice if I get two crits in a row. Other than that there is a very noticeable sound effect AND graphical animation. If this isn't enough I suppose you can tie it to a SCT event or any other kind of addon quite easily.

Gheed · 10/06/08, 9:11 PM

Originally Posted by xiaoxin21

What is the best method to track hot streak procs? By observing the crits you get on mobs and counting internally or by observing the buff bars or any other methods like an addon?

I am having some problems noticing hot streak procs to the point I am not really looking at the fight itself and just staring at my buff bar for the icon to hit pyroblast,it get even worse with a fury warrior with rampage, the buff colour looks similar and sometimes I will hit pyroblast on mistake.

This is easily solved with any add-on such as SCT or even easier with Power Auras, or just use the blizzard built in scrolling combat text, it also shows buff gains. Never had problem with not noticing hot streak at least.

pikapewpew · 10/06/08, 9:31 PM

Originally Posted by xiaoxin21

What is the best method to track hot streak procs? By observing the crits you get on mobs and counting internally or by observing the buff bars or any other methods like an addon?

I am having some problems noticing hot streak procs to the point I am not really looking at the fight itself and just staring at my buff bar for the icon to hit pyroblast,it get even worse with a fury warrior with rampage, the buff colour looks similar and sometimes I will hit pyroblast on mistake.

You can also use Classtimer and add yourself the timer.

Roywyn · 10/06/08, 10:40 PM

Originally Posted by xiaoxin21

What is the best method to track hot streak procs? By observing the crits you get on mobs and counting internally or by observing the buff bars or any other methods like an addon?

I am having some problems noticing hot streak procs to the point I am not really looking at the fight itself and just staring at my buff bar for the icon to hit pyroblast,it get even worse with a fury warrior with rampage, the buff colour looks similar and sometimes I will hit pyroblast on mistake.

MikScrollingBattleText (WotLK) : WoWInterface Downloads : WotLK Beta

Gets you flashy pop-ups for any kind of proc and any class. It been maintained pretty well throughout beta.
Pyros, Frostbite, Brain Freeze, Fingers, Barrage, you can't miss them with this mod.

Xentropy · 10/06/08, 11:59 PM

Originally Posted by khemael

The fact is TotW actually requires One person to constantly nerf 10% of his dps in order to give +12% dps to 2 maybe max 3 person in the raid ! Whereas Balance Druid for Example doesn't nerf their dps to give +13% dps to ALL casters.

<snip>

EDIT : Idea : Has anyone tryed to modify the Lag= setting from simcraft ? It's quite interesting that some spec benefits more than others from lesser lag. At 0.1 Lag down from 0.2, DestroLocks are just under Affli. The others remain quite the same. Also, at High Lag= values (e.g. 0.5) Fire and FrostFire, like all Proc Spec are destroyed. It seems any laggy man would have to go with Frost-Like specs.

To your first point, the even larger issue with that is if our dps is balanced around having that 12% dps, having less than about three mages means just bring one for the crit buff and no more. And if our dps is not balanced around having that 12% dps, top guilds will stack ten mages to increase rdps as close to 12% as possible. It's not a sustainable talent, at all. It cannot be balanced if it works on bosses, ever.

To your second, the last time I tested, the lag setting had a gigantic effect on dps, and Frost was actually among the heaviest affected by higher lag, because frostbolt is a faster cast than fireball. Adding half a second of lag to a 2.5 second cast is a bigger impact on dps than adding the same 0.5 seconds to a 3.0 second cast. Just lowering lag from 0.2 to 0.05 increased Frost by about 200dps and Fire by only 50dps, closing a good amount of the dps gap (though this was before Spell Impact gave Fire an insurmountable lead). I'd actually like to see lag=0 or at least something such as lag=0.05 as the standard for our dps testing, since we're trying to simulate the ideal, not the real, and in an ideal setting there will be no lag at all. Those with high normal latencies who are unable to compensate by the cast mechanics introduced in 2.3 will probably have learned for themselves by now to avoid short cast-time builds. The less often they have to press a button in that situation, the better.

Just a quick rundown of the dps output with the settings I'm using (which don't include TtW since I don't find it a sustainable balance):
Build - DPS w/ 200ms lag (Simcraft's default) - DPS w/ 50ms lag - delta
Mage_FrostFire - 4899 - 5267 - +7.51%
Mage_Fire - 4891 - 5259 - +7.52%
Mage_Arcane - 4350 - 4565 - +4.94%
Mage_Frost - 4208 - 4609 - +9.53%

(If nothing else, this shows how important hitting your cast button at the right time is--an input delay of just a tenth of a second per cast costs you around 5% of your dps regardless of spec. A large part of the "skill" of dpsing in WoW is just this timing. Due to the "partial GCD" you get if you hit your button too early, spamming the button will err on the side of extra lag vs proper single-press timing, unless you're fighting on the GCD with instants or 1.5s casts.)

Søndag · 10/07/08, 2:18 AM

[quote=khemael;924780]The point is that... is any arcane mage really going to serve as a slave in order to refresh Slow ? Look at the dps of Mage_Arcane_Slow. No one want to be at this place.

The fact is TotW actually requires One person to constantly nerf 10% of his dps in order to give +12% dps to 2 maybe max 3 person in the raid ! Whereas Balance Druid for Example doesn't nerf their dps to give +13% dps to ALL casters.

Class serving as refresh-slave for 3max player is not something Blizzard does habitually.
Balancing Dps around something like TotW is not something Anyone can agree. Remember when Evocation was a talent and Arcane Points where compulsary. I thought Blizzard didn't want that anymore :/

And for now, Only Frostfire build can bypass this TotW-around balancing. That's why I'm sad.

If at least it worked with FB/FFB...but no. It isn't the case.
QUOTE]

Khemael could you post some links to the builds and perhaps rotations you use? i'd like to see what you do differently in your models compared to mine. It's also hard to see why one build is better then another when we cant see the actionlist, gear settings and builds.

khemael · 10/07/08, 2:22 AM

Just a quick rundown of the dps output with the settings I'm using (which don't include TtW since I don't find it a sustainable balance):

Build - DPS w/ 200ms lag (Simcraft's default) - DPS w/ 50ms lag - delta
Mage_FrostFire - 4899 - 5267 - +7.51%
Mage_Fire - 4891 - 5259 - +7.52%
Mage_Arcane - 4350 - 4565 - +4.94%
Mage_Frost - 4208 - 4609 - +9.53%

Why is you're Fire build so high compared to FrostFire if you say you're not using TtW ? I got 300-400 dps difference for FrostFire VS Fire by not using TtW.

To your first point, the even larger issue with that is if our dps is balanced around having that 12% dps, having less than about three mages means just bring one for the crit buff and no more. And if our dps is not balanced around having that 12% dps, top guilds will stack ten mages to increase rdps as close to 12% as possible. It's not a sustainable talent, at all. It cannot be balanced if it works on bosses, ever

.

Then it seems I'm right here when I say that Only FrostFire could really compete for now ?

Khemael could you post some links to the builds and perhaps rotations you use? i'd like to see what you do differently in your models compared to mine. It's also hard to see why one build is better then another when we cant see the actionlist, gear settings and builds.

The builds are strictly the same of the LastReport of SimulationCraft, just as the rotations Are. The only thing which was modified from one graph to another was Mage_Arcane_Slow Deleted and a new Circle for Focus Magic properly working.

mage=Mage_Arcane
actions=flask,type=blinding_light/wizard_oil/molten_armor/focus_magic,target=Mage_Fire/mana_gem,trigger=3500/mana_potion,trigger=3500
/evocation,trigger=10000/arcane_missiles,barrage=1/presence_of_mind,pyroblast/arcane_barrage/fire_ball
talents=http://talent.mmo-champion.com/?mage=0055030012003000000000000000000000000000000000000000000023000503110033015032300 0515321

mage=Mage_Fire
actions=flask,type=pure_death/wizard_oil/arcane_brilliance/
molten_armor/focus_magic,target=Mage_Frost/mana_gem,trigger=3500/mana_potion,trigger=3500/evocation,trigger=10000/combustion/pyroblast,hot_streak=1/fire_ball
talents=http://talent.mmo-champion.com/?mage=0355030012003330053120300350000000000000000000000000000023000503110033000000000 0000000

mage=Mage_Frost
actions=flask,type=pure_death/wizard_oil/arcane_brilliance/molten_armor/focus_magic,target=Mage_Arcane/mana_gem,trigger=3500/mana_potion,trigger=2500
/ice_lance,frozen=1,fb_priority=1/evocation,trigger=10000/water_elemental/icy_veins/cold_snap/fire_ball,brain_freeze=1/frost_bolt
talents=http://talent.mmo-champion.com/?mage=0000000000000000000000000000053303031023310023005223135023000503110003000000000 0000000

mage=Mage_FrostFire
actions=flask,type=pure_death/wizard_oil/arcane_brilliance/molten_armor/mana_gem,trigger=3500/mana_potion,trigger=3500/scorch,debuff=1/
evocation,trigger=10000/combustion/icy_veins/pyroblast,hot_streak=1/frostfire_bolt
talents=http://talent.mmo-champion.com/?mage=2305030012303331053120301351203303031003000000000000000000000000000000000000000 0000000

Setia · 10/07/08, 2:40 AM

Originally Posted by Xentropy

Due to the "partial GCD" you get if you hit your button too early, spamming the button will err on the side of extra lag vs proper single-press timing, unless you're fighting on the GCD with instants or 1.5s casts.)

Do you have any proof of that? It sure doesn't work that way on Live - spamming a key works just fine...

And I just tested it on the beta server. Spamming my key every 10 ms (through AutoHotkey) gave me zero noticeable lag on average. On 35 Frostfire Bolts, the average time between two bolts is 2.87 seconds, which is exactly my cast time given my Haste rating, so any lag I actually get must be under 10 ms (the expected average delay, given my the speed at which I spam the button, is 5 ms). And my ping is 200-300 ms on the Beta server.

chase · 10/07/08, 2:41 AM

Originally Posted by Bladestrom

Ah ok, I assumed following implied Focus Magic would get overwritten.

Patch 3.0.2 - On Debuffs, Buffs, and Raid Stacking
..

Increased Spell Power Buff: Focus Magic, Improved Divine Spirit, Flametongue Totem, Totem of Wrath, Demonic Pact

Focus magic has not been a spell power buff for several patches now.

Soul · 10/07/08, 3:37 AM

Originally Posted by Setia

Do you have any proof of that? It sure doesn't work that way on Live - spamming a key works just fine...

And I just tested it on the beta server. Spamming my key every 10 ms (through AutoHotkey) gave me zero noticeable lag on average. On 35 Frostfire Bolts, the average time between two bolts is 2.87 seconds, which is exactly my cast time given my Haste rating, so any lag I actually get must be under 10 ms (the expected average delay, given my the speed at which I spam the button, is 5 ms). And my ping is 200-300 ms on the Beta server.

I've definitely seen it work that way on Live. My latency tends to be 300ms or so and it's quite noticeable what happens if you miscast. That's why I had that whole long section on lag and spellcasting in the TTT article.

Søndag · 10/07/08, 4:52 AM

I've just updated my scoreboard, now the list has been split in three categories and i have tried to model the builds proposed by Manly on the last page.

I'd appreciate more input on the realist builds, especially frost - since i have not played a frost mage since levelling.

You can find the list using the link in my signature!

grayrest · 10/07/08, 5:18 AM

Originally Posted by khemael

Why is you're Fire build so high compared to FrostFire if you say you're not using TtW ? I got 300-400 dps difference for FrostFire VS Fire by not using TtW.

Since TtW affects Fireball and Frostfire Bolt equally, there shouldn't be a huge dps difference between the two with/without.

Originally Posted by Ghostcrawler

We think we have Elemental's dps to a pretty good point now. There are some classes that are doing too much dps in your beta build, which may be one of the reason you feel low. But in our tests, Elemental is above the tanks and healers and on par with everyone who isn't broken.

This was just posted on the shaman changes thread. It's interesting for us to know who's broken.

Xentropy · 10/07/08, 5:31 AM

Originally Posted by khemael

Why is you're Fire build so high compared to FrostFire if you say you're not using TtW ? I got 300-400 dps difference for FrostFire VS Fire by not using TtW.

I noticed I was accidentally not using icy veins, but that still only gives my FFB 5387 dps vs FB's 5257. I'd need more information to determine why yours is higher or mine is lower than expected (or some of each).

By the way, please split the action= lines in your quote. The long lines break the forum and cause a horizontal scrollbar to appear onscreen, making reading posts a pain.

Originally Posted by Setia

Do you have any proof of that? It sure doesn't work that way on Live - spamming a key works just fine...

I'm positive, and it works exactly that way Live. When you press a key and are not on GCD, whether or not the spell is ready to cast, a GCD starts on your client. No keypresses (of abilities on the GCD) are sent to the server during this GCD. When the command you sent reaches the server and the server determines it was received too soon, it sends back a cast failed, not ready error. When your client receives the error, the GCD that started when you hit the key is cancelled and it'll send server commands again. Therefore, pressing the button too early causes a GCD equal to your latency to occur, during which further keypresses are ignored. With lag lower than your normal reaction time, this makes no difference, but for people with, say, 200ms latencies, spamming instead of using Quartz and timing properly is costing them an average of half their latency per cast.

Timeline (for 500ms round-trip latency to show the issue most prominently, offset by 10ms to show the worst possible case -- real time lost to spamming will average half your latency):
0.00 Client begins cast of 3s spell.
0.25 Server receives cast command and starts cast.
0.50 Client receives cast success and Quartz adjusts the cast bar properly for latency.
1.50 Original GCD completes, button-spammer tries to cast again.
1.99 "Fake" GCD cancelled by server response. Button-spammer tries to cast yet again.
2.49 Another "fake" GCD cancelled by server response, Button-spammer tries to cast again.
2.99 Another "fake" GCD cancelled by server response. Button-spammer tries to cast again.
3.01 Spell enters the "cast next" red section of Quartz castbar. Non-spammer presses next cast. Spammer's cast attempts are ignored and not even sent to the server because client is in GCD!
3.25 Original spell finishes casting.
3.26 Non-spammer next cast begins.
3.49 "Fake" GCD cancelled by server response and button-spammer finally sends the cast attempt that will work.
3.74 Spammer next cast begins.

The higher your lag, the LESS reliant you can be on button-spamming due to the temporary GCD's incurred by the client. If you have 10ms latency all the time (which I suspect is the case since you haven't noticed your "fake" GCD's--they're very noticable to someone with 200ms like I tend to have), by all means spam away (though use of AutoHotKey to autorepeat a key has been explicitly mentioned by blue as bannable botting, so I'd be careful there). If you have latency that is higher than half your reaction time to a Quartz castbar line moving into a red area, get Quartz and time your casts properly to gain free dps (not to mention save your jackhammering finger).

Selmarix · 10/07/08, 5:54 AM

That was the old behavior which severely penalized high latency players. So they changed it some time ago and now the "fake" gcd does not prevent your client from sending the command again.
The result is that button spamming does work.

Althor · 10/07/08, 5:55 AM

Originally Posted by Xentropy

I noticed I was accidentally not using icy veins, but that still only gives my FFB 5387 dps vs FB's 5257. I'd need more information to determine why yours is higher or mine is lower than expected (or some of each).

By the way, please split the action= lines in your quote. The long lines break the forum and cause a horizontal scrollbar to appear onscreen, making reading posts a pain.

I'm positive, and it works exactly that way Live. When you press a key and are not on GCD, whether or not the spell is ready to cast, a GCD starts on your client. No keypresses (of abilities on the GCD) are sent to the server during this GCD. When the command you sent reaches the server and the server determines it was received too soon, it sends back a cast failed, not ready error. When your client receives the error, the GCD that started when you hit the key is cancelled and it'll send server commands again. Therefore, pressing the button too early causes a GCD equal to your latency to occur, during which further keypresses are ignored. With lag lower than your normal reaction time, this makes no difference, but for people with, say, 200ms latencies, spamming instead of using Quartz and timing properly is costing them an average of half their latency per cast.

Timeline (for 500ms round-trip latency to show the issue most prominently, offset by 10ms to show the worst possible case -- real time lost to spamming will average half your latency):
0.00 Client begins cast of 3s spell.
0.25 Server receives cast command and starts cast.
0.50 Client receives cast success and Quartz adjusts the cast bar properly for latency.
1.50 Original GCD completes, button-spammer tries to cast again.
1.99 "Fake" GCD cancelled by server response. Button-spammer tries to cast yet again.
2.49 Another "fake" GCD cancelled by server response, Button-spammer tries to cast again.
2.99 Another "fake" GCD cancelled by server response. Button-spammer tries to cast again.
3.01 Spell enters the "cast next" red section of Quartz castbar. Non-spammer presses next cast. Spammer's cast attempts are ignored and not even sent to the server because client is in GCD!
3.25 Original spell finishes casting.
3.26 Non-spammer next cast begins.
3.49 "Fake" GCD cancelled by server response and button-spammer finally sends the cast attempt that will work.
3.74 Spammer next cast begins.

The higher your lag, the LESS reliant you can be on button-spamming due to the temporary GCD's incurred by the client. If you have 10ms latency all the time (which I suspect is the case since you haven't noticed your "fake" GCD's--they're very noticable to someone with 200ms like I tend to have), by all means spam away (though use of AutoHotKey to autorepeat a key has been explicitly mentioned by blue as bannable botting, so I'd be careful there). If you have latency that is higher than half your reaction time to a Quartz castbar line moving into a red area, get Quartz and time your casts properly to gain free dps (not to mention save your jackhammering finger).

As someone that plays on a 300-400ms Quartz cast time latency and has a G15 keyboard and tested that casting changes when they were introduced in 2.3 or 2.4 I can say this is wrong.

I was quite capable of macroing a rank 1 lesser heal to a G-key with it set to auto-repeat every 10ms and recording the combat log and seeing that it would indeed chain cast it perfectly fine with no added latency. i.e. my heals (with no haste gear) were going off about every 1.51 to 1.53 seconds. All this while I had a 250-450 variable cast latency.

It's a simple test for anyone with a G15 keyboard or something like AutoHotKey.
I really suggest those of you that don't believe it, try it out for yourselves.

The only type of spells where it's detrimental to spam the button are channeled spells, because for whatever reason Blizzard decided that they should be interruptable just by starting a new cast (once you're past the GCD). This by the way is one of the big problems I have with their ideas about trying to make Shadow Priests scale equivalently to other classes. The fact that they don't recognise this as issue really blows my mind.

Roywyn · 10/07/08, 6:28 AM

Updated most things on the front page.
Still have a level 70 spec link to do.

Also lacking info about Mirror Image scaling-

Søndag · 10/07/08, 7:33 AM

Originally Posted by khemael

The point is that... is any arcane mage really going to serve as a slave in order to refresh Slow ? Look at the dps of Mage_Arcane_Slow. No one want to be at this place.

Actually throwing a slow does not have to be a massive dps drop for an arcane mage, and a group of mages could rotate who spams and all would end up gaining.. but let's see if totw keeps working like it does now.

Maje · 10/07/08, 7:47 AM

Originally Posted by Søndag

Actually throwing a slow does not have to be a massive dps drop for an arcane mage, and a group of mages could rotate who spams and all would end up gaining.. but let's see if totw keeps working like it does now.

I'm not sure I see that happening, maybe if slow was a learned spell, but it's not, you need 41 arcane to cast it. Somehow I don't see a raid with arcane mages (more then one that is), and even if it was possible it would be a horrendous design flaw which pigeonholes every mage into 40 arcane.

As it is now, having one arcane mage to be a slow 'bitch' and another mage that took torment is already broken, that means you perform subpar when the slow isn't in effect. But I guess it's too early to tell (7 days till the patch...), the whole talent seems broken, you can't take it if your frostfire and you can't apply it if you're fire. It might be a boost to frost if it could be applied by frostbolt but then 12% would make every mage frost.

Jonny_Monroe · 10/07/08, 8:25 AM

Can anyone report on DPCT values for the arcane spell cycle? I'm looking at a priority system rather than a cast cycle system. After my awesome-looking ABar/Fireball combination launches (which I've lovingly named the Super Haduken), MBAM normally comes up and I'm not sure whick spell to cast next. Know MBAM out-damages fireball on the same cast time, but how does it compare to ABar? My testing puts it as 'close', and my gut tells me the best thing is blow the MBAM once the current GCD is up then ABar then back to my super hadukens (As an aside - it is very easy to hit ABar at the last 40-30% of a cast and have both spells launch simultaneously; you really don't need to spam, its an easy window to hit).

Also; does clearcasting/potency change priorities? With fireball having a much nicer damage ratio but MBAM eating into your mana pool it would be nice to see DPCT/MPD under normal, clearcast, potency and both to work out the best priority.

Loco7 · 10/07/08, 9:23 AM

Originally Posted by Xentropy

The higher your lag, the LESS reliant you can be on button-spamming due to the temporary GCD's incurred by the client. If you have 10ms latency all the time (which I suspect is the case since you haven't noticed your "fake" GCD's--they're very noticable to someone with 200ms like I tend to have), by all means spam away (though use of AutoHotKey to autorepeat a key has been explicitly mentioned by blue as bannable botting, so I'd be careful there). If you have latency that is higher than half your reaction time to a Quartz castbar line moving into a red area, get Quartz and time your casts properly to gain free dps (not to mention save your jackhammering finger).

This is very accurate.

I have been using single press, timed casting for 3 years now. The original version of Quartz (i cannot remember the name) pointed to this.

The only other casting technique I have found to significantly up my dps, that I haven't seen noted anywhere, is your spells do not actually cast on key press, they cast on key release.

That means you can make up a very small amount of time per cast by holding down your next spell's hotkey, till the GCD/spell is ready.

The only drawback I have found to this is using reactionary spells such as insta pyro. If you are holding down say your fireball key, then to cast a reactionary you need to use your proper hotkey, while not releasing the first hotkey. Or you get a GCD.