3.1 PTR - Page 11

Promethia · 03/01/09, 5:33 AM

Originally Posted by Thistlebee

You bring up a great point with this and I would love to see some numbers on haste vs crit as far as mana regen in concerned.

Originally Posted by Havoc12

I think people should rexamine the uptime formula very carefully

$1 - (1-C)^{n}$

Both n and C have diminishing returns, but the diminishing returns applies to both values.

If we redefine n to be the number of (crit capable) casts in the last 8 seconds prior to any haste, then we could use

$HC_{uptime} = 1 - (1-C)^{n(1 + H)}$

where C is the crit percentage and H is the haste percentage. That allows us to consider the effects of adding haste as well. We can then take a few (partial) derivatives to see how changing crit, haste, and n affect HC uptime. In particular:

$\displaystyle{\frac{\partial HC}{\partial C} = n(1 + H)(1-C)^{n(1 + H)-1}}$

$\displaystyle{\frac{\partial HC}{\partial H} = -n\ln(1 - C)(1-C)^{n(1 + H)}}$

$\displaystyle{\frac{\partial HC}{\partial n} = -(1 + H)\ln(1 - C)(1-C)^{n(1 + H)}}$

So the above formulas show how changing crit, haste, and n respectively affect HC uptime. Just on inspection, it is apparent that changing crit has a bigger effect on HC uptime than changing haste. The (1 - C) term is <1 so having one less of them in a product is a good thing. But more formally, if we look at the effect of haste relative to crit on HC uptime, we get:

$\displaystyle{\frac{HC_{haste}}{HC_{crit}} = -\frac{(1 - C)\ln(1 - C)}{1 + H}}$

The above expression will always be less than 1, although that's maybe not obvious:

$\displaystyle{\text{Let }R = \frac{1}{1 - C}\text{, then}}$

$\displaystyle{\frac{HC_{haste}}{HC_{crit}} = \frac{\ln(R)}{R}\frac{1}{1 + H}}$

Since ln R is strictly less than R for all R, ln(R) / R is less than 1. Similarly, 1 / (1 + H) is always less than one, so the product of the two fractions is also less than one as well. Or if that doesn't convince you, plug a bunch of numbers into a spreadsheet and test it out.

Bottom line: haste is never better than crit for improving HC uptime.

Originally Posted by Havoc12

Well its been confirmed and its 100% proc on crit plus it also includes renew!, thus the new regen formula for spirit regen for priests is

$R= 0.005 + k'*(1-0.5*FSR)*(1+0.5*(1-(1-C)^{n}) )*Spi*\sqrt{Int}$

I was wondering where the 0.005 came from. Anyone know?

Also, since the tooltip language is a bit ambiguous, I'm wondering if anyone has verified the effects of HC on mana regen. Specifically:

1. Does it only affect mana regen from spirit (similar to meditation)? I've heard that, but has it been proven?

2. If so, does it affect regen additively or multiplicatively. For instance, while inside the five second rule, does HC work to increases the mana regen while casting to 100% (50% from meditation plus another 50%) or to 75% (multiplying the meditation bonus by 150% as reflected in the above formula)?

Bjork · 03/01/09, 7:35 AM

Originally Posted by typobox

I mathed Serendipity out here, although I did forget to account for GCD capping on the FHx3 rotation. Short story is that FHx2->GH is definitely a large improvement over straight GH spam or FH->GH, and FHx3->GH is likely a small improvement over that. (Feel free to point out any other glaring errors.)

Thanks.

So what we're looking at is - at best (and that's 100% wrong because you didn't adjust for the GCD gap) - a 12% increase in single target HPS from pure GH-spam to 3*FH+GH. That 12% increase came at a very, very high cost as we've lost our old Serendipity. What we've gained is peak-HPS roughly every 6-7 seconds - isn't that what paladins have had with Holy Shock and druids with Swiftmend all along just that these things are instant?

So now people think holypriests are amazing tankhealers because we've got a weak version of Holy Shock, we're still miles behind on pure single target power and we also have the worst mana efficency on single target healing. Hallelujah.

Only valid reason for having a priest on tankhealing is Inspiration and then you should be disc. Or you just bring a restoshaman doing the same job and you have Mana Tide.

Don't get me wrong, new Serendipity is amazing, it makes holypreists extremly good healers and it's a fun mechanic, but we're still gimp tankhealers compared to any other healer.

maldran · 03/01/09, 8:15 AM

Originally Posted by Promethia

Also, since the tooltip language is a bit ambiguous, I'm wondering if anyone has verified the effects of HC on mana regen. Specifically:

1. Does it only affect mana regen from spirit (similar to meditation)? I've heard that, but has it been proven?

2. If so, does it affect regen additively or multiplicatively. For instance, while inside the five second rule, does HC work to increases the mana regen while casting to 100% (50% from meditation plus another 50%) or to 75% (multiplying the meditation bonus by 150% as reflected in the above formula)?

HC gives you a 1.5 multiplier to the regen formula for 8 seconds.
I've tested it out on the PTR, my regen changes from 336(oo5s)/168(i5s) to 505(oo5s)/252(i5s) (not using any mp5 pieces).

The spirit and int values stay the same, only the effective regen out of it is changed. If you crit a second time during the HC buff its duration resets to 8 seconds. Leaving the 5second rule or not doesn't change the buff either obviously.
If I buff myself with spirit while HC is up, its effects get immediately applied to my regen. I suppose trinket procs would work similar.

MavSteele · 03/01/09, 8:29 AM

Originally Posted by Bjork

Thanks.

So what we're looking at is - at best (and that's 100% wrong because you didn't adjust for the GCD gap) - a 12% increase in single target HPS from pure GH-spam to 3*FH+GH. That 12% increase came at a very, very high cost as we've lost our old Serendipity. What we've gained is peak-HPS roughly every 6-7 seconds - isn't that what paladins have had with Holy Shock and druids with Swiftmend all along just that these things are instant?

So now people think holypriests are amazing tankhealers because we've got a weak version of Holy Shock, we're still miles behind on pure single target power and we also have the worst mana efficency on single target healing. Hallelujah.

Only valid reason for having a priest on tankhealing is Inspiration and then you should be disc. Or you just bring a restoshaman doing the same job and you have Mana Tide.

Don't get me wrong, new Serendipity is amazing, it makes holypreists extremly good healers and it's a fun mechanic, but we're still gimp tankhealers compared to any other healer.

Fair enough, but with dual specs (and depending on your raid composition) I'm not sure that it matters much. If you're going to be healing a tank full time then a holy paladin or disc priest is better suited for the job because of mana efficiency and burst healing on a short CD. Depending on how the LB "change" works out, resto druids may still be up there slightly ahead of holy priests, but we'll see on that one.

Additionally, if BH is going to continue to give two stacks of serendipity, we have the ability to trade efficiency for burst pretty easily through BH->GH combos.

Bjork · 03/01/09, 8:39 AM

Originally Posted by MavSteele

Fair enough, but with dual specs (and depending on your raid composition) I'm not sure that it matters much. If you're going to be healing a tank full time then a holy paladin or disc priest is better suited for the job because of mana efficiency and burst healing on a short CD. Depending on how the LB "change" works out, resto druids may still be up there slightly ahead of holy priests, but we'll see on that one.

Additionally, if BH is going to continue to give two stacks of serendipity, we have the ability to trade efficiency for burst pretty easily through BH->GH combos.

Our restodruid was actually slightly above holypaladins on the Patchwerk dummy last night. The dummy that doesn't do hatefuls and increase damage over time on one tank only. Holypriests not even close (neither of us had optimal specs though).

E: That was a lot of sniping going on though, as he starts hitting very weak.

Elimbras · 03/01/09, 8:49 AM

Originally Posted by Dagma

Now, this isn't a completely general model. I solved with a fixed n = 4 (2 second cast intervals), and this model assumes even cast intervals. We can't be sure that the quality of the estimate doesn't depend strongly upon n, until we solve for either a more general model or more other n values. Thankfully, that's not a hard task. But a model of uneven cast intervals would be a little trickier, perhaps.

But absent the additional modeling effort, I am at least comfortable to conduct rough theorycraft using the 1-(1-c)^n approximation. It seems pretty good, to get a sense of the uptime.

Yeh, the formula [latex]1 - (1-C)^n[\latex] is exact for the steady state, when your spell inter-time is such that it divides 8 sec. And the steady state approximation is really good for any fight where mana regen is important (this excludes shorts fights, but manapool is the main component there).

If you want to deal with uneven cast intervals, it's pretty easy.
Just use the probability to have the buff at time t (assuming t > 8s) :
$P_{HC}(t) = 1 - (1-C)^{N(t)}$ , where N(t) is the number of spells that can trigger HC in the last 8 seconds.
Then, you can compute the mean HC uptime :
$HC_{Uptime} = \frac{1}{L} \int_0^L P_{HC}(t) dt = \sum_{i=0}^8 (1 - (1-C)^i ) P(N = i)$ ,
where P(N=i) is the probability to have i spell that can triggers HC in the last 8 sec
( $P(N=i) = \frac{1}{L} \int_0^L 1_{N(t)=i} dt$ ).

So, basically, all what you need is the distribution of N(t), instead of the mean value N.
And anyway, you may have an acceptable approximation using the mean value N directly.

The difference between the exact formula and the mean-value formula is :
$diff = (1-C)^{\sum p_i * i } - \sum p_i * (1-C)^i = 1 - \sum p_i * (1-C)^{i - N_{mean}}$

Note that the function x -> a^x is convex, so the mean value of the function is less than the function taken at the mean value. So, diff is negative, and using the mean value formula is overestimation.

Worst case is when P(N=0) = 0.5, and P(N=8) = 0.5.
Then the mean value formula gives HC = 1 - (1-C)^4
The real uptime is 0.5 * (1 - (1-C) ^ 8 ).
For the following values of crit rate, we get :

Crit rate	Exact Uptime	"Mean value" uptime	difference
10 %	28.5 %	34.4 %
15 %	36.4 %	47.8 %
20 %	41.6 %	59 %
25 %	45 %	68.4 %
30 %	47.1 %	76 %
35 %	48.4 %	82.1 %

For low crit rate, the difference is not such big. For high crit rate, it's more important, but that's basically because the real HC uptime is capped by one half (it can't proc half of the time).
If you take less variance in N(t), the approximation would be quite closer.

Safiyania · 03/01/09, 1:55 PM

Having spent some time trying out Empowered Renew on my priest alt I found that my renew ticks were ticking for a lot less than I had expected them to be assuming the formula for determining healing per tick is as follows:

(Improved Renew)*(Twin Disciplines)*(Spiritual Healing)*(Glyph)*(Base+Spellpower*C*(Empowered Renew))

After quite a bit of jumping off of high places I have found that the spell formula for renew ticks is actually a better fit to the following:

(Sum of Improved Renew,Twin Disciplines, Spiritual Healing and Glyph Effects)*(Base+Spellpower*C*(Empowered Renew))

Having stripped my toon of all gear and talents I have confirmed that Base = 280. By throwing gear on but having no talents and glyphs I tentatively determined C to be approximately 0.376. Through gradual addition of the talents and the glyph of renew to the mix I found that Improved Renew, Twin Disciplines, Spiritual Healing and the glyph of renew appear to be additive rather than multiplicative in their effects. Empowered Renew is indeed a spellpower coefficient modifier as is clearly indicated by the tooltip information.

For my priest, with 2231 Spellpower, including Inner Fire but no Divine Spirit active (to compare to live) the formula, as derived from my empirical testing yields the following prediction for a renew tick:

(1+0.15+0.05+0.1+0.25)*(280+2231*0.376*1.15)=1929.26.

In PTR, observed renew ticks were 1929-1930. Upfront healing component was on the order of 1213-1214 (more on this to come)

Stripping off gear to a lower spellpower amount of 963 (an order of magnitude difference) yields:

Predicted tick: 1079.42
Observed range:1079-1080, Upfront component: 679.

With regards to the Upfront healing component of an Empowered renew it appears that it is currently being determined as follows:

Upfront Healing=0.15*(Total Amount Healed by 5 Unglyphed Renew ticks)

Again, for 2231 Spellpower, if Upfront healing were determined by the ACTUAL amount of total healing done by my glyphed renew it would be in the range of:

0.15*1929*4<=Observed Upfront Healing<=0.15*1930*4 -> 1157.4<=Observed Upfront Healing<=1158, which clearly it was not from my testing.

But what if it were calculating based on 5 ticks, not 4? Then the observed range should have been on the order of
1446.75<=Observed Upfront Healing<= 1447.5. So clearly the actual healing done by your glyphed renew ticks is not how the upfront component is being calculated.

Now, what if we remove the glyph effect from the healing? Going back to my predictive model for renew ticks I estimate an unglyphed renew tick at 2231 spellpower to be 1618 (which also appears to correspond with observed ticks for fully talented but unglyphed empowered renew). 5 ticks @ 1618 healing = 8090 Healing Done, 15% of which is 1213.5 which falls within the observed range of the Upfront healing component of Empowered Renew. At 963 Spellpower, 5 unglyphed ticks should yield ~4527 total healing done, 15% of which would be 679, again corresponding to the observed upfront healing component.

Tainter · 03/01/09, 6:45 PM

There's something funny about the Renew glyph. It's meant to increase healing per tick, but with the same total healing. Somehow the way it works is screwed up though and while it does increase the tick size the total healing is currently less than for a non-glyphed Renew.

Edit: On Live anyway.

Sinndir · 03/01/09, 7:48 PM

Originally Posted by Tainter

There's something funny about the Renew glyph. It's meant to increase healing per tick, but with the same total healing. Somehow the way it works is screwed up though and while it does increase the tick size the total healing is currently less than for a non-glyphed Renew.

Edit: On Live anyway.

How do you figure? If your renew healed for 10,000 health (ticks of 2000) then the glyph would change it to 12,500 (ticks of 3125)

[edit]
Fair enough lets try this another way (also not that I do not current have the glyph of renew, just using a hypothetical situation.

Renew that heals for 10,000 health (ticks of 2000)

Glyph changes the ticks to be 2500, but only have four of them. Still manages to be 10,000 ?

constantius · 03/01/09, 7:48 PM

Playing around with the Patchwerk test, and our tank healing (as Holy) has taken a huge hit to the regen nuts. I simply can't sustain tank healing anymore. It used to be, with careful use of IHC, I *could* spam GHeal at the same rate as paladins and their HL. On PTR (with current build), it's not even close. The paladins can sustain a mixed healing for the entire duration of 30 million HP ... I cannot.

I'd try Disc, but I'm not Glyph'd.

[e] @ Sinn: no, he's right. The math doesn't work out. The glyph is supposed to increase each tick by 25%, and reduce total ticks by 1. But the total healing goes down -- something's wrong with their implementation. You don't get (healing from a 5-tick renew) over (4 ticks). You get slightly less.

[e2] Alright, tried some more pulls. With 4-piece T7, two regen trinkets, and a druid keeping Rejuv (Revivify) up on me, I could get through a fight without sucking fumes. Much. When HC proc'd in that gear, it gave me the type of regen I enjoy on live. When it was down, I was hurting pretty bad.

Here's hoping they avoid Patchwerk-style stand-and-nuke-heal fights for Ulduar, because Holy is pretty bad at them now. Disc seems to be ok, though, so dual-spec, go.

toth · 03/01/09, 9:44 PM

I was thinking today that especially with what we're seeing on the PTR, mp5 cloth needs to go in 3.1. It's polluting the loot tables with stuff nobody wants.

Nidaba has some points in this post about mp5 vs. spirit for mana regen. My take on it is that spirit gives you almost the same static mana regen assuming 100% I5SR. In addition holy priests gain a portion of spirit as spellpower and mp5 gains no benefit under the new Holy Concentration while spirit regen does. This means that spirit allows our mana regen to scale with crit rating.

Clearly mp5 is wasted itemization for mages, warlocks and shadow priests. I believe it is also wasted itemization for holy priests. You can make an argument that mp5 is better than spirit for disc priests but even then it's pretty marginal.

I'd personally like to see Blizzard replace mp5 with spirit on future cloth drops. For me, mp5 cloth is something I grab until I get a replacement with spirit simply because nobody else wants it.

Sarkli · 03/01/09, 9:45 PM

Ok.. to talk about some of the things this new patch is going to bring forth that has either tickled the anger nerve in my brain or had a question about. All in all i really have no complaints about what they are doing to Priests. They are finally getting rid of pointless spells and talents, buffing or giving us other ones and definitely giving us more to have fun with.

Based on MMO-Champion

With how easily fear can be interupted, trinketed out of or be immune to i see nothing but a waste of effort been put into <Glyph of Psychic Scream -- Increases the duration of your Psychic Scream by 2 sec. and increases its cooldown by 8 sec. (Old: Increases the duration of your Psychic Scream by 1 sec.)> Maybe its just me.. but what id rather see is a cooldown reducer put in for this fear glyph instead of having the duration lengthened. Anyone else agree? I mean i dont ever see myself using this glyph ESPECIALLY putting an 8sec INCREASE on its CD.

Blackout: This talent has been removed. Did anyone else catch this? I had to do a double-take..why is this being removed? I havent read up on much around the forums lately about possible changes etc etc but it seems to me that this is a big mistake on Blizzards part to remove this talent. IMO its one of the biggest pains in the ass to have happen to you in a pvp setting and is awesome for a pvp Spriest to have available to him. Only thing i can think of to why they are getting rid of it is because of diminishing returns. Is it currently affected by that? If so why cant they just do to it what they're doing to a warriors charge?

PvP Trinkets will now break Shackle Undead. Question on this one. What is it used on other than a pet of sorts? DK w/ Lichborne?

cs-cam · 03/01/09, 10:17 PM

Originally Posted by Sarkli

Blackout: This talent has been removed. Did anyone else catch this? I had to do a double-take..why is this being removed? I havent read up on much around the forums lately about possible changes etc etc but it seems to me that this is a big mistake on Blizzards part to remove this talent. IMO its one of the biggest pains in the ass to have happen to you in a pvp setting and is awesome for a pvp Spriest to have available to him. Only thing i can think of to why they are getting rid of it is because of diminishing returns. Is it currently affected by that? If so why cant they just do to it what they're doing to a warriors charge?

Most RnG-based stuns have been removed due to PvP implications, it was Blackouts turn to go.

Iluminati · 03/01/09, 10:32 PM

That's true, however mages still have impact and shadow priests are much worse than mages in pvp, so there must be some other logic behind the removal. It's possible they were just shuffling talents and didnt want to put it back after the darkness change.

As for psychic scream:

"Psychic Horror: Has been redesigned and is now a 1-pt talent. You terrify the target, causing them to tremble in horror for 3 sec. and drop all weapons (disarm effect: including bows) for 10 sec. 1 minute cooldown. Instant cast. The horror effect can be dispelled, but the disarm cannot."

It's up on mmo-champion but not in their talent calc. Can anyone confirm/deny this change?

Yaltus · 03/01/09, 11:29 PM

Impact is almost certainly on the way out as well, it's just that Blizzard has to figure out something else to put there before they want to yonk it. Since they were already messing with the shadow tree, they took advantage of this chance to remove blackout.

World of Warcraft - English (NA) Forums -> Remove RNG stun = buff damage?