WotLK Healing Compendium v3.2: Same Old Thing - Page 3

Zomgdie · 04/23/09, 10:50 AM

Originally Posted by Liriel

For raid-healing you do not safe people with more SP. In most cases it is irrelevant when you safe someone if he ends up with 50% or 55% life. But it may be relevant if you heal him .2s earlier or later. So haste is a stat that can safe lifes. SP in most cases is not. Yes, haste hurts our mana but it is the best stat to safe people. Raidhealing is about saving people not getting the highest possible HPS.

This is my thought process as well. Our job as raid healers is not to drop massive gheals on that tank. We are to keep everyone else alive. Look at it like triage. Who needs a heal right this second and who can wait for the shamans chain heal or the druid hots. High amounts of crit and haste allow us to get to low targets faster and bring them to an acceptable level. I think a lot of people forget that unlike dps being competition, healing is a team sport. We all fill our niche. After you've taken care of the critical people then you can start looking at topping people up or getting serendipity charges up in preparation for the AOE burst. Where yet again it's better to get the heal off and move to the next target as fast as possible as opposed to healing for more.

I don't think any math is needed to realize that a target who gets a heal and survives at 30% hp is better off than the one who would have gotten a heal a half second to late and been at 40%.

I run with about 450-500 haste depending on a few items I swap in and out. I have no issues with throughput, rarely have mana concerns and the only people with more output than me(usually) are damn OP resto druids :P I'll agree spellpower is important, sure, but I have enough of it. Using regen centric buffs, Ie Flask of Distilled Wisdom, Cuttlesteak or Nettlefish allow me to sustain the healing for longer periods of time AND increase my crit. Who doesn't love free flash heals?(and holy concentration procs)

tedv · 04/23/09, 11:46 AM

Originally Posted by Liriel

Can you please give an argument why 125 SP should be better than 65 Int?

To be honest I don't care for Spellpower and the last time I did was MC. When was the last time someone died because your heal was 300 to small? We have a good chunk of overheal. Not every heal is an overheal but nearly every time someone is healed to full one has a big overheal. So healing for some more only will result in more overheal. It takes a long run for more Spellpower to make a difference in heals you need to cast for a given dmg.

I don't think this is the proper way to approach the question. If you are in a situation where there's a lot of overhealing, it's generally because the healing is light and you can afford to overheal. In that case, yes, spell power doesn't help. But neither does haste, crit, or regen, as you've already "solved" the healing problem. What's probably happening is that you have too many healers in the raid, and you should cut them for DPS until you are in a situation where there isn't a lot of overhealing.

If there's not a lot of overhealing though (which in Ulduar is typical a heavy AoE damage situation), then extra spell power really does save lives. When you heal for an extra 300 damage but don't top someone off, that margin can mean topping them off isn't as important as healing someone else instead.

The real impact is for group healing though. I did a quick analysis of gemming in Constantius' gear for example, and if he switched his gems from int/haste focused to spell power, and started using Frost Wyrm instead of Distilled Wisdom, here's the net change in stats:

+192 spell power
-97 Int (-1.5% crit)
-16 Spirit (-4.5 spell power)
-16 Haste (-0.5% haste)

How does this change affect a Prayer of Healing? For prayer, it's an extra 188 * 1.1 (Spiritual Healing) * 1.1 (Divine Providence) * 1.35 (crit rate) per target. That's 307, or 1535 total, not counting heals from the glyph and ignoring overheals (for now). We need to account for the loss of 0.5% crit and haste though. With 2700-ish spell power, a typical non-crit prayer of healing hits for 22500 for the whole group. Including a 35% crit rate, this is an average 30375. The crit loss costs 0.015 * 0.5 * 22500 = 168 healing and the haste lost costs 0.005 * 30375 = 152, for a net loss of 320 healing.

That makes the total net change be 1535 - 320 = 1215 extra damage healed per prayer, or 243 per target. Even with something like 40% overhealing, this is still a clear win.

Originally Posted by MavSteele

I find this discussion is interesting, although it seems like people are pretty defensive about the subject. I think the fact that there can be such varying opinions on something as simple as stats is both a testament to the fact that healing is more an art than a science and that healing philosophy and it's ramifications can vary widely from guild to guild.

I think there are some parts of healing that are art and other parts that are science. But lets not call the science parts "art" just because some people don't have math backing up their opinions. The math is out there and it seems to disprove these intuitions. We should treat healing with as much science as possible, even if it will never be 100% science like DPS can be.

Originally Posted by Zomgdie

This is my thought process as well. Our job as raid healers is not to drop massive gheals on that tank. We are to keep everyone else alive. Look at it like triage. Who needs a heal right this second and who can wait for the shamans chain heal or the druid hots. High amounts of crit and haste allow us to get to low targets faster and bring them to an acceptable level. I think a lot of people forget that unlike dps being competition, healing is a team sport. We all fill our niche. After you've taken care of the critical people then you can start looking at topping people up or getting serendipity charges up in preparation for the AOE burst. Where yet again it's better to get the heal off and move to the next target as fast as possible as opposed to healing for more.

I don't think any math is needed to realize that a target who gets a heal and survives at 30% hp is better off than the one who would have gotten a heal a half second to late and been at 40%.

You need to look at the actual numbers though and see what you are really getting. You are arguing that extra spell power is useless because it can overheal, but haste is great because it can top people off faster. Looking at the numbers extracted from Constantius' armory, the actual loss would be 0.5% haste and the gain would be 188 spell power. That much spell power will increase the healing of a flash heal by 5% but make it cast 0.5% slower. And by 0.5% I mean 1/200th. That's 7.5 milliseconds slower. How can anyone argue that you're saving lives in the extra 7.5 milliseconds of haste, but not in the extra 120 damage healed by Flash Heal?

RootBreaker · 04/23/09, 12:02 PM

Originally Posted by tedv

The real impact is for group healing though. I did a quick analysis of gemming in Constantius' gear for example, and if he switched his gems from int/haste focused to spell power, and started using Frost Wyrm instead of Distilled Wisdom, here's the net change in stats:

+192 spell power
-32 Int (-0.5% crit)
-16 Spirit (-4.5 spell power)
-16 Haste (-0.5% haste)

How does this change affect a Prayer of Healing? For prayer, it's an extra 188 * 1.1 (Spiritual Healing) * 1.1 (Divine Providence) * 1.35 (crit rate) per target. That's 307, or 1535 total, not counting heals from the glyph and ignoring overheals (for now). We need to account for the loss of 0.5% crit and haste though. With 2700-ish spell power, a typical non-crit prayer of healing hits for 22500 for the whole group. Including a 35% crit rate, this is an average 30375. The crit loss costs 0.005 * 0.5 * 22500 = 56 healing and the haste lost costs 0.005 * 30375 = 152, for a net loss of 206 healing.

That makes the total net change be 1535 - 206 = 1329 extra damage healed per prayer, or 265 per target. Even with something like 40% overhealing, this is still a clear win.

I'm not sure how you can only lose 32 int if you're switching from distilled wisdom to frost wyrm.

Also, your math assumes that 35% crit rate is a 35% increase. Since heals only crit for +50%, a 35% crit rate is a 17.5% increase. Also, you're not including the spellpower coefficient of prayer of healing.

Besides, no one's arguing that distilled wisdom is a superior throughput flask. It's used for mana regen.

tedv · 04/23/09, 12:13 PM

Originally Posted by RootBreaker

I'm not sure how you can only lose 32 int if you're switching from distilled wisdom to frost wyrm.

Also, your math assumes that 35% crit rate is a 35% increase. Since heals only crit for +50%, a 35% crit rate is a 17.5% increase. Also, you're not including the spellpower coefficient of prayer of healing.

Besides, no one's arguing that distilled wisdom is a superior throughput flask. It's used for mana regen.

Sorry that should have been 97 int, which is 1.5% lost. Edited the post with the new numbers, but it's still heavily biased towards spell power.

I did count the "crits heal for 50% more" in the formula:

The crit loss costs 0.015 * 0.5 * 22500 = 168 healing

The 0.5 is the 50% coefficient.

The spell power coefficient of Prayer of Healing is 1.0 I thought, so it was implicit. Even if it's 3.0/3.5 though, the net change is still over 1000 extra healing from the proposed gear/consumable swap.

TheDoctor · 04/23/09, 12:32 PM

Originally Posted by tedv

-snip-
The real impact is for group healing though. I did a quick analysis of gemming in Constantius' gear for example, and if he switched his gems from int/haste focused to spell power, and started using Frost Wyrm instead of Distilled Wisdom, here's the net change in stats:

+192 spell power
-32 Int (-0.5% crit)
-16 Spirit (-4.5 spell power)
-16 Haste (-0.5% haste)

How does this change affect a Prayer of Healing? For prayer, it's an extra 188 * 1.1 (Spiritual Healing) * 1.1 (Divine Providence) * 1.35 (crit rate) per target. That's 307, or 1535 total, not counting heals from the glyph and ignoring overheals (for now). We need to account for the loss of 0.5% crit and haste though. With 2700-ish spell power, a typical non-crit prayer of healing hits for 22500 for the whole group. Including a 35% crit rate, this is an average 30375. The crit loss costs 0.005 * 0.5 * 22500 = 56 healing and the haste lost costs 0.005 * 30375 = 152, for a net loss of 206 healing.

That makes the total net change be 1535 - 206 = 1329 extra damage healed per prayer, or 265 per target. Even with something like 40% overhealing, this is still a clear win.

The the +sp case you have as 188*1.1*1.1*1.35 when I actually think you should be showing it as 192*.8057*1.1*1.1*(.65+1.5*.35) which instead yields a value of ~220.

Obviously switching from heavier regen stats to output stats will result in an output gain. Whether that is needed/desired is a completely different evaluation.

Rather than looking at absolutes you should probably look at it from the % gain/loss standpoint. So really what you are modifying is the (base+sp*coeff) portion of the healing value so looking at the comparison it should be shown that it is a (2150+2892*coeff)/(2150+2700*coeff) = 1.0357% of the original value, @2700 SP... You should notice that as SP from other sources increases the gain from a static value source like Frost Wyrm or an individual gem decreases. Then if you evaluate the haste & crit cost you would find that ((1+.5(Crit-0.015))(Haste - .005)/((1+.5Crit)(Haste)) @20% haste 30% crit = .9893% the spell power loss is less than .001%..

So +SP you gain ~3.57% throughput at the loss of ~1.07%, for a net gain of ~2.5%. In the process you lose mana regen, which should be accounted for if you want a valid comparison of the trade off considering it is obvious that the gemming/flask isn't for it's throughput it is for the regen.

RootBreaker · 04/23/09, 1:24 PM

Originally Posted by tedv

Sorry that should have been 97 int, which is 1.5% lost. Edited the post with the new numbers, but it's still heavily biased towards spell power.

I did count the "crits heal for 50% more" in the formula:

The crit loss costs 0.015 * 0.5 * 22500 = 168 healing

The 0.5 is the 50% coefficient.

The spell power coefficient of Prayer of Healing is 1.0 I thought, so it was implicit. Even if it's 3.0/3.5 though, the net change is still over 1000 extra healing from the proposed gear/consumable swap.

According to the 1st post in this thread, the PoH coefficient is 80.7%. I think that's based off 1.88 (for being a healing spell) * 3/3.5 (for cast time) * 0.5 (for being an aoe spell).

Also, that's not the only formula in your post that used crit.

How does this change affect a Prayer of Healing? For prayer, it's an extra 188 * 1.1 (Spiritual Healing) * 1.1 (Divine Providence) * 1.35 (crit rate) per target. That's 307, or 1535 total, not counting heals from the glyph and ignoring overheals (for now). We need to account for the loss of 0.5% crit and haste though. With 2700-ish spell power, a typical non-crit prayer of healing hits for 22500 for the whole group. Including a 35% crit rate, this is an average 30375.

That should be:

188 * 1.1 (Spiritual Healing) * 1.1 (Divine Providence) * 1.175 (benefit from crit rate) * 0.807 (poh coefficient)

With 2700-ish spell power, a typical non-crit prayer of healing hits for 22500 for the whole group. Including a 35% crit rate, this is an average 26437.

tedv · 04/23/09, 1:57 PM

I've been looking at Ulduar boss fights on wowwebstats.com and wowmeteronline.com for estimates of general mana usage and efficiency. Here are some rough data points to provide context:

* For a typical 5 minute fight, one third of the mana pool comes from the base mana and one third is from Spirit regeneration. About 30% is replenishment and shadow fiend and other things that scale based on total mana. The other 5% is from random procs, potions, and such. This is on the order of 60k mana total.
* Increases in intellect are linearly correlated with around 63% of total available man and have a square root correlation with the 33% from spirit. In other words, if you increase your base mana pool by 10%, you'll increase your total available mana by maybe 7%.

All of this is for holy by the way. Sorry, discipline priests. (Side note: we might want to fork two separate theorycrafting threads, as the equations change quite a bit for discipline.)

So with that in mind, we can calculate the mana lost from losing 97 intellect and 16 spirit. After accounting for kings, this is 107 int and 18 spirit on the character sheet.

Intellect: Worth 1605 base mana, or 3210 throughout the fight. Also increases total spirit regen by 5%. This is another 976 mana, for 4186 total.
Spirit: Worth 390 mana in a 5 minute fight with 30% holy concentration uptime.

So the total gain is 4576 mana. That's... decent but not amazing. How much of that mana is actually spent? If you're ending fights above the 4.5k mana mark, then the intellect was useless. If you're ending exactly at 0 mana, then I understand the justification. I ran some quick numbers on how much of that extra mana you'd need to spend for it to beat spell power, and if you even end the fight with 1200 extra mana, spell power still won.

Ceralyn · 04/23/09, 2:36 PM

For your side note, there is a Discipline compendium as companion to this one. Right now I think most of us are waiting for TheDoctor's spreadsheet.

constantius · 04/23/09, 2:40 PM

I started churning through some numbers, but found some major discrepancies between mine and yours, and don't have time to finalize it now. One note that I definitely can point at where you're off is your estimate of 1/3rd base + 1/3rd spirit. If you are limiting it to *only* spirit-based regen, that's fine and it works out about correct, but if you're just basing it off "regen that comes every 5 seconds without procs", then it's more like 40% from that, and under 25% from mana pool size.

Typical holy mana pool size is ~ 23,000. At 500 Mp5 I5SR, you get a minimum of 30,000 mana back from I5SR regen over a 5 minute fight. HC procs only increase this number (by up to 6-7k).

The other thing I noticed is that you say 107 intellect (raid-buffed) is worth 4.2k mana. This is incorrect. It is 1605 base + 803 shadowfiend + 578 mana tide + 1203 (replenish) + spirit-based ~ 5.1k.

tedv · 04/23/09, 3:09 PM

Originally Posted by constantius

I started churning through some numbers, but found some major discrepancies between mine and yours, and don't have time to finalize it now. One note that I definitely can point at where you're off is your estimate of 1/3rd base + 1/3rd spirit. If you are limiting it to *only* spirit-based regen, that's fine and it works out about correct, but if you're just basing it off "regen that comes every 5 seconds without procs", then it's more like 40% from that, and under 25% from mana pool size.

Typical holy mana pool size is ~ 23,000. At 500 Mp5 I5SR, you get a minimum of 30,000 mana back from I5SR regen over a 5 minute fight. HC procs only increase this number (by up to 6-7k).

The other thing I noticed is that you say 107 intellect (raid-buffed) is worth 4.2k mana. This is incorrect. It is 1605 base + 803 shadowfiend + 578 mana tide + 1203 (replenish) + spirit-based ~ 5.1k.

It's true those calculations ignore m/5, but Holy priests don't have any of that on their gear (if they can help it). The estimates from WWS reports were:

* 1/3 Base mana
* 1/3 Spirit-based Regen
* 1/3 Replenishment, Shadow Fiend, and other stuff that scales off base mana

What you term "regen that comes every 5 seconds without procs" is really the merger of the second two categories. I separated the two categories react differently as your base mana pool increases.

My estimate was 4500 mana from 107 intellect included base, shadow fiend, replenish, and spirit altogether. I ignored mana tide because I never get it, and frankly never needed it. (We normally give it to shaman, paladins, and occasionally DPS casters.) If you include that 600 mana from mana tide, you get the 5.1k number you cite. So our estimates are in agreement there.

For the record, a typical holy mana pool is only 23k for priests intentionally stacking int. For priests stacking spell power, it's between 20k and 21k. You are estimating 37k total mana from general continual effects like replenishment and spirit. When added to a base mana pool of 20k to 23k, this is 57k to 60k total mana. Add another 3k for incidental procs (metagem, potion) and you have between 60k to 63k total mana, which is in line with my estimate of 60k total mana in a 5 minute. I think we are in agreement on this point too.

So what's the major discrepancy?

Zomgdie · 04/23/09, 3:23 PM

Originally Posted by tedv

How can anyone argue that you're saving lives in the extra 7.5 milliseconds of haste, but not in the extra 120 damage healed by Flash Heal?

It's quite simple really. That amount of time could be the difference between the heal actually hitting the target or it already being dead. Now I'll admit 7.5ms isn't a long period of time, that is true. However thats that much sooner I can start my next cast, the next one after that and the next one after that. 120 dmg healed means nothing when the person ticks to death the instant before your heal lands. Higher probability of a living target or a sneeze worth of extra dmg healed, that is the question.

RootBreaker · 04/23/09, 3:31 PM

Originally Posted by Zomgdie

Originally Posted by tedv

How can anyone argue that you're saving lives in the extra 7.5 milliseconds of haste, but not in the extra 120 damage healed by Flash Heal?

It's quite simple really. That amount of time could be the difference between the heal actually hitting the target or it already being dead. Now I'll admit 7.5ms isn't a long period of time, that is true. However thats that much sooner I can start my next cast, the next one after that and the next one after that. 120 dmg healed means nothing when the person ticks to death the instant before your heal lands. Higher probability of a living target or a sneeze worth of extra dmg healed, that is the question.

That's true for fights like Ignus where you basically only need one heal on each person to save their life from flame jets, but for something like Hodir where each player is taking 40k damage over 20 seconds, its very possible for spell power to make the difference between life or death. Obviously haste is great for Hodir too.

tedv · 04/23/09, 3:49 PM

You can make the same argument in reverse though. The vast majority of the time, that 7.5 millisecond window is NOT enough to kill the person. Perhaps that 7.5 millisecond window is only lethal when you have 188 less spell power and your previous heals hit for less. Maybe when you stack spell power, casting a heal 7.5 milliseconds later isn't as dangerous because people have been topped off better. At any rate, you are making a really dubious argument based on hypothetical situations with no probabilities attached to them. I guess I'm not convinced by that.

On the subject of haste and the impact of having 21%, I ran some more numbers on this. Note that 8% of the haste is from raid buffs, leaving 13% from gear. As a holy priest, I have actively tried to avoid haste as a stat and have unfortunately ended up with 11% on my gear. Lets assume someone has done a better job of gearing and has similar ilvl gear with only 9% haste on it. This is the difference between 17% (low haste value) and 21%.

There's the claim that you can "really feel the difference" between 17% haste and 21%. But how much of an actual difference is there in the global cooldowns between these haste values?

21% haste: 1.5 / (1 + .21) = 1.2397 seconds
17% haste: 1.5 / (1 + .17) = 1.2821 seconds

This is a net difference of 0.0424 seconds, or 42 milliseconds. The average human has a reaction time between 120ms and 370ms. A really good player who is paying attention has a reaction time of around 100ms. So the time differential on global cooldowns between 17% and 21% haste is at best a third of your physically possible reaction time.

I'm sorry, but I have to call BS on the importance of 21% haste given these numbers. Yes, you will feel it if you go as low as 10% haste. That's a 120 millisecond difference. But honestly, who only has 2% haste on their gear? You can't not get it.

To summarize:

#1) While more haste is still better, you can't really tell the difference between 21% and 17%. Therefore there is no artificial haste threshold where healing will magically get much better.

#2) It doesn't make sense to argue that people should stack haste for a meager increase in throughput while simultaneously arguing they should stack int for longevity. Either argue for stacking haste and spell power for throughput, or stack spell power and int/spirit for longevity.

constantius · 04/23/09, 5:07 PM

Originally Posted by tedv

<snip>
which is in line with my estimate of 60k total mana in a 5 minute. <snip>
So what's the major discrepancy?

My back-of-the-envelope numbers show ~ 100k mana over a 5 minute period. That's almost twice as much as you show. That's a discrepancy.

Initial mana pool: 23k
Fiend: 11k
Tide: 8k
Potion: 4k
Spirit-and-buff-based-regen (including HC): 36k [avg of 600 Mp5 assumed]
Replenishment: 17.2k
Procs: 2k (cloak enchant, metagem, trinkets)
= 101.2k mana

tedv · 04/23/09, 5:42 PM

Originally Posted by constantius

My back-of-the-envelope numbers show ~ 100k mana over a 5 minute period. That's almost twice as much as you show. That's a discrepancy.

Initial mana pool: 23k
Fiend: 11k
Tide: 8k
Potion: 4k
Spirit-and-buff-based-regen (including HC): 36k [avg of 600 Mp5 assumed]
Replenishment: 17.2k
Procs: 2k (cloak enchant, metagem, trinkets)
= 101.2k mana

I think I forgot Blessing of Wisdom / Mana Stream totem, which explains why you are getting larger mana numbers. The 17.2k number for replenishment seems very high to me. What WWS did you use for the data? Despite having two to three sources of replenishment per raid, I rarely get that much mana back. Perhaps it's just because Replenishment picks the lowest raid members on mana and that's rarely me. I suppose if I blew more mana early, then I'd get better replenishment returns, but that would be at the expense of mana to other raid members.

I'll note that the merged spirit-and-buff-based-regen total of 36k is not useful for this analysis. Part of that number is from spirit (which scales with int) and part of that is from m/5 (which doesn't scale). If we want to analyze the impact of getting more intellect, we only need look at the sources that scale. That's the base pool, fiend, replenishment, optional tide, and spirit. Everything else just adds confusion.

By the way, I think one reason you might be running low on mana and I'm not is your talent spec. You don't have any points in Test of Faith. Point for point, it's roughly as good as Spiritual Healing. You're down roughly 6% healing from that alone. Mana is great, but it's only as good as the spells you can buy. Making your heals 12% better when healing is needed most seems like a no-brainer, and I can see how lacking it would lead to a situation where you have to cast more heals (and thus spend more mana).

The second point in Surge of Light also helps mana more than you'd think, partially from free heals and partially from better OO5SR spirit regen.