It's worth noting that the more haste you have, the better Mystical Skyfire is, because less of the proc is wasted on unhasted time. That is, if the buff lasts 4 seconds and you proc it on a mind flay when your flays take 3 seconds, you have 1 second to use it, so you only get one hasted spell. But if your mind flays are 2.5 seconds, you'll have 1.5 seconds of haste buff, so you can cast some instant cast spell (invoking a close to 1 second GCD) and the cast a hasted mind flay afterward, so you get two hasted spells.

Also, the spell data on wowhead definitely says 6 seconds: Spell Focus Trigger - Spells - World of Warcraft

Do we know if the tooltip is bugged or the buff? A 4 second duration will bring the value down to 24 spell damage guaranteed, and probably closer to 20 (but increasing towards 24 as you have more spell haste).

Against that however, the more haste you have, the better +dmg becomes (relatively speaking). So the +14 dmg 2% int would be worth comparatively more.

I'm not sure which effect would be bigger, but it could be that the MSD becomes less useful as you get more haste.

One point to note is that if it pushes your haste rating above 780ish (i.e. reducing the GCD to 1 second), there will be very little additional benefit beyond that point. That would be a huge amount of haste however.

Its 4 seconds on PTR

Yes, we know.

Is that intended, is the question, given that the tooltip says 6 seconds?

Is it possible to get a 3d graph?

x-axis: spell dmg 1200-2000
y-axis: haste rating 0-500
z-axis: dps

I think that would answer the more important questions: at what spell damage is 1 haste = 1 spell damage.

I've been meaning to do this for a while I just have been too preocupied doing other things with my computer like playing WoW. Each simulation of 500 steps takes about 5 minutes so doing a graph like that would take about 65 hours. It'll also be quite bumpy as most of these are, if I went for 100,000 seconds instead of 10,000, which I have done on a few occasions for looking at things in a bit more detail it would take 650 hours or about a month. However, since I think the linear relation ship is pretty well established the overall goal of gear should simply be to make your SP+Haste value be as high as possible, or perhaps a SP+.9*Haste.

If I compiled it into a program I'm sure it would run faster but I wrote it to run in mathematica which is a program that is quite similar to matlab. I really don't think that at this point it would be valuable to do anything more with the program since the purpose of it, that is find the value of haste rating in terms of spell power, has largely been accomplished. What would probably be better would be to start now looking at specific gear sets and then either using it to calculate specific dps or to use it to determine what gemming would be ideal. I would imagine in either case that Aevie's and Mefadin's lists from Shadowpriest.com would hold all the answers anyone has about haste and the gear they should be going for in sunwell.

Id bet that the tooltip is wrong. That seems more likely than the spell itself messing up.


I am a little disappointed in this nerf. Was there a pvp motivation behind nerfing MSF? In pve raiding I think SPs are the only ones that are using it (or, should be using it) and it seems hard to believe that a specific pve raid dps nerf to Shadow Priests was necessary... at all. I was kinda hoping there would be a new gem that would fit for SPs better than the MSF of 2.3 although I have no idea what an "ideal" gem would be for us. Chance on spell cast to gain +damage of some kind?

Actually, if they extended the duration on "new" MSD to 8-10 seconds, it would have been a buff with the new mechanics. Even at 6 it would have been tolerable. At 4... not so much.

If it stays at 4, it's a tossup whether ESD or MSD is better - MSD probably wins out if you can use the single blue for MSD to net a meaningful socket bonus, while ESD probably wins out if you can get socket bonuses out of haste/dmg gems. Looking through the Sunwell gear, it seems like the only item where a blue nets you a socket bonus is T6 boots, while you can legitimately use dmg/haste gemming in quite a few locations to net socket bonuses.

Based on the gear trend, and the current state of MSD, I think we'll wind up shifting to ESD (as mediocre as it is) just so we can use orange gems.

Do you really think we will need to have a majority of [Quick Lionseye] to get the numbers we want or need? I believe perfect amounts of haste are already attainable, and almost every piece of new gear for shadows from 2.4 has haste on it, including the new slots of t6. I'd assume the new gear plus the [Reckless Pyrestone] will be best.

Well, we've said that 1 damage = 1 haste rating, right? And that it might be closer to 1 haste rating = .8-.9 damage. That means unless you're trying to get a socket bonus, a +12 damage gem is still better by a fair margin than a 10 haste gem, specifically 2 to 4 damage better.

Well, if we're discussing socketing for the theoretical "perfect" amount of haste then it is relevant. If I'm reading the graph on the previous page correctly, at around 150 haste it starts to get "worse" in comparison to spell damage. Yes, it's still a damage increase, however, if you're socketing the haste/damage gem to get a bonus and your haste is below 150, the haste is worth more than if you were socketing the same item but you already had 200 haste.

If the ratio of haste:damage is 1:1 when you have less than 150 haste, and 1:.8 or 1:.9 when you have greater than 150 haste, it most certainly makes a difference in gem choices.

I could be reading the graph wrong though.

Plus the whole thing for the 'perfect' amount revolves around having a 'clean' rotation and not having your abilities come out of CD or drop off the target at weird times where you can't possibly put them up again without a dps decrease etc. The perfect amount refers to an equal increase in haste where I believe 2 flays fit perfectly between MBs etc. I haven't got intensely into Haste just yet and all my knowledge is based on spreadsheet, lists, and theories at SP.com and not here. I'm rather new here on EJ.com

The general thing I'm seeing is that haste is getting a boost from .508 to a full 1 or very close, that and it affect GCDs so the simple math of 'more is better' doesn't really apply to this very VERY dynamic stat, which relies on the +dmg you have at any given time.

for example you just said it's always 1 to 1 haste to dmg.. well can't I have 0 dmg and 2k haste and do more dps then? I somehow think that theory breaks right there. The numbers vary greatly every time you add more of each.

No one said anything of the sort. That said, it looks to be ~1:1 for the first 200 or so haste, and converts quite favorably up to very high levels; using a 1:1 or 1:.9 approximation for quick evals is probably just fine.

The MB MF MF ideal is a bunch of shit, honestly. None of the simulation work cheeba or I have done has come anywhere near showing a spike in DPS at the "ideal haste" point because cooldown clashes, lag, and a number of other things will break that ideal cycle anyway. Not to mention the whole concept of a waitstate and clipped flays making MB MF MF less important that it's presented to be.

Looking at the itemization, generally you'd be trading 6 damage for 5 haste and a socket bonus... and the socket bonuses seem to be 3-5 damage in and of themselves. I think that haste/dmg gemming for socket bonuses will be quite common for spriests.

That's only true on items without blue sockets, which is almost none of them right now. Each blue sockets still trades you 6 damage for a socket bonus (2 damage) and either 7 stamina or 4 spirit. That seems never worth it to me. Even if the piece had two yellow and one blue socket for +5 damage, you have a net loss of 2 spell damage for 4 spirit or 7 stamina. The only blue socket you use is the T6 helm, because it's -1 damage for +7 stamina, or twice as good.

Are there enough pieces with only yellow or red sockets that you can gain the 6 to 10 spell damage loss by trading Mystical Skyfire Diamond for +14 spell damage, +2% int? I just don't see it.

The new JC neck for one... Sunfire set...

Thanks for the analysis. I'm glad that at least 40% of my gems won't be spinels. I'm not sure how much it's worth back-gemming T6 gloves is worth it. Depends on whether you include it in your 4 piece bonus, but that seems like the best long term option. Interestingly, this increases the value of T6 legs for keeping the 4 piece bonus, as they also have a yellow slot. Short term that decision will most likely be decided by loot drops.

Speaking as someone with a fairly high eval of haste: I won't be regemming my T6 gloves. That's just silly. I might consider picking up the T6 legs and using them on a shadow token, but that's really hard to justify - an extra leg enchant plus wasting the LCE/3xspinel I'm running as is? Since ESD has the easiest requirement EVER (>3 red), there's no real reason to regem a 50/50.

Keeping the T6 bonus as you gear is going to be interesting, for sure - on the brightish side, the tokens drop before any of the H/C/S/G replacement items do, so you can probably get a belt/bracer/boot before you lose a T6 piece.

After going through the list of gear for 2.4 it's pretty obvious that the 3 new pieces of T6 are going to be best in slot or very close it, so I'm wondering what people consider the best 4th piece, or is it even worth it. I'm of the mind that the helm is most likely the best trade off however I've seen a lot of people going with the shoulders as well.

Yeah helm is pretty clearly best 4th t6 item according to my numbers at least.
[Helm of Arcane Purity] -> [Hood of Absolution] is a smaller loss than [Amice of the Convoker] -> [Shoulderpads of Absolution]

The only items where I'd consider it a clear win to go for the socket bonus over straight 12dmg gems would be the shoulders above and the gloves with the same socket setup. If you're keeping the t6 helm where it is only a 1dmg loss to use the blue gem it seems like a tossup with the current numbers for the metas which to use. Still hoping they might listen to the pleas to change the 2% int part on the ESD to something not quite so worthless though.

What about keeping the T6 gloves? Gloves have fewer stat points allocated to them than other slots, so it's the natural choice. I count a 21 spell damage loss from keeping the T6 helm by the way, which is not insubstantial.

Er nevermind. [Gloves of Tyri's Power] is grotesquely better than [Handguards of Absolution].

OK i'm trying to code my own dps simulator to see how spell haste will affect shadowpriest's spell rotation but i'm having some problems with it... Can anyone help with it? I'm a bit noob about this programming stuff...

Here it is:


#include <stdio.h>
#include <stdlib.h>

void globalcd(int vt, int swp, int swd, int mb, int ve, int t, int gcd)
// This function represents the time flow, we consider the time has passed equal to gcd everytime we call this function.
// vt,swp,swd,mb,ve values are spell cooldown counters (dot durations are considered cooldowns). Gcd is global cooldown.
// When a spell is NOT on cooldown, that means its cooldown counter is equal to a greater number than it's cooldown.
// t is time (miliseconds)
{
t=t+gcd;
swp=swp+gcd;
vt=vt+gcd;
ve=ve+gcd;
mb=mb+gcd;
swd=swd+gcd;
}

int main()
{
int vt, swp, swd, mb, ve, t, gcd, tmax, n_vt, n_swp, n_mb, n_ve, n_swd, n_mf, totalmanaspent, spellhasterating, spellhaste;
// n_vt represents how many times we cast v_nt in our spell rotation


t=0;
swp=vt=swd=mb=ve= 100000; // We set the initial values high, to represent no spell is on cooldown.
n_vt= n_swp= n_mb= n_ve= n_swd= n_mf= 0;
printf("Enter your spell haste rating\n");
scanf("%d",&spellhasterating);
spellhaste = spellhasterating / 15;
gcd = 1500 - spellhaste;
printf("How long is the combat? (In miliseconds)\n");
scanf("%d",&tmax);

while (t<=tmax);
{

if (vt>= 15000) // Casts VT if not on cd
{
globalcd(vt, swp, swd, mb, ve, t, gcd);
vt = 0;
n_vt ++;
}
else if (swp>=24000) // Casts SW: P
{
globalcd(vt, swp, swd, mb, ve, t, gcd);
swp = 0;
n_swp ++;
}

else if (ve>=60000) // Casts VE
{
globalcd(vt, swp, swd, mb, ve, t, gcd);
ve = 0;
n_ve ++;
}

else if (mb>=5500) // Casts MB
{
globalcd(vt, swp, swd, mb, ve, t, gcd);
mb = 0;
n_mb ++;
}

else if (swd>=12000) // Casts SW: D
{
globalcd(vt, swp, swd, mb, ve, t, gcd);
swd = 0;
n_swd ++;
}
else //Casts Mind Flay
{
globalcd(vt, swp, swd, mb, ve, t, gcd);
globalcd(vt, swp, swd, mb, ve, t, gcd);
n_mf ++;
}
}

totalmanaspent = (195*n_mf) + (382*n_mb) + (425*n_vt) + (52*n_ve) + (309*n_swd) + (575*n_swp);
printf("In your spell rotation; \n Number of VT casts = %d\n Number of SW: P casts = %d\n Number of MB casts = %d\n Number of SW: D casts = %d\n Number of VE casts = %d\n Number of MF casts = %d\n", n_vt,n_swp,n_mb,n_swd,n_ve,n_mf);
printf("Total mana spent = %d\n", totalmanaspent);
return 0;
}

Currently the output is this:
Enter your spellhaste rating:
286
How long is the combat (in miliseconds)?
360000

And that's it. It never gets to the loop part. Where is the problem?

Thanks all