Ele'

I'd say that the destro. warlocks and shadow priests will make a better use of haste, because affliction warlocks have more haste-capped spells (corruption, syphon).

There where some talks about that in the warlocks megathread around post #2140 - 2150.

As far as I know, the pets don't scale with your haste rating (as they don't scale with your hit rating, and so on). It's a shame, in my opinion :-/

 

Noxtenebrae

Yeah, Roywyn, these rumors get perpetuated, unfortunately. And yes, it seems that the GCD remains unaffected by both Heroism and Bloodlust.

 

Xelopheris

Skimmed and didn't find anything about Ferals in here.

White Damage
My white damage, depending on group composition, averages from 30 to 40% of my total damage on a bleedable mob. Druids do a LOT more of their damage from yellow attacks than whites, so haste rating has a minimal immediate effect on their DPS.

Omen of Clarity
Based on numerous WWS reports that I've compiled with myself, Omen of Clarity has about a 2% proc rate.

Assuming that for every time I mangle, there is 2.5 shreds average, and one rip, that means I spend a total of 175 energy over 4.5 attacks, or 38.88 energy per attack average.

A 2% chance on hit will proc once every 50 seconds on white damage with no haste. Thus, every 50 seconds you get 38.88 energy, which equates to 0.77~ energy/second. Because of the way haste multiplies, 100% haste will give you 100% more Omen of Clarity procs. However, given that your normal regen is 10/sec, and omen of clarity averages 0.77~/sec, so you can see how little of your DPS is represented by OoC (7.7% of 60% is 4.62%)

2t4 - Malorne 2 piece cat bonus

4% chance on attack to restore 20 energy. This is pretty simple, every 25 attacks, you get 20 energy. This is an average of 0.8 energy/second. Very simple math here, based on Omen of Clarity 7.7 ~ 4.62% total DPS, 0.8 is equivalent to 4.8% of your yellow DPS.

Final result

Assuming at first 40% white damage, 60% yellow damage, and then a bonus if 4.62% yellow and 4.8% yellow from Omen of Clarity and 2 piece tier 4, white damage now represents 36% of your white DPS. Your yellow damage from non-natural regen represents represents 8.6% of your total damage. A total of 44.6% of your damage is affected by haste. This is FAR less than warriors, rogues, hunters, enh shaman. Even ret paladins benefit from haste more.

Notes

I did not account for bonus energy procs off of attacks generated by bonus energy. Thus, slightly more of your DPS is hasteable, but I wouldn't stretch much farther than a total of 45% hasteable DPS.

What this means
If you are currently doing 1000 DPS, and get a bloodlust, the 30% increase only applies to 45% of your DPS. Thus, you only get about 135 extra DPS from it. Bloodlust gives a mere 13.5% increase in damage dealt over it's duration.

 

lorelye

Can we recap for casters?

"When chain casting a spell, latency adds a flat idle duration in between each cast. "
- This was removed in patch 2.3, no?

- 15.7 is the correct number as of patch 2.3 for conversion? 15.7 spell haste = 1%?

You also say that spell haste is calculated before talents? Are you certain?
- So, fireball for example, 3.5 sec cast time. Remove 1% = -.035 sec. Then we remove the .5 sec reduction Talent afterwards, for 2.965 sec cast time. In an ideal world, this would be slightly more than a 1% increase in damage per second (1% = 1.0118%). Odd, no?

- Is it correct that spell haste will not apply to spells with cast time 1.5 seconds or less?

I can imagine that 10% spell haste might be noticeable. But would you, even with Quartz, notice a 1% difference? Is anyone that good?

Are Quartz users now removing their /stopcasting macros?

Can someone explain better, what was the plus that fire/ice got in 2.3? They removed what exactly?

Thx
Lorelye

 

Daidalos

Short answer yes. Longer answer it there is a gcd check that happens where client cannot attempt to recast again while spamming the button so latency has some affect.

yes

It is after talents.

Not exactly. It does apply it just does not allow the next spell to be cast under the gcd


1%? not really

yes
That has nothing to do with haste. Mages no longer have a coefficient reduction when lowering thier cast time of fireball/frostbolt

 

tedv

I've been thinking about how to mathematically model the DPS increase of H% spell haste, and I believe I've come up with a reasonable model. First, the assumptions.

1) Any time saved from spell haste is converted into extra casts of a spell. I've heard arguments that you can't really account for a .1 second decrease in time on a 3 second cast because of latency, but I contend that a 3 second cast that takes 3.2 seconds on average will turn into a 2.9 second cast that takes 3.1 seconds on average, since you can't perfectly account for the exact cast time without spell haste either.

2) There is a constant amount of cast time spend on spells affected by haste and the remainder cannot be hasted. Technically this is false because of spells with cooldowns whose cast time is 1.5 seconds or less. For example, extra haste could mean you cast proportionately more Mind Blasts per minute because your mind flays line up closer with the Mind Blast cooldown. Or you could have so much spell haste that you must either wait a second for the Mind Blast cooldown to be ready, or cast Mind Flay now and use Mind Blast when when the flay finishes, effectively casting fewer Mind Blasts per minute. It's not clear how (in)significant this effect is, but it can push things in either direction, so I assume it averages out to no effect at all.

Next the variables:

X = Normal cast time of the spell you cast when you have nothing else to cast (ie. what spell the extra cast time gained from spell haste turns into)
D = DPS of spell X
T = Percent of the time you spend casting spells affected by spell haste (ie. cast time > 1.5 seconds)
H = Percent of spell haste gained


And the derivations:

X'

[top] Hasted cast time of X. Note that this is NOT X*(1-H).
X'

X / (1 + H)

G

[top] Time saved per second of casting a hasted spell versus the same spell unhasted.
G

1 - X'/X

G'

[top] Overall time saved throughout all casting
G'

T * G

D'

[top] Extra DPS gained from haste
D'

G' * D * X/X'

That last formula requires a bit of explanation. For every 1 second of cast time that would be spent without haste, when you have spell haste, you "save" G' seconds. Those G' seconds are spent casting a spell whose unhasted DPS is D. However, you spend that time casting a hasted version of the spell, which deals damage X/X' faster. The hasted spell has D * X/X' DPS, so you gain a total of G' * D * X/X' DPS.

Simplifying D' to the original values:

D' = G' * D * X/X'
D' = T * (1 - X'/X) * D * X/X'
D' = T*D * [ (1 - X'/X) * X/X' ]
D' = T*D * [ X/X' - 1 ]
D' = T*D * [ X / (X/[1+H]) - 1 ]
D' = T*D * [ [1+H] - 1 ]
D' = T*D*H

This means that an increase in haste provides a linear increase in DPS. Obviously it provides no increase in DPM, unlike spell damage.

So if you gain 1% spell haste and spend 100% of your time casting hastable spells, and your standard spell deals 1000 DPS, you'll gain 50 DPS (1% of 1000). Here's how this breaks out for shadow priests and fire mages in T6 content:

Shadow Priest:
D = 850 (Mind Flay)
T = 50%
1% haste: .01 * .5 * 850 = 4.25 DPS

Fire Mage:
D = 1400 (average Fireball including ignites)
T = 94% (for 8:1 Fireball:Scorch ratio)
1% haste: 13.16 DPS

Given that you need 15.6 spell haste for 1% haste, this works out to .27 DPS per point for shadow priests and .84 DPS per point for mages. At a very rough estimate of +1 spell damage = +0.5 DPS, we have the following equivalence estimates:

Shadow Priest: 1 spell haste = 0.54 spell damage
Fire Mage: 1 spell haste = 1.68 spell damage

I assume the numbers for destruction warlocks will be similar to fire mages, and the number for affliction warlocks will be similar to shadow priests. The dominating factors are obviously the power of the spammable damage spell and the ideal percent of time casting it. The more single minded a class's play style is, the better spell haste is. It's worth noting that spell haste is NEVER useless the same way spell crit is for shadow priests and affliction warlocks. But it's definitely not something these classes want to stack. The reverse seems to be true for mages.

 

Muphrid

tedv, I feel your analysis is very thorough, save for one particular spot...

The number for the Fire mage in particular strikes me as far, far higher than any previous valuation of spell haste, and I fear it is due to an underestimate of the DPS benefit of +1 spell damage.

Furthermore, let me present the following: while I feel your math is correct if you assume that a constant amount of time is spent on any given spell (no matter how fast it casts), I don't feel this is very useful, as rotations are picked based on how well the spells in the rotation "line up," so there is minimal lost time. Speeding up some aspects of a rotation but not others flies in the face of this and makes rotations break.

Thus, let me introduce the following variables, under the assumption that the actual rotation remains fixed by number of casts of a spell.

Z = total time of rotation
Z' = new total time of rotation
n

[top] number of casts of spell X

Z

n1*X1 + n2*X2. This is a two-spell rotation.
Z' = n1*X1/(1+H) + n2*X2. A two-spell rotation where only one spell receives haste benefit.

D' = D*(Z/Z'-1). The new DPS is just D*Z/Z'. Subtract D to get the change.

Z/Z'-1 = (Z-Z')/Z' = (n1*X1 + n2*X2 - n1*X1/(1+H) - n2*X2) / (n1*X1/(1+H) + n2*X2) = (n1*X1*(1-1/(1+h)) / (n1*X1/(1+h) + n2*X2) = (1-1/(1+H)) / (1/(1+H) + n2/n1*X2/X1) = H / (1 + (1+H)*n2/n1*X2/X1))

Note that, in your example, T = n1*X1 / (n1*X1 + n2*X2) = 1 / (1 + n2/n1*X2/X1). This is very close, but missing the (1+H) in the second term of the denominator, which is basically accounting for how the percentage of time spent on a specific, haste-able spell has changed due to haste itself.

D must also be the DPS of the whole rotation: since the expected damage of the rotation remains fixed (only the time over which it is applied changes), the actual DPS increase depends on all components (both haste-able and non-haste-able parts) of the rotation.

 

F.U.B.A.R.

So, i've read the entire thread and I know that now Improved Fireball/Frostbolt doesn't affect dmg coefficient, but i found little confirmation about if Spell Haste Rating does or doesn't affect Dmg Coefficient by lowering the casting speed of spells.
Does someone have some info about this?

 

Prinsesa

The cast time reduction is calculated AFTER talents, with one exception: Holy Light's interaction with Light's Grace. Any spell haste you have (as in Dawnsteel Shoulders) calculates its cast time reduction based on the unbuffed Holy Light's 2.5 second cast time. Ostensibly, this is because Holy Light is only reduced to a 2.0 second cast if Light's Grace is up, whereas Fireball will always cast at 3.0 seconds with Improved Fireball.

Spell haste will not affect your damage coefficients, as in your Fireballs will hit just as hard with or without Bloodlust.

http://armory.mmo-champion.com.nyud....63815BffMm.png

 

• Xei

Is there anyway I can quantify that? Now moving into the BT/Hyjal/ZA Haste loot world, I was interested in the stat ... but I am almost certain playing nightly at an average 450ms will have a pretty hefty impact on the effectiveness of Caster Haste.

I am not a huge TC'er, but I do like to read and am willing to do some testing to quantify the effect latency has on caster haste rating - for myself, but also for all other Oceanic WoW players.

"Being a leader is not a position of power. It is a position of service." ~ Barestomper

WTB Oceanic Horde PvE Players

 

Daidalos

Well this is dicussed a few times in more detail, but it would most likely be best to actually see if some other oceanic / high ping players can post some wws or logs for us to look at. My understanding on the mechanic is basically when you attempt to cast the client goes on GCD then checks with the server if it is truely on GCD. If you are not (due to already being casting) the GCD is removed and you are then free to attempt to cast again.

What we need to verify is if spell cast requests are still sent to the server when client is on GCD. I would assume the client blocks spell requests from being sent while the client is on GCD. I know bliz was tinkering with this for awhile after the 2.3 patch due to some issues this system. It may be that spell requests are sent till teh client has confirmed GCD.. or the server checks for GCD server side and rejects spell casts requests. If anyone has or can point to some data on this it would clear this up.

 

Sota

As is noted in the big patch thread elsewhere on EJ, 2.4 is scheduled to change spell haste so that it also reduces the global cooldown. Insta-casters rejoice.

Link: WoW-Europe.com Forums -> 01/02 2.4 Patch notes

 

yaya@EU-hyjal

With this change, 788 spell haste rating will bring the global cooldown from 1.5s to 1s. With bloodlust, 243 haste rating is needed.

Therefore haste will become a very powerful stat for all casters up to at least 243 (its usefulness beyond that point depends on several factors such as class, spec, and group composition).

edit : corrected the value with bloodlust, since it stacks multiplicatively with haste as pointed out by muphrid

 

Muphrid

Only 243 haste rating is needed; Bloodlust/Heroism stack multiplicatively, not additively, with spell haste.

 

Goatharder

I think that there are two different schools of thought on shadowpriest DPS: that our gear choices maximize our DPS, or that our gear choices maximize mana return. At first glance, these seem to be the exact same (more dps = more mana return), but it ignores one of the key requirements for ideal mana return to exist: as near to 100% VT uptime as the shadow priest can achieve. Haste throws that balance out of whack.

What haste does, in affecting only mind flay, is minimal. Even if you can reduce cast time on mind flay, your actual DPS will not go up if you are attempting 100% VT uptime, unless you have huge amounts of haste. The key to this is you must eke out 1.5 seconds of extra cast time every 15 seconds. Assuming absolutely ideal circumstances (SW: D on cooldown such that you can only use it once, SW:P refreshed right before the VT), you're looking at a cast sequence like this (assuming zero lag and a perfect player):

00:00 VT
01:30 MB
03:00 Flay
06:00 Flay
09:00 MB
10:30 SW: D
12:00 Flay
15:00 VT

As the VT casts cannot move, they 'must' be 15 seconds apart, you would need to shave off a full half second on a single mind flay to fit one more GCD in there. The benifit from that would also only be situational, as the spells you would 'fit' into the extra space are not spammable spells, and actually being able to cast more would rely on a specific window of opportunity (both MB and SW: D being off cooldown, or very near to it, at the beginning of the 'VT cycle'.

Post 2.4, this 'magic number' of haste that will allow you to cast one more GCD worth of spells within the timeframe will be easier to achieve, but it will still not be worth the value that most simulations give it, simply because extra cast opportunity time isn't worth it with the VT limitation.

Something to note, and I mentioned this in the beginning part of the post, is that all of the above is dependent upon the shadowpriest prioritizing regen over raw DPS numbers. If you were willing to lose the 'every 15 seconds' VT requirement, I am sure your overall DPS would go up more linearly with haste than without, however you still will not reach the DPS numbers of 'pure' classes, and would be sacrificing one of the most important reasons shadow priests are brought to a raid.

 

Muphrid

Goatherder, if you would be so kind as to properly attribute that quote to tedv instead of to me, that'd be fantastic.

In addition, your concern would be addressed if haste properly increased the tick speed of VT as it should (ideally), essentially allowing you to use the time that would be lost to waiting for VT recasts.

 

Nerhtal

Right i must have not known this before or, well, im simply a retard. Well i am a bit i guess as im a full haste Spriest (i did the +dmg thing pre-tbc throughout MC, BWL, AQ40 and The 2 and a half areas of naxx my guild did before the expansion pack hit)

15.7 Haste Rating = 1% Haste

So 246 Haste rating should be more then 13.5% haste then?

My Rez is 8.65s (10s Normal)
My Flay is 2.6s (3s Normal)
My 1.5s spells are now 1.3s

Me and my friend sat down and thuoght about this and seemed to come out to the fact that each items rating is applied separately. Well mostly my friend, i was busy melting BT mobs faces off.

So the first item reduces the cast time, the second item uses the new cast time value for the new calculation.

Anyone able to confirm if this is how it is meant to be, or wether I simply mis-interpreted how the 15.7 Rating to % conversion is done.

As in, 246/15.7=15.67% (rounded up)

Working on that my Rez should be under 8.5s (8.43s) yet its 8.65s which is a 13.5% reduction instead. I unfortunately didnt do the calculations per item my friend (arcane haste mage experiment) did and he said that it seems that wht i mentioned above is happening.

So, using my 246 shr value and the actual effect im on 18.22 Spell Haste Rating = 1% effective spell haste. Is this to be expected?

 

yaya@EU-hyjal

If your haste rating is x and the base cast time of a spell is c, the effective cast time of this spell is

c / ( 1 + x/1577)

So for instance for your 10s rez spell, with 246 haste, the cast time is 10 / (1 + 246/1577) = 8.65 s

Hope it answers your question.

 

Nerhtal

Ah right ok, so not quite how i expected it. Tyvm.

 

Covale

Has there been any word on the actual formula being used to calculate the amount of reduction in the global cooldown per point of spell haste?