You didn't read my post clearly. If you use DRW behind the target it doesn't Parry haste the target. The reason my post states you wouldn't use it if you were tanking. You would go gargoyle. If you plan on tanking a lot then its not the spec for you.
Actually for blood tanking I'd figure a 50/14/7 For toughness, annihlation and lichborne. Cause you still need 8 in frost for Imp touch and Toughness. Both gargoyle and DRM don't have any obvious tanking benefit. They contribute no threat towards you and spend RP on DC, which at least is threat.
With all the changes currently flying around and one of my raid leaders informing me I might be called on to tank as a DK (oh noes!) I've started paying attention to the Frost and Unholy trees again. Specifically looking at main tanking specs. I'll front porch the spec below with this disclaimer: All my effort has been in really perfecting my play style as a blood spec'd dps DK. I've still got a long way to go before I ever tank.
That being said how does this look for a main tank spec? I mean from my standpoint it seems to allow for excellent mitigation, ample rune uptime, good rune power regen and healthy aggro. Are the points people have been putting into Unholy (Anticipation,Epidemic) really that much better?
Anticipation really helps at the lower gear levels, while you're still trying for the defense cap. I find epidemic handy in evening out moves, so as I'm not spending time constantly refreshing both diseases on the target.
There should be a few builds people have been using lately a few pages back.
Here's just one option I quickly pulled together Talent Calculator - World of Warcraft
With all the changes currently flying around and one of my raid leaders informing me I might be called on to tank as a DK (oh noes!) I've started paying attention to the Frost and Unholy trees again. Specifically looking at main tanking specs. I'll front porch the spec below with this disclaimer: All my effort has been in really perfecting my play style as a blood spec'd dps DK. I've still got a long way to go before I ever tank.
That being said how does this look for a main tank spec? I mean from my standpoint it seems to allow for excellent mitigation, ample rune uptime, good rune power regen and healthy aggro. Are the points people have been putting into Unholy (Anticipation,Epidemic) really that much better?
Here's the spec I'd use as a main tank. Note that there's still 3 points unspent; feel free to distribute them however. Depending on raid comp, you may need to put them into Virulence for the spell hit (frost especially relies on spell hit), or you may be able to put them into Acclimation or Subversion. Other considerations are Morbidity, Two-Handed Weapon Specialization*, Butchery, and Rune Tap.
Comments on the differences between the builds:
Anticipation (as all Avoidance talents) is unaffected by the Diminishing Returns and breaks the Avoidance Caps. ANY avoidance you can get from talents is worth putting points into.
Epidemic is invaluable for tanking because it gives you a bit of extra time to recover your normal rotation; tanks are far more likely to break rotation than DPS is, since they have to deal with up to 15% boss parry PLUS they may have to burn runes on tanking cooldowns.
Scent of Blood is a HORRIBLE talent. No one should ever take it. Ever. Putting 2 points in it over Bladed Armor is a crime. Bladed Armor is an astoundingly good talent; even with cruddy premade PvP gear and not in Frost Presence, it gives almost 400 AP. In actual tank gear and +armor trinkets (which are good for us), you could easily push that to 750-800 AP with 5/5 Bladed Armor. That's a huge amount of passive threat increase for 5 talent points.
Icy Talons/Improved Icy Talons don't stack with Windfury. I'm assuming that even if you're going to run 10 mans, you're almost certainly going to have SOME kind of shaman, and WF is by far the best air totem. 6 talent points isn't worth a 5% passive haste and a maybe-4%-higher windfury totem (if your shaman isn't enhancement), not in a tank build.
Killing Machine is worthless in a tank build*. You're going to be replacing a very large portion of your auto-attacks with Rune Strikes. Runic Power Mastery is bad for the same reason: You'll NEVER be at 100+ Runic Power as a tank unless you're specifically saving it up, so the extra 30 maximum is worthless. For the same reason again, you need Chill of the Grave, because you're going to be able to use every single shred of RP you get.
No Acclimation by default because, last I heard, it doesn't stack with totems/auras. It'd be worth grabbing if you don't have regular resist auras/totems somehow, or if the supposed Rune Strike buff addition is worthwhile; you do have the points for it.
Death Rune Mastery isn't really useful for non-Blood builds. As Frost or Unholy, you don't really need or want more blood runes (which is basically what it gives you).
Spell Deflection is OK, but not worth giving up what you need to give up for it. It's too much of a specialized-yet-RNG-dependant talent.
Not taking Glacier Rot or Black Ice in a Frost build is just puzzling as all get-out. Why would you choose 20% non-stacking autoattack damage (Icy Talons) over stacking +30% Frost damage (which affects Icy Touch, Frost Fever, Howling Blast, and Frost Strike)? Glacier Rot is a very slightly lower return on investment (doesn't affect Frost Fever and is only 5% per point instead of 6%) but it's still dramatically better than a buff you're going to have anyway without spending 5 talent points on it.
Merciless Combat and Endless Winter are mostly filler. MC will help with threat at the end of fights... but usually threat's not as big a deal that late in a fight. Endless Winter does provide free counterspells, which can be very useful in certain fights and helps reinforce our roles as caster tanks, at least on the rare boss that can be counterspelled.
Hmmm.... I think that's it. Hope that helps. The best advice I can give you outside of actual talent specs is to go to Ebon Hold or a capital city and just test out Frost DPS rotations. Frost feels a LOT different from Blood. It's a lot more reactive, IMO. Don't get too set in a rotation, though, because as a MT you're going to be needing to use all the awesome short cooldown mitigation buttons they give us, all of which mess with the rotation to one degree or another.
* - this is assuming you're not going to Dual Wield as a tank. If you must DW, then Killing Machine becomes dramatically better and DW Spec becomes a pretty important talent to grab (not vital, since it only affects white damage, but important).
Last edited by Zurai : 10/10/08 at 6:45 AM.
Reason: added DW caveat
I would be interested to know what gems are going to be best in slot. This would be for both tanking as well as dps.
So for DPS...
Red = Bold (R) (+Str)
Yellow = Etched (O) (+Str/+Hit)
Blue = Sovereign (P) (+Str/+Stam)
There are also some nice +Exp/+Hit and +Str/+Crit Orange gems as well.
I've never played a tank before so I don't really have any idea for tanking gems.
I should be able to answer this question after this weekend (for DPS), assuming I don't hit any snags in my Rawr module. From what I currently understand, stat priority is something like this for an unholy DPS death knight (based on ilvl cost):
str > crit > hit > expertise > haste > armor pen
(the hit and expertise of course only being true until cap)
Formally Xyrm/Zurm, the Ret Pally. Now playing my rogue, Zyrm, more casually with RL friends.
[*]Icy Talons/Improved Icy Talons don't stack with Windfury. I'm assuming that even if you're going to run 10 mans, you're almost certainly going to have SOME kind of shaman, and WF is by far the best air totem. 6 talent points isn't worth a 5% passive haste and a maybe-4%-higher windfury totem (if your shaman isn't enhancement), not in a tank build.
WF may be the best air totem, but if you brought Improved Icy Talons, they could use another air totem, say Wrath of Air, which seems to be the only buff that provides spell haste.
In general, I find it interesting that people tend to skip over certain raid talents because they assume that someone else is going to take care of it for them. I've seen plans of enhancement shamans to skip Improved WF because, well, Deathknights bring it (not even considering that a) only Frost DKs can have it and b) even some Frost DKs don't have it). Sort of a vicious circle.
If you can be absolutely certain that there's always someone else in your raid who provides the buff, then sure, go ahead and skip it without second thought.
That's an excellent point. Something else to keep in mind about these buffs that multiple classes provide is the ease in which they can use and acquire them. For example, more warlocks would consider the 13% magic vulnerability they get as something that is a waste of a GCD, as it requires a specific effort on their part. Moonkins can apply it easily enough on a single target without any trouble, but only a deathknight can easily spread the debuff to multiple targets at once (hence, an unholy DK should probably be in every raid). For shaman, improved SoE (making it better than Horn of Winter) and imp WF (making it the same as imp icy talons) are easy to get because they only have one DPS tree and not many places to put those points, plus the cost mid-fight is lower for them (20 RP, even 10 with the glyph is still a larger impact on rotations than dropping 2-3 cheap, 1 gcd totems considering how much downtime they have between abilities).
Formally Xyrm/Zurm, the Ret Pally. Now playing my rogue, Zyrm, more casually with RL friends.
I'm pretty sure it's been discussed before, but I wonder about your choice of Icy Reach. What does it give you? I don't see its usefulness at all. KM is better filler, IMO. Maybe for chain-pulling aoe packs and maximizing aoe dmg... or the HB increased range?
You mention a couple "useless" talents but then you get Merciless Combat. You even pointed it out but I would submit that Merciless Combat has absolutely no value except for a boss/trash mob dying a little faster once he's below 35%. Threat is NOT a factor at that point. It's at least as questionable as Killing Machine (which is about the same thing with less magnitude but will proc throughout the entire fight). KM has good synergies with Guile of Gorefiend. Up to pure preference, but I wouldn't write either off completely. Once I get down to HB I usually grab I usually take Frost Aura as well, depending on how it ends up stacking.
Dark Conviction could also be a decent consideration, though I'd probably stick with other, better talents. 5% crit on everything goes a long way and it's right there for us to take in almost any tanking build.
Icy Talons, unless you desperately need the raid buff, is not optimal. Speeding up my melee swings, even with a 2H, is not something I really want to do as a tank until I get some Serious gear.
Hungering Cold really just isn't worth it, is it? Unless you'd be tanking something like Hyjal. Even then... it's pretty much a pvp talent? According to GC "Acclimation buffs Rune Strike (in addition to its normal effects)." which makes it considering worth taking in the future. Perhaps not worth what you have to give up. I like how our tank build/trees basically ended up like Warriors where there's lots of variations on a basic build.
I wouldn't say merciless combat is useless. Threat at that stage can be an issue sometimes. Case in point would be Vashj. Tank kiting a boss reduces the amount of threat they can put down and an extra amount can help. Same with a boss that has an aggro dump or reduces aggro though out the fight.
I personally doubt hungering cold is a good tanking talent. I'm sure there will be situations where it will have uses, but I don't feel it would make that much of an impact overall. Great pvp talent though.
Icy talons depends on if you have a shaman really. Even then, an elemental shaman would probably not want to drop windfury. It wouldn't be that hard to get imp icy talons for that situation. Dropping icy reach and endless winter can help with that easily.
Been trying to model BS uptime and I'm starting to doubt a few of my models, let me know what you think.
H = NPC hit chance
Chance boss attack will consume a charge = H - H^2 - H^4 - H^6 - H^8 - H^10
This is assuming the npc's attack speed is such that only the previous attack can put the present one under the charge cooldown(is that 2s or 3.5s, btw?). My basic idea is (chance to hit) - (chance that attack is under cooldown). I'm doubting this because I'm thinking some of the higher power parts of the equations are redundant and already covered by the H^2.
Anyway, once you get average chance any attack will consume a charge, you use that to find average uptime.
5/(chance to consume charge)*(attack speed) = BS uptime
Only problem with that is it's a binomial distribution. That equation would be accurate IF the average BS duration was notably shorter than the cooldown, such that the chances of the buff lasting beyond cooldown were very small and didn't affect the average much(because that's the previous equation is, an average).
Does anyone know if there's a way to calculate average uptime while taking into considering the buff clipping itself? Maybe an integration of the binomial distribution from 0 to 60? Not even sure how to accomplish that.
WF may be the best air totem, but if you brought Improved Icy Talons, they could use another air totem, say Wrath of Air, which seems to be the only buff that provides spell haste.
In general, I find it interesting that people tend to skip over certain raid talents because they assume that someone else is going to take care of it for them. I've seen plans of enhancement shamans to skip Improved WF because, well, Deathknights bring it (not even considering that a) only Frost DKs can have it and b) even some Frost DKs don't have it). Sort of a vicious circle.
If you can be absolutely certain that there's always someone else in your raid who provides the buff, then sure, go ahead and skip it without second thought.
The thing is - you pay 6 talent points to provide Windfury Totem (and 5% stacking haste to yourself). That's not worth it. As a tank, would you rather have spent 6 talent points on providing a buff that another person in the raid can almost certainly provide, or would you rather have increased security? Don't forget that the first four to five points spent do absolutely nothing for you if there's either someone with Windfury or if you yourself take Improved Icy Talons. That is NOT a wise investment of talents. There are far better things to do with 6 talent points. Also don't forget that IIT is only a 20 yard range, which means you're almost certain to miss any hunters in the raid with it, while even if WF Totem's range wasn't longer, the shamans are going to be closer to the hunters in the first place. A 20 yard range from you on large bosses like Malygos that have to be pointed away from the raid due to frontal AOE could even miss the melee.
Originally Posted by Daedalix
I'm pretty sure it's been discussed before, but I wonder about your choice of Icy Reach. What does it give you? I don't see its usefulness at all. KM is better filler, IMO. Maybe for chain-pulling aoe packs and maximizing aoe dmg... or the HB increased range?
You mention a couple "useless" talents but then you get Merciless Combat. You even pointed it out but I would submit that Merciless Combat has absolutely no value except for a boss/trash mob dying a little faster once he's below 35%. Threat is NOT a factor at that point. It's at least as questionable as Killing Machine (which is about the same thing with less magnitude but will proc throughout the entire fight). KM has good synergies with Guile of Gorefiend. Up to pure preference, but I wouldn't write either off completely. Once I get down to HB I usually grab I usually take Frost Aura as well, depending on how it ends up stacking.
Icy Reach is mainly for increased HB radius. That extra 10 yards helps a lot for AOE aggro on a frost build. Nothing sucks worse than one or two elites being just out of HB range (which means they'll be out of Death and Decay range too).
As far as Killing Machine is concerned... as a 2H tank, you're going to be replacing at the least every other auto attack with a Rune Strike; 50% avoidance is trivial for DKs, and the bosses are going to be attacking at least as fast as you are. Rune Strikes (once they're fixed to actually replace the auto attack like they're supposed to...) don't trigger KM. Further, KM is a 50% chance if your auto attacks crit. With the build I provided, your auto attack crit rate will be something like 25% WITH full raid buffs, at the most. That's a 50% chance on a 25% chance every 6-7+ seconds (at the low end; I suspect Rune Strikes will be more frequent than every other auto-attack, especially againts bosses that attack faster than the default 3.0 speed). Assuming 6 second average after parry haste and WF totem, that gives you 10 12.5% chances every minute. On average, 1.25 of those will trigger KM. 1.25 guaranteed crits a minute is already pretty poor for 5 talent points; now factor in that you've already got a 25% crit rate, so 0.31 of those 1.25 would have crit anyway. That leaves us with 0.94 extra crits per minute, assuming you don't get any KM refreshes (which can happen; it doesn't seem to have much of an internal cooldown if it has one at all, and the buff lasts 30 seconds). Even if all those crits are on Frost Strike, you only gained about 2-3k damage, or about 40 DPS. That's less than 10 DPS per talent point. Awful.
Like I added as a caveat, though, KM is hugely better for DW tanks. There's no possible way you can Rune Strike even half of your auto-attacks as a DW; you simply won't generate enough Runic Power for that. That leaves you a lot more auto-attacks per minute, closer to the order of 40 to 50 even with a very high RS rate. With the same 25% crit, you've suddenly got 40 to 50 12.5% chances, leaving you with ~6 average procs per minute. That's much, much more worthwhile.
Left up to the reader is whether DW tanking is worth it in the first place, though
As for Merciless Combat... assuming you spend only 20% of the fight under 35% of the boss's health, it's still a 2.4% DPS increase to your most powerful abilities. It's overall about a 1.5% to 2% DPS increase if only 20% of the fight is "in Execute range". That more or less fits the "1% DPS increase per point" metric for DPS talents - far better than Killing Machine does for tanks, anyway. And, besides, I flat out said it was filler. If you MUST have Killing Machine as a 2H tank, drop MC and put those two points and the three free points from that build into Killing Machine. I recommend against it almost as strongly as I recommend against Icy Talons, but it's your build.
Originally Posted by Deathwing
This is assuming the npc's attack speed is such that only the previous attack can put the present one under the charge cooldown(is that 2s or 3.5s, btw?).
It's supposed to be 3.5, but it never got changed from 2.0.
Like I added as a caveat, though, KM is hugely better for DW tanks. There's no possible way you can Rune Strike even half of your auto-attacks as a DW; you simply won't generate enough Runic Power for that. That leaves you a lot more auto-attacks per minute, closer to the order of 40 to 50 even with a very high RS rate. With the same 25% crit, you've suddenly got 40 to 50 12.5% chances, leaving you with ~6 average procs per minute. That's much, much more worthwhile.
It's supposed to be 3.5, but it never got changed from 2.0.
If you have a prot paladin in your raid, every swing will be a runestrike - we get a LOT of rp from BoS.
More usefully, here's a slab of python code for a basic uptime calculator on boneshield.
import random
debug = 0
connectchance = .5
swingtime = 1.5
proctime = 2.0
numswings = 100000
numprocs = 0
numhits = 0
swings = range(0,numswings)
results = [0]
#recast every minute.. lasts 4 procs
recast = 60.0 / swingtime
lastcast = 0
numcharges = 4
uptime = 0
for s in swings:
if numcharges > 0:
uptime += 1
if random.random() > connectchance:
results.append(0)
if debug: print ("miss!")
elif swingtime < proctime and results[s] == -.6 and numcharges > 0:
results.append(.6)
numhits += 1
numprocs += 1
if debug: print("hit, charges up, inside cd")
elif numcharges == 0:
results.append(1)
numhits += 1
if debug: print("hit! no reduction")
else:
results.append(-.6)
numhits += 1
numprocs += 1
numcharges -= 1
if debug: print("hit! removed a charge")
if lastcast + recast < s:
lastcast = s
numcharges = 4
if debug: print("recasting!")
else:
if debug: print("lastcast is %s, recast at %s, currently %s - No recast!" % (lastcast, recast, s))
print("%d procs, %d hits in %d seconds" % (numprocs, numhits, numswings*swingtime))
print("uptime: %d / %d = %s %%" % (uptime, numswings, str(100 * uptime/(1.0* numswings))))
print("net damage reduction: %s %%" % (100 * (1 - (numhits - 0.4 * numprocs)/numhits)))
Kind of sloppy code, but I didn't feel like making anything bigger than a script today.
Edit: so you don't all have to install python, here's a brief table of results:
Icy Reach is mainly for increased HB radius. That extra 10 yards helps a lot for AOE aggro on a frost build. Nothing sucks worse than one or two elites being just out of HB range (which means they'll be out of Death and Decay range too).
Zurai, what do you mean by this? Icy reach does not make the AoE portion of HB go from 10 yards to 20 yards, it simply extends the range of how far you can target something from 20 yards to 30 yards.
Zurai, what do you mean by this? Icy reach does not make the AoE portion of HB go from 10 yards to 20 yards, it simply extends the range of how far you can target something from 20 yards to 30 yards.
You know what, I never noticed that. I just assumed it worked like the mage +range talents that changed +range to +area for AOEs.
So in that case, you have 4 unspent points, and I guess one lone point in Killing Machine, much as it pains me.
You mention a couple "useless" talents but then you get Merciless Combat. You even pointed it out but I would submit that Merciless Combat has absolutely no value except for a boss/trash mob dying a little faster once he's below 35%. Threat is NOT a factor at that point. It's at least as questionable as Killing Machine (which is about the same thing with less magnitude but will proc throughout the entire fight). KM has good synergies with Guile of Gorefiend. Up to pure preference, but I wouldn't write either off completely. Once I get down to HB I usually grab I usually take Frost Aura as well, depending on how it ends up stacking.
I always favored Merciless Combat in my builds because I expect a handful of dps classes to have similar talents. To me this seemed like a no trainer to stay on par with their threat generation passively.
The thing is - you pay 6 talent points to provide Windfury Totem (and 5% stacking haste to yourself). That's not worth it.
You forgot 5% spellhaste for the raid by allowing the shaman to use Wrath of Air Totem. That's pretty significant. One could even make a case for Nature Resistance or Grounding Totem for that matter, though I admit those are very situational. Still, with only one shaman present, those 6 talent points are worth a more than what's obvious at first, not less.
With two shamans or more in the raid, one of which is an enhancement shaman, or another Frost DK, I'd agree with you that those 6 points could perhaps be better spent elsewhere. Even so, I'd probably still take the talents for 5- and 10-mans.
The range problem for hunters is valid of course. But then again, if range is a concern, two windfuries would provide more room for everyone. Drop a totem near the hunters and make use of Icy Talons in melee.
Ah, and why put 2 points in Frost Aura, if there's almost certainly a druid with Improved MotW in the raid? Ok, resistance stacking seems out of whack at present, from what I've heard (everything stacks), but still...
You forgot 5% spellhaste for the raid by allowing the shaman to use Wrath of Air Totem. That's pretty significant. One could even make a case for Nature Resistance or Grounding Totem for that matter, though I admit those are very situational. Still, with only one shaman present, those 6 talent points are worth a more than what's obvious at first, not less.
The range problem for hunters is valid of course. But then again, if range is a concern, two windfuries would provide more room for everyone. Drop a totem near the hunters and make use of Icy Talons in melee.
Ah, and why put 2 points in Frost Aura, if there's almost certainly a druid with Improved MotW in the raid? Ok, resistance stacking seems out of whack at present, from what I've heard (everything stacks), but still...
No, I don't forget 5% spellhaste to the raid. 6 talent points for 5% spell haste to the raid is still godawful.
The range problem is for more than Hunters. There are a lot of very large bosses in the game. I don't know the actual size of Malygos in the raid encounter, but when you see him when you quest in the area around The Nexus, he's easily twice the size of Nefarion - more than 20 yards wide all by himself. Only the DK would get the IIT haste at that point, which makes those points a complete and utter waste. I'm sure there are other bosses where the raid is more spread out as well; Gothik comes to mind, as do Noth, Grobulus, and the Four Horsemen. Totems have both a wider area of effect AND can be placed in the optimal position, rather than being centered on the tank (who is the last person you want to center any raid DPS buff on).
Originally Posted by Mild Confusion
Why not deathchill? Does the same thing more or less, but more controlled and reliable.
This answer and the answer to "Why Frost Aura" are the same: you must spend those points to advance down the tree, and they're the best available options. You don't get the choice between 1 point of KM vs Deathchill, or believe me, I'd take Deathchill. Similarly, you get the choice between Deathchill, KM, or Frost Aura to advance past that tier, and Frost Aura provides a greater benefit than IMOTW (not much greater, but 80 > 75). It also might save a non-Resto druid a couple talent points.
Last edited by Zurai : 10/10/08 at 4:02 PM.
Reason: fixed quote attribution
Here's the spec I'd use as a main tank. Note that there's still 3 points unspent; feel free to distribute them however. Depending on raid comp, you may need to put them into Virulence for the spell hit (frost especially relies on spell hit), or you may be able to put them into Acclimation or Subversion. Other considerations are Morbidity, Two-Handed Weapon Specialization*, Butchery, and Rune Tap.
Comments on the differences between the builds:
....
I've shortened it so as to not take up too much space.
A couple things I would add:
I found morbidity to be invaluable, especially on fights like Gothik the Harvester.
I also tend to find I have more RP than I can spend (perhaps I'm doing it wrong), so I generally skip endless winter.
I always pick up Death Chill for initial threat (essp on aoe pulls).
I would like to pick up unholy command, but I never find the talent points for it.
Ah, I see what you mean Zurai, gotta blow a point somewhere to get to the next tier of talents. 1 in killing machine would be the best option if there is a shaman dropping windfury. If not, then using icy talons as a talent dump would take priority.
If you have a prot paladin in your raid, every swing will be a runestrike - we get a LOT of rp from BoS.
More usefully, here's a slab of python code for a basic uptime calculator on boneshield.
import random
debug = 0
connectchance = .5
swingtime = 1.5
proctime = 2.0
numswings = 100000
numprocs = 0
numhits = 0
swings = range(0,numswings)
results = [0]
#recast every minute.. lasts 4 procs
recast = 60.0 / swingtime
lastcast = 0
numcharges = 4
uptime = 0
for s in swings:
if numcharges > 0:
uptime += 1
if random.random() > connectchance:
results.append(0)
if debug: print ("miss!")
elif swingtime < proctime and results[s] == -.6 and numcharges > 0:
results.append(.6)
numhits += 1
numprocs += 1
if debug: print("hit, charges up, inside cd")
elif numcharges == 0:
results.append(1)
numhits += 1
if debug: print("hit! no reduction")
else:
results.append(-.6)
numhits += 1
numprocs += 1
numcharges -= 1
if debug: print("hit! removed a charge")
if lastcast + recast < s:
lastcast = s
numcharges = 4
if debug: print("recasting!")
else:
if debug: print("lastcast is %s, recast at %s, currently %s - No recast!" % (lastcast, recast, s))
print("%d procs, %d hits in %d seconds" % (numprocs, numhits, numswings*swingtime))
print("uptime: %d / %d = %s %%" % (uptime, numswings, str(100 * uptime/(1.0* numswings))))
print("net damage reduction: %s %%" % (100 * (1 - (numhits - 0.4 * numprocs)/numhits)))
Kind of sloppy code, but I didn't feel like making anything bigger than a script today.
Edit: so you don't all have to install python, here's a brief table of results:
speed
avoidance
uptime
netreduction
1.5
50%
26.7%
10.7%
1.5
65%
35.2%
14.0%
2.0
50%
25.8%
10.3%
2.0
65%
36.8%
13.7%
2.5
50%
31.8%
12.8%
2.5
65%
45.6%
18.2%
3.3
50%
42%
16.8%
3.3
65%
58.8%
23.5%
I came over here to post my updated DK Tank Simulator, and saw this post. Considering I've just spent a ridiculous amount of time on Bone Shield equations I thought I would interject.
I'll be honest I'm not a mathematician but I do tend to manage. My original formula was based on the idea that the formula would involved a step function. The reason I thought this might be the case, is that a charge only has a chance to be removed if the mob hits them. This seemed to me to indicate a situation where it either could happen or couldn't with no partial possibilities.
Well to make a long story short I was very wrong. I later built two rough models to see what would happen and immediately noticed that the results could not be reconciled with a step function but they did help me come up with a better BS Up Time Equation.
After seeing your post I thought I would post my models in case anyone could use them or if some one had the time/knowledge to create better ones.
And on a side note, heres your values run through My Models with 4 charges. I'm getting quite different values are you figuring in the Cooldown?
speed
avoidance
model uptime
1.5
50%
35.293%
1.5
65%
44.858%
2.0
50%
33.827%
2.0
65%
45.550%
2.5
50%
41.883%
2.5
65%
56.975%
3.3
50%
56.387%
3.3
65%
73.915%
Heres the Files, the Death Wing Formulas are extraneous, I was trying to prove a different formula wasn't even close to the models. My simulators don't do percent uptime they do seconds by the way.
Been trying to model BS uptime and I'm starting to doubt a few of my models, let me know what you think.
H = NPC hit chance
Chance boss attack will consume a charge = H - H^2 - H^4 - H^6 - H^8 - H^10
This is assuming the npc's attack speed is such that only the previous attack can put the present one under the charge cooldown(is that 2s or 3.5s, btw?). My basic idea is (chance to hit) - (chance that attack is under cooldown). I'm doubting this because I'm thinking some of the higher power parts of the equations are redundant and already covered by the H^2.
Anyway, once you get average chance any attack will consume a charge, you use that to find average uptime.
5/(chance to consume charge)*(attack speed) = BS uptime
Only problem with that is it's a binomial distribution. That equation would be accurate IF the average BS duration was notably shorter than the cooldown, such that the chances of the buff lasting beyond cooldown were very small and didn't affect the average much(because that's the previous equation is, an average).
Does anyone know if there's a way to calculate average uptime while taking into considering the buff clipping itself? Maybe an integration of the binomial distribution from 0 to 60? Not even sure how to accomplish that.
The average BS up time is much easier to model than any confusing code. Consider:
Ignoring CD for the moment, let
T(N) = The expected uptime for N BS charges
A = attack speed
H = hit chance
then recursively the formula is
T(0) = 0
T(N) = H*(A + T(N-1)) + (1-H)*(A + T(N))
= HA + H*T(N-1) + A - HA + T(N) - H*T(N)
= A/H + T(N-1)
= N*A/H
Depending on whether you're glyphed or not I'll just use N for the number of charges. Considering the CD regardless of what the number is, if A > CD it doesn't matter. If A < CD but 2A > CD then exactly N-1 swings are inside the CD and have no effect, the rest are exactly as the above formula describes, so the total uptime is A*(N-1) + N*A/H. If 2A < CD but 3A > CD, then there are exactly 2*N-2 swings that are inside the CD and therefore don't effect, so the total uptime is 2*A*(N-1) + N*A/H. And so on.
This is still a little simplified since you don't always have a constant chance to be hit, and the mob doesn't always have a constant attack speed (especially considering parry haste and auto attacks + specials). But I'd say this formula is more than sufficient.
The results Janraea posted are with a 2s CD, did you change yours?
Banda,
That works, relatively. In fact, you look at the second half of my message that you quoted, that's what I'm using to figure in average uptime. Unfortunately, that results is going to be skewed higher than it should be because any specific cases where BS would have lasted more than 60s are counted inherently in the average, but are of course discarded in real life.
Another way to put it: that average equation will eventually obtain 100% uptime, but you know that's impossible.
Of course this all under the presumption that the 3.5 second CD either is or will be implemented like was stated by the developers.
@Tantlin
I was using the 2.0 cooldown, as reported in game. tweak the 'proctime' variable to use 3.5 if you desire. I only check one swing ago, so if two swings can fall into the cooldown period, my sim will give bad results. Also, I was using 4 charges, because I forgot the glyph, alter the 'numcharges = 4' line in the 'if lastcast + recast < s' conditional to fix that.
When I threw in the proctime of 3.5, my table matched yours to within 1% for every value I tried. It's likely the difference is that I only count uptime in terms of number of swings affected divided by number of swings, rather than in seconds - I'll tend to shave off the last piece of the uptime you're counting.
I am a mathematician, but this involves no math, it's just a quick and dirty sim. Any attempt to solve this problem with pure math will have to step fairly heavily into number theory, where I was never strong (and I recommend against it).
Edit: My assertion about requiring number theory was because I formulated the problem as having infinite cases - it doesn't, since swing speeds don't go below about 1.0.
@Banda
The mathematical model you pose here has one significant problem Banda - it's continuous and smooth. That matters rather a lot, since going from swing speed 1.9 to 2.1 increases uptime drastically - simulation is really the only way to calculate these numbers. In particular, 1.9 swing speed should have the same uptime as 3.8, etc.
Edit: Ah, I see the conditional branch at the bottom now. It looks reasonable to me, but I'll always trust simulations more :-)
@Deathwing Python Programming Language -- Official Website
I find the language a pleasure to work with compared to the others I've used. If you want additional resources to help with learning to use it or anything, pm me :-)
Incidentally, we cross 90% uptime at around 66% avoidance, using the 3.5 number and the glyph.
Although I have to comment the IT not stacking with IIT and shaman totems seems contradicting. There is a huge and probably somewhat unproductive thread on this same matter on the DK forums. Jist of it is haste talents/abilities of other classes stack with WF totem, and if IT stacks with WF totems but our IIT WF replacement doesn't, then a shaman becomes better, which defeats the idea of equalizing raid buffs.
Also Imp IT I think was put at a 40 yard buff some build ago.
I find the Icy Reach is good for pulls and range never hurts. But in most builds that also get IT and Imp IT don't have points for it.
And at some point Hungering Cold needs to be useful.
10/10/08, 4:56 AM
Similar Threads
