Jump to content


Photo

Advanced Death Knight Tactics


  • Please log in to reply
125 replies to this topic

#1 Mendenbarr

Mendenbarr

    Oh, I never leave home without my party cannon.

  • • Guide Author
  • 180 posts

Posted 20 October 2012 - 12:58 AM

This post has been promoted to an article

<Midwinter> - 14/14 HM -25m - LF exceptional raiders
http://www.mwguild.net/
DK Theorycrafting/Game Dev Blog
DK Gear/log reviews/audits


#2 Jessamy

Jessamy

    Struck by Diax's Rake

  • Moderators
  • 4,803 posts

Posted 20 October 2012 - 03:41 AM

Getting obliterate to crit as often as possible involves sitting on runes. Simply don't spend them unless you are about to lose a resource, via over-capping, or tier 5 talents, or you have killing machine procs.

The oposite treatment for frost strike is to never use frost strike when killing machine is up, even if you have the runic for it. Only use frost strike when you are about to overcap runic, or you don't have killing machine.

This sounds like smart play -- managing your abilities to get the most use out of procs. However, it's in direct conflict with your first priority, not wasting resources. You list 5 resources to manage, but forgot to include a hidden 6th, time (or gcds, if it makes more sense to think of it that way to you). The more time you spend attacking, the better.

That's not to say that trying to use the correct skill when killing machine is active is wrong. Just make sure you don't do so by waiting while you could use one of your abilities. In general, it's better to push the wrong button than spend time waiting for the right button to become available. (This general rule holds for all dps classes, not just DKs.)

#3 Mendenbarr

Mendenbarr

    Oh, I never leave home without my party cannon.

  • • Guide Author
  • 180 posts

Posted 20 October 2012 - 04:57 AM

This sounds like smart play -- managing your abilities to get the most use out of procs. However, it's in direct conflict with your first priority, not wasting resources. You list 5 resources to manage, but forgot to include a hidden 6th, time (or gcds, if it makes more sense to think of it that way to you). The more time you spend attacking, the better.

That's not to say that trying to use the correct skill when killing machine is active is wrong. Just make sure you don't do so by waiting while you could use one of your abilities. In general, it's better to push the wrong button than spend time waiting for the right button to become available. (This general rule holds for all dps classes, not just DKs.)


It's most definitely more complicated than it sounds. I'm finding it rather counter-intuitive, personally. The strategy also revolves around a fact that will not be true forever, but it's true now. That fact is that we have quite a fair number of GCDs, at present gear levels, that don't get used for anything. Under the traditional priority, and for most classes, these "free gcds" happen when you have no resources left to spend. You have no runes up, you have 8 runic power, you have no rime procs, and HoW is on CD, for example. I'm trying to propose moving those free gcds to in-between frost strikes, instead of at the end of a big spam, or to when you have KM procced but not enough runes.

From a logical perspective, if all 5 resources listed are not capped, you are generating attacks at 100% of the possible amount of attacks generated. It doesn't matter if those resources are used as soon as they are up, or 3 seconds later. This obviously changes if you're about to hit a stop dps phase, about to switch targets, about to kill the target, and for a multitude of other reasons. For most boss fights however, this is not the case.

Perhaps I should make a section on managing GCDs, as you are quire correct it's one more resource to manage. If you hold off on anything, and can't spend resources fast enough in the GCDs you have, that counts as over-capping. It takes a bit of getting used to, but I'm personally noticing a definitive improvement over the traditional spam tactics.


Edit:
Ignoring KM procs being overridden, and proper management of resources to avoid wasting any, I can't see any way this would reduce overall damage at all. KM procs being overridden is the biggest concern I have with this play-style, but considering that using 2 KM procs on a FS is about equal to 1 proc on an Obliterate, I'm still confident it's a damage increase. I'm not, however, confident enough to write a sim around this play-style, or adapt an existing one. Logic feels right, but I'm hoping for feedback on the subject.

#4 xbit

xbit

    Von Kaiser

  • Members
  • 54 posts

Posted 20 October 2012 - 05:33 AM

From a logical perspective, if all 5 resources listed are not capped, you are generating attacks at 100% of the possible amount of attacks generated.


True

It doesn't matter if those resources are used as soon as they are up, or 3 seconds later.


Not necessarily. You're forgetting to consider potential future resource caps. If you wait too long, even if you never waste any resource during that waiting, a sequence of good RNG can cause you to generate more resources than you are able to spend, causing waste. As an example, consider a case where you have 70 RP, KM is up, you have no capped runes, and you can Obliterate in 3 seconds (at which point you still won't be wasting any regen). Based on what you said, you should hold off and wait for that Obliterate to eat the KM. Assume even further than your KM isn't overwritten in that window.

But, what happens when your runes are in the state where you have 1F 1D fully depleted, 1U is up the other FU pair has 3 seconds remaining on regen, and the other D has 4 seconds remaining. You wait for 3 seconds, Obliterate. Rime procs. You use Rime. KM procs (and the D rune just came off CD so you can Obliterate). Except now your RP is at 94 so you can either Obliterate the KM and waste 18 RP or FS and waste the KM.

If you had used FS on the KM immediately you wasted the KM, but could then have Obliterated the next KM and not lost the 18 RP in the process. Net gain. Furthermore, the early FS could have procced a rune which could have caused another Obliterate and potentially another Rime proc. Alternatively, RNG could have worked out such that the second KM didn't proc and you could FS as normal, in which case you would have been worse off by not waiting.

This isn't exactly easy to reason your way out of because there are a lot of corner cases whose probabilities are non-trivial to calculate. Sims indicate that waiting is almost universally a DPS loss, so I trust that the probabilities work out in favor of not waiting. If you can find a case where adding a wait to the simc action priority results in a DPS gain I would be interested in seeing it.

#5 Mendenbarr

Mendenbarr

    Oh, I never leave home without my party cannon.

  • • Guide Author
  • 180 posts

Posted 20 October 2012 - 05:32 PM

Not necessarily. You're forgetting to consider potential future resource caps. If you wait too long, even if you never waste any resource during that waiting, a sequence of good RNG can cause you to generate more resources than you are able to spend, causing waste. As an example, consider a case where you have 70 RP, KM is up, you have no capped runes, and you can Obliterate in 3 seconds (at which point you still won't be wasting any regen). Based on what you said, you should hold off and wait for that Obliterate to eat the KM. Assume even further than your KM isn't overwritten in that window.

But, what happens when your runes are in the state where you have 1F 1D fully depleted, 1U is up the other FU pair has 3 seconds remaining on regen, and the other D has 4 seconds remaining. You wait for 3 seconds, Obliterate. Rime procs. You use Rime. KM procs (and the D rune just came off CD so you can Obliterate). Except now your RP is at 94 so you can either Obliterate the KM and waste 18 RP or FS and waste the KM.

If you had used FS on the KM immediately you wasted the KM, but could then have Obliterated the next KM and not lost the 18 RP in the process. Net gain. Furthermore, the early FS could have procced a rune which could have caused another Obliterate and potentially another Rime proc. Alternatively, RNG could have worked out such that the second KM didn't proc and you could FS as normal, in which case you would have been worse off by not waiting.

This isn't exactly easy to reason your way out of because there are a lot of corner cases whose probabilities are non-trivial to calculate. Sims indicate that waiting is almost universally a DPS loss, so I trust that the probabilities work out in favor of not waiting. If you can find a case where adding a wait to the simc action priority results in a DPS gain I would be interested in seeing it.



Let's look at the case you provided following the old priority, and the new one.

Case new:
70 runic, 1F 1D fully depleted, 1U is up the other FU pair has 3 seconds remaining on regen, and the other D has 4 seconds remaining, KM up. You wait 3 seconds, instead of using FS. 1 obliterate with KM. Rime procs, and you suggest using rime, however, I'd say hold off on the rime proc until before you need to obliterate again. Since KM in not up the second after the first obliterate, you can safely tap FS for a non-crit. We're assuming KM procs at this point, so now you use rime, and then obliterate. You hold off on a GCD where you could be using obliterate, because you used FS before the obliterate to use up the proc, but you don't overcap runic. You start at 70, jump up to 94, down to 74, and up to 98. At this point, you are dangerously close to capping runic, so you want to tap out as many FSs as possible, waiting for that next proc. 90% of the time, you can get off 2-4 FS or more before that happens.


Total: 2 KM obliterate, 1 FS, 1 Rime, no over-capped resources, no wasted resources.


Case you:
But we're not done yet! Let's take the same case as above, but use rime when you suggest, leaving us with 94 runic, the runes for an obliterate, and a KM proc. The solution is simple. Use frost strike! You do end up wasting the KM proc, but you waste the second one, and not the first one. Admittedly, this isn't better than the traditional spam strategy, however, you break even. I've yet to find a case where you can't break even by playing perfectly, and many many cases where you can go higher than even. I'm sure there is a case where this is a loss, but it's rare enough to hardly ever show up.

Total: 1 KM obliterate (first one) 1 obliterate. 1 KM FS, 1 Rime, no over-capped resources, no wasted resources.

Case traditional:
You'd use the first KM proc on the FS, since you have the runic for it now. Then, on the second one, you would use obliterate.

Total: 1 KM obliterate (second one) 1 obliterate. 1 KM FS, 1 Rime, no over-capped resources, no wasted resources.

#6 Dossou

Dossou

    Glass Joe

  • Members
  • 11 posts

Posted 30 October 2012 - 02:09 AM

The way I play is essentially not ever waiting for anything with 1-2 exceptions.

1. Obliterate with KM up before end of next swing/GCD
2. If not waiting causes me to have a Frost pair or an Unholy pair not on CD

In my opinion waiting on Obliterates isn't really worth it unless one of the above conditions is the case.

Mendenbarr's statement about never using Killing Machine Frost Strike isn't really correct at all. If you're sitting on RP with no runes up and KM procs, you're wasting more potential procs by not using Frost Strike. My first condition applies this rule. I monitor my swing timer when I have a KM proc stored but no runes to Obliterate. If OB will be available before my next swing, I save the KM for OB. If not, I use Frost Strike. In my gear, my swing timer is ~3 seconds unbuffed, so it's not very difficult to track and hold/use KMs appropriately.

I also disagree with you when you say don't Obliterate with Rime up. The same tracking your swing timer applies here. If KM is procced and your next swing is about to land, it's best to use that KM with Obliterate even if you have Rime procced to avoid wasting a potential 2nd KM proc. However if you can safely use 2 Globals before your swing hits, Rime and then KM Obliterate is the way to go.

What it all boils down to is optimizing your usage of KM procs and ensuring you don't waste any.
When I got a swing timer and starting using to store/use KM procs for Obliterate my DPS increased by a noticeable margin.

Of course the practicality of monitoring a swing timer with the high speed of 2H frost with as the expansion goes on with more and more haste becomes increasingly difficult, but right now it's pretty reasonable.

P.S. Mendenbarr - Epeen Bot - RaidBots - Web Tools for WoW Nerds

#7 Mendenbarr

Mendenbarr

    Oh, I never leave home without my party cannon.

  • • Guide Author
  • 180 posts

Posted 02 November 2012 - 06:38 PM

The way I play is essentially not ever waiting for anything with 1-2 exceptions.

1. Obliterate with KM up before end of next swing/GCD
2. If not waiting causes me to have a Frost pair or an Unholy pair not on CD

In my opinion waiting on Obliterates isn't really worth it unless one of the above conditions is the case.

Mendenbarr's statement about never using Killing Machine Frost Strike isn't really correct at all. If you're sitting on RP with no runes up and KM procs, you're wasting more potential procs by not using Frost Strike. My first condition applies this rule. I monitor my swing timer when I have a KM proc stored but no runes to Obliterate. If OB will be available before my next swing, I save the KM for OB. If not, I use Frost Strike. In my gear, my swing timer is ~3 seconds unbuffed, so it's not very difficult to track and hold/use KMs appropriately.

I also disagree with you when you say don't Obliterate with Rime up. The same tracking your swing timer applies here. If KM is procced and your next swing is about to land, it's best to use that KM with Obliterate even if you have Rime procced to avoid wasting a potential 2nd KM proc. However if you can safely use 2 Globals before your swing hits, Rime and then KM Obliterate is the way to go.

What it all boils down to is optimizing your usage of KM procs and ensuring you don't waste any.
When I got a swing timer and starting using to store/use KM procs for Obliterate my DPS increased by a noticeable margin.

Of course the practicality of monitoring a swing timer with the high speed of 2H frost with as the expansion goes on with more and more haste becomes increasingly difficult, but right now it's pretty reasonable.

P.S. Mendenbarr - Epeen Bot - RaidBots - Web Tools for WoW Nerds


I can definitely agree that maximizing damage is all about proper usage of procs. You play-style seems to involve not wasting a single KM proc, by using them as soon as you get them. However, would you not agree that, when comparing the damage increase of a km proc used on a frost strike, compared to a km proc used on an obliterate, would you not agree that the proc is "half" wasted?

By choosing to not wait for obliterate, you surely are using more KM procs per minute, but is it enough to make up for the fact that some procs are only doing half the damage they could be doing?

In regards to your criticism on rime, there may be a legitimate concern there, so let's look at some math. Assuming you are not GCD capped, you have rime procced, let's say you have both death's recharging, and one frost and one unholy up, and KM just procced. Following my rules, you would use rime first, and that would take a full GCD, during which there is a chance you proc KM again, overwriting the proc.

The average amount of procs wasted is (KM PPM)/60 = KM PPS (procs per second)
KM PPS * value of KM proc = loss

Whereas, following your suggestion of using the obliterate before the rime:
Rime proc chance * value of rime = loss


Now we can delve into messy numbers.
Sub in your own numbers if you want to.

KM is 6 PPM before haste, and while I'm unable to check at the moment, being with no power, I believe 2hand frost has around 40% haste, with 6k haste rating. That means that you'll get 6 *1.4 PPM, or 8.4 PPM. 8.4/60 gives us PPS, which happens to be 0.14 PPS. Assuming you manage to use 60% of procs on obliterate, and 40% on FS, and that oblit hits for 60k noncrit and 120k crit, with no major str cds, and that FS hits for 30k noncrit and 60k crit, .4*30 + .6*60 = 48k, being the average value of a KM proc.

0.14 * 48k = 6.72k dmg being the loss of using rime.

.45 * 50k (HB's dmg) = 22.5k dmg being the loss of not using rime

I'll update the numbers to be more accurate when I get power back, but the difference seems big enough that slight alterations won't make much of a difference.


While I appreciate the criticism of my proposed frost tactics, I'd appreciate it even more if you had some math or at least in-depth reasoning to back up your criticisms.

P.S. Extenuating circumstances, which I would love to go over in detail sometime as we're on the same server, prevent me from personally being able to show the fruits of my labor so early in a tier. These circumstances, in fact, are preventing me from properly testing this method of frost tactics, for lack of a better name, at the level it deserves to be tested. So I feel the need to ask, based of the point of this thread, did you even try anything in it?

#8 Dossou

Dossou

    Glass Joe

  • Members
  • 11 posts

Posted 02 November 2012 - 09:12 PM

I can definitely agree that maximizing damage is all about proper usage of procs. You play-style seems to involve not wasting a single KM proc, by using them as soon as you get them. However, would you not agree that, when comparing the damage increase of a km proc used on a frost strike, compared to a km proc used on an obliterate, would you not agree that the proc is "half" wasted?

By choosing to not wait for obliterate, you surely are using more KM procs per minute, but is it enough to make up for the fact that some procs are only doing half the damage they could be doing?

In regards to your criticism on rime, there may be a legitimate concern there, so let's look at some math. Assuming you are not GCD capped, you have rime procced, let's say you have both death's recharging, and one frost and one unholy up, and KM just procced. Following my rules, you would use rime first, and that would take a full GCD, during which there is a chance you proc KM again, overwriting the proc.

The average amount of procs wasted is (KM PPM)/60 = KM PPS (procs per second)
KM PPS * value of KM proc = loss

Whereas, following your suggestion of using the obliterate before the rime:
Rime proc chance * value of rime = loss


Now we can delve into messy numbers.
Sub in your own numbers if you want to.

KM is 6 PPM before haste, and while I'm unable to check at the moment, being with no power, I believe 2hand frost has around 40% haste, with 6k haste rating. That means that you'll get 6 *1.4 PPM, or 8.4 PPM. 8.4/60 gives us PPS, which happens to be 0.14 PPS. Assuming you manage to use 60% of procs on obliterate, and 40% on FS, and that oblit hits for 60k noncrit and 120k crit, with no major str cds, and that FS hits for 30k noncrit and 60k crit, .4*30 + .6*60 = 48k, being the average value of a KM proc.

0.14 * 48k = 6.72k dmg being the loss of using rime.

.45 * 50k (HB's dmg) = 22.5k dmg being the loss of not using rime

I'll update the numbers to be more accurate when I get power back, but the difference seems big enough that slight alterations won't make much of a difference.


While I appreciate the criticism of my proposed frost tactics, I'd appreciate it even more if you had some math or at least in-depth reasoning to back up your criticisms.

P.S. Extenuating circumstances, which I would love to go over in detail sometime as we're on the same server, prevent me from personally being able to show the fruits of my labor so early in a tier. These circumstances, in fact, are preventing me from properly testing this method of frost tactics, for lack of a better name, at the level it deserves to be tested. So I feel the need to ask, based of the point of this thread, did you even try anything in it?


I don't think I need math to back up the fact that I play the way I state and consistently push Top 10 Ranks. I did 107k on Gara'jal going down twice (I didn't log myself though) and 86k on Spirit Kings. Gara'jal would have been Rank 1 and Spirit Kings was Rank 1. Like I said, maximizing your KM procs is a DPS increase over delaying for Obliterates unless you're not autoswinging in between. I really can't even monitor autoswings that accurately during Spirit Kings because there's actually mechanics.

Your PPM on KM is pretty inaccurate accounting for many DKs having extremely varying levels of haste. For example, someone without a Heroic Weapon is still probably gemming str over haste. I had 8k haste before I got Tier which dropped me down to 7k, still well above your estimated 6k. This leads to more autoswings, more KM procs, and therefore more Frost Strike crits, therefore more damage.

Additionally, the fact that you are assuming waiting for Obliterates with KM procced, why are you assuming only 60% accuracy of this in your math?

Using Math to value doing certain things that rely on KM procs to be accurate have little to no meaning due to the fact that Frost is so RNG based. Running 50k iterations with my gear simmed, the difference between my lowest and highest sims was 26k. TWENTY-SIX. That's an insane amount of RNG over a 6-7 minute fight.

When you ask if I've even tried anything in this thread, the answer is yes, I have tried most everything because there isn't anything advanced about it. It's just basic frost raiding information. I'm trying hard not to knock on you but there really isn't any new information in this post.

#9 Mendenbarr

Mendenbarr

    Oh, I never leave home without my party cannon.

  • • Guide Author
  • 180 posts

Posted 02 November 2012 - 10:47 PM

I don't think I need math to back up the fact that I play the way I state and consistently push Top 10 Ranks. I did 107k on Gara'jal going down twice (I didn't log myself though) and 86k on Spirit Kings. Gara'jal would have been Rank 1 and Spirit Kings was Rank 1. Like I said, maximizing your KM procs is a DPS increase over delaying for Obliterates unless you're not autoswinging in between. I really can't even monitor autoswings that accurately during Spirit Kings because there's actually mechanics.

Your PPM on KM is pretty inaccurate accounting for many DKs having extremely varying levels of haste. For example, someone without a Heroic Weapon is still probably gemming str over haste. I had 8k haste before I got Tier which dropped me down to 7k, still well above your estimated 6k. This leads to more autoswings, more KM procs, and therefore more Frost Strike crits, therefore more damage.

Additionally, the fact that you are assuming waiting for Obliterates with KM procced, why are you assuming only 60% accuracy of this in your math?

Using Math to value doing certain things that rely on KM procs to be accurate have little to no meaning due to the fact that Frost is so RNG based. Running 50k iterations with my gear simmed, the difference between my lowest and highest sims was 26k. TWENTY-SIX. That's an insane amount of RNG over a 6-7 minute fight.

When you ask if I've even tried anything in this thread, the answer is yes, I have tried most everything because there isn't anything advanced about it. It's just basic frost raiding information. I'm trying hard not to knock on you but there really isn't any new information in this post.


I want to clear the air and say I meant no disrespect towards you or your play-style. No one is doubting that you get very impressive results in logs.

However, unless I am mistaken, the point of elitist jerk's forums is to have a place to discuss theory-crafting of class mechanics. I'm unaware of another thread, or website, where anyone is discussing, in detail, the benefits OR problems with waiting on runes. I brought up this issue because I feel it's a damage increase, and I've presented both logic and math to support that.

Like I said, maximizing your KM procs is a DPS increase over delaying for Obliterates unless you're not autoswinging in between.


The auto-swing is a chance for KM to proc again, before you've used the current proc, over-riding it and causing a dps loss, because a KM proc becomes worth 0 (The wasted one)

Using a KM proc on FS instead of Oblit is a damage loss for that particular attack. The KM proc becomes worth (FS crit - FS hit), instead of (oblit crit - oblit hit)

Those are facts, not up for debate.

What we can argue about is which is a bigger loss.


Using Math to value doing certain things that rely on KM procs to be accurate have little to no meaning due to the fact that Frost is so RNG based. Running 50k iterations with my gear simmed, the difference between my lowest and highest sims was 26k. TWENTY-SIX. That's an insane amount of RNG over a 6-7 minute fight


While RNG may change your priority around, as well as the situations that you have to deal with, the nice thing about math is that is averaged out. The averages have meaning, and value, if we can interpret them correctly. I believe I prefaced my calculations with, the numbers are about to get messy. Luckily, I also put in all the formulas, so it should be easy for anyone else to calculate if it's a gain or a loss for their own particular haste value.

Some interesting math ~
0.14 * 48k = 6.72k dmg being the loss of using rime with 6k haste
.45 * 50k (HB's dmg) = 22.5k dmg being the loss of not using rime

In order to have enough haste to have not using rime be worth it,
x * 48k = 22.5
22.5/48 = x
x = 0.46875
0.46875 * 60 = 28.125 PPM
or 470% haste
or 178500 haste rating

We should chat sometime!

#10 Qaajn

Qaajn

    Von Kaiser

  • Members
  • 43 posts

Posted 03 November 2012 - 12:02 AM

Since you are discussing optimal proc usage, I wanted to throw out a somewhat crazy idea here. It was something that struck me with the pandaria changes to frost.

How much worse is it to actually DPS in Unholy Presence as a 2-hand Frost DK?

While this may sound completely crazy at first, think of it for a moment. You lose 20% runic power regen, and your Frost Strikes now cost 35 runic power instead. In exchange you get 10% faster rune regeneration and attack-speed. It sounds like a really bad trade, but is it that bad? Especially if used during periods of heavy haste.

Ponder this:
  • 2H-frost rely heavily on Obliterate, you get "more" of those (for simplicity ignoring less use of t5) .
  • You want to use as many KM procs as possible on Obliterate, and the increased attack-speed give you more procs.
  • You want to avoid using KM with Frost Strike. And you obviously do fewer of those.

I did some very minor tests at 85, and surprisingly the DPS weren't that much behind. I also realised that the changes had taken the fun away from frost entirely, along with never liking unholy, so now I run 2x blood and can't even try it out myself, as I have no gear even if I respec.

Though the result is most likely that it's worse, I believe the discussion and the increased understanding of how the abilities work together make it worth evaluating still. Particular interest might be if this scale linearly with gear? Is there certain conditions where using unholy aura may be better?

#11 Dossou

Dossou

    Glass Joe

  • Members
  • 11 posts

Posted 03 November 2012 - 12:55 AM

Since you are discussing optimal proc usage, I wanted to throw out a somewhat crazy idea here. It was something that struck me with the pandaria changes to frost.

How much worse is it to actually DPS in Unholy Presence as a 2-hand Frost DK?

While this may sound completely crazy at first, think of it for a moment. You lose 20% runic power regen, and your Frost Strikes now cost 35 runic power instead. In exchange you get 10% faster rune regeneration and attack-speed. It sounds like a really bad trade, but is it that bad? Especially if used during periods of heavy haste.

Ponder this:

  • 2H-frost rely heavily on Obliterate, you get "more" of those (for simplicity ignoring less use of t5) .
  • You want to use as many KM procs as possible on Obliterate, and the increased attack-speed give you more procs.
  • You want to avoid using KM with Frost Strike. And you obviously do fewer of those.

I did some very minor tests at 85, and surprisingly the DPS weren't that much behind. I also realised that the changes had taken the fun away from frost entirely, along with never liking unholy, so now I run 2x blood and can't even try it out myself, as I have no gear even if I respec.

Though the result is most likely that it's worse, I believe the discussion and the increased understanding of how the abilities work together make it worth evaluating still. Particular interest might be if this scale linearly with gear? Is there certain conditions where using unholy aura may be better?


You get 4 Frost Strikes for every 7 if you were in Frost Presence, which means less Runic Empowerment, which means less Obliterates overall.

#12 Qaajn

Qaajn

    Von Kaiser

  • Members
  • 43 posts

Posted 03 November 2012 - 02:10 AM

You get 4 Frost Strikes for every 7 if you were in Frost Presence, which means less Runic Empowerment, which means less Obliterates overall.


While it has a clear effect, it may not be as big as you think. Napkin math:

45% of the FS will return 1/2 of an obliterate. During 1 "rune recharge period" you gain 1.5 Obliterates

In unholy presence:
To get the 140 RP you need for 4 frost strikes, it takes 7 obliterates. Time taken to generate the runes needed to cast 4 FS: runespeed/UnholyAura*(7OB-4FS*45%/2)/1.5

Frost presence:
To make 7 FS you need 140/24 OB (5.8333). Time required to generate those runes: runespeed*(140/24OB-7FS*45%/2)/1.5

This show how long the rune cycles to generate 140 RP are. The relation is always the same, regardless of haste, unholy being 30.2259% longer. But since frost generates runic power 20% faster, we have to compensate for this to get the relation on rune generation. This make the final result that frost get runes 8.5216% faster then unholy.

Which is very much noticable, but perhaps not as vast as first assumed. And while you do lose out on frost strikes, you get more autoattacks for more KM, and a larger portion of those would also go to OB, due to less overall FS use. While under normal use, frost would obviously be the better option, with enough haste, you will fill out your free GCD, losing the option to decide what to KM and not cap. (OB:FS ratio would be <5:6 due to some runes not used for OB). So at one point, with enough haste, Unholy could still win out. But I think I reached the limit of my napkin math here... Perhaps it's not something normally usable, but the insight and the special-case might make it interesting enough to simulate.

#13 irongnome

irongnome

    Glass Joe

  • Members
  • 24 posts

Posted 04 November 2012 - 12:24 AM

I think there is something very basic here that we are missing.
The feedback loop.

As we are all aware, holding on to resources as a DK is a DPS loss. If you don't spend your runes, you don't generate runic power. If you don't generate runic power, you don't generate RE/RC/BT procs. If you don't generate procs, you lose out on DPS that is directly quantifiable as a percentage of your DPS equal to the amount of time you held Vs not having the talent at all.

So here is my problem. I don't have the math to quantify RE/RC/BT. It was something that literally want to make me rip my hair out. I also don't see this math accounted for in the OP's defense of the priority list detailed within. You are only comparing abilities against each other 1:1 without accounting for the opportunity cost.

I will try and take another crack at the problem in order to see what exactly that opportunity cost is but no promises. I would also try and take a stab at the problem via simulation using simcraft but I would need to compile version 505-5 in order to give you accurate results due to 505-4 (the latest public download) has a bug in it that affects frost.

#14 Mendenbarr

Mendenbarr

    Oh, I never leave home without my party cannon.

  • • Guide Author
  • 180 posts

Posted 04 November 2012 - 06:15 AM

.

As we are all aware, holding on to resources as a DK is a DPS loss. If you don't spend your runes, you don't generate runic power. If you don't generate runic power, you don't generate RE/RC/BT procs. If you don't generate procs, you lose out on DPS that is directly quantifiable as a percentage of your DPS equal to the amount of time you held Vs not having the talent at all.


While I felt the same way at first, surpringsly, this is not always the case.

Let's say you do nothing during a fight. Of the five resources listed, the only one you would be generating is runes, which would be capped, and therefore wasted. You don't generate runic unles you use abilities, you don't generate t5 procs unless you use fs, you don't get Rime procs unless you obliterate, and km doesn't proc unless you are meleeing. Spending those runes generates the runic and Rime procs, and auto-attacking and spending the runic generates km and t5 procs. We're very proc based.

Over the course of a fight, of any particular duration, we will only regenerate a certain numbers of runes. As long as you spend these runes before they get capped, you will not waste any, or generate any less than if you had used them the instant they became available. Every rune you use gives a certain number of runic, always the same a,out as long you are not runic capped. Every fs gives a chance at a t5 proc, this chance doesn't decrease if you wait a second, it only decreases if you "cap" on the way it generates, for example having no completely uncharged runes with re, or having 12 charges of bt.

Holding onto to your resources is NOT a dips loss, overcapping them it.

The math concerning the t5 profs is rather simple

100 runic = 5 fs = 5* .45 runes (using re as an example) = 2.25 runes.
100 runic always gives an average of 2.25 runes, as long as you don't cap anything.





In regards to up vs fp, I'd like to share more thoughts on that when I don't have to type everything out of an iPad, but it is worth a look.

Edit: Powers back on, heres UP vs FP.

Assuming RE for math
UP = 10% rune regen and attack speed
FP = 20% runic and FS cost 20 instead of 35

1 cycle is the time it takes to generate each rune 10 times (without unholy pres).

no pres cycle:
10 frost, 10 unholy, 10 death, 300 runic, 3.857 runes via T5, 38.571 runic
15 obliterates, 8.571 frost strikes, 1.928 oblits, 1.102 frost strikes, 7.617 rime procs, and let's say 5 HoW, for 50 more runic, which is 1.428 frost strikes,
All in all, about 17 obliterates, 11 frost strikes, 8 rime procs, and 5 HoW.
That's 41 GCDs, and I'd guess that cycle would take around 80 seconds.

frost pres cycle
10 frost, 10 unholy, 10 death, 410 runic, 9.225 runes via T5, 110.7 runic,
15 obliterates, 5 HoW, 20.5 frost strikes, 4.612 oblits, 5.535 frost strikes, 1.24 oblits, 1.5 frost strikes,
All in all, about 21 obliterates, 28 frost strikes, 9 rime procs, and 5 HoW.
That's 63 GCDs, and I'd guess that cycle would take around 80 seconds.

unholy pres cycle
11 frost, 11 unholy, 11 death, 380 runic,
16.5 oblits, 10.857 frost strikes, 2.442 oblits, 1.39 frost strikes,
Note: also contains 10% more auto-attack dmg and therefore KM procs.
All in all, about 19 obliterates, 12 frost strikes, 9 rime procs, and 5 HoW.
That's 45 GCDs, and I'd guess that cycle would take around 80 seconds.

The clear winner, with an 80 second cycle, is frost presence. But what if you have a 63 second cycle? (6.3 sec rune recharge rate) Then frost presence still wins, and is actually ideal, because you're perfectly GCD capped. However, as you'd be GCD capped, you wouldn't be able to game KM procs at all, or barley, and lucky rime/T5 procs would be wasting resources.

With frost pres, you ratio of oblits to frost strikes is .75 to 1, whereas for unholy presence, the ratio is 1.583 to 1.

Conclusion: Somewhere between a rune recharge time (not using unholy pres) of 6.3 seconds and 4.5 seconds, unholy presence becomes a dps gain over frost presence. More calculations would be needed to find exactly where, but I'm not sure where to start frankly.

#15 Qaajn

Qaajn

    Von Kaiser

  • Members
  • 43 posts

Posted 05 November 2012 - 01:07 AM

Conclusion: Somewhere between a rune recharge time (not using unholy pres) of 6.3 seconds and 4.5 seconds, unholy presence becomes a dps gain over frost presence. More calculations would be needed to find exactly where, but I'm not sure where to start frankly.


Very interesting to see some more detailed math and analysis on this!

Assuming ~40% haste from gear and bloodlust of 35%, we get a rune recharge of 5.29 seconds outside of unholy. This gets us decently into the lower half of the interval where unholy may take over from frost. With no special boss mechanics used, it seems as if this might need to be considered for use in some fights, especially once gear increases and more haste is available. I imagine it might not be so hard to simulate this if you are familiar with it. With frost being GCD capped and unholy closing approaching, I imagine you don't have to pay special attention to "timing" KM much either in the simulation. Don't forget about the increase in attack-speed in unholy, I assume it will have a decent impact on the final numbers, though the time used for switching presence need also be accounted for.

#16 Ellianne

Ellianne

    Glass Joe

  • Members
  • 3 posts

Posted 05 November 2012 - 09:57 PM

I love how this entire thread is totally based on hard statistics garnered through actual logs and simulations... oh wait...

Over every single 2h frost dk's cycle till the end of time you will average 2.25 runes. However that doesn't change the fact that in a particular snapshot of time, say 1 fight, you may get an over abundance of runes or a drought of runes.

The theoretical dps gain you get by holding onto your resources is easily wiped out by the fact you could potentially go into a rune drought thus leading to less frost strikes and overall less runes generated.

There is a reason that sims have such a GIANT variation in dps distributions. It's called RNG and it's something you need to account for.

#17 Brushmonkey

Brushmonkey

    Glass Joe

  • Members
  • 1 posts

Posted 06 November 2012 - 04:26 AM

I love how this entire thread is totally based on hard statistics garnered through actual logs and simulations... oh wait...

Over every single 2h frost dk's cycle till the end of time you will average 2.25 runes. However that doesn't change the fact that in a particular snapshot of time, say 1 fight, you may get an over abundance of runes or a drought of runes.

The theoretical dps gain you get by holding onto your resources is easily wiped out by the fact you could potentially go into a rune drought thus leading to less frost strikes and overall less runes generated.

There is a reason that sims have such a GIANT variation in dps distributions. It's called RNG and it's something you need to account for.


This is certainly the case with RE and RC, but the amount of runes gained through BT is entirely consistant, and you can choose when the runes appear. The only random resources are km and rime, and you can at least predict when rime will occur or be overwritten. There should never be an unpredictable gcd shortage.

I've found there is dps increase to be had from a degree of waiting for km procs to use on obliterate, but the more gear lvl progresses the less important it will get and the more difficult it will become to execute without stumbling over excess runes.

#18 Mendenbarr

Mendenbarr

    Oh, I never leave home without my party cannon.

  • • Guide Author
  • 180 posts

Posted 06 November 2012 - 10:36 AM

I love how this entire thread is totally based on hard statistics garnered through actual logs and simulations... oh wait...

Over every single 2h frost dk's cycle till the end of time you will average 2.25 runes. However that doesn't change the fact that in a particular snapshot of time, say 1 fight, you may get an over abundance of runes or a drought of runes.

The theoretical dps gain you get by holding onto your resources is easily wiped out by the fact you could potentially go into a rune drought thus leading to less frost strikes and overall less runes generated.

There is a reason that sims have such a GIANT variation in dps distributions. It's called RNG and it's something you need to account for.


The thread is based off logic, with minor numbers and math where needed.
And whether you get twice the runes you get on average, or half the runes, it's still logically a damage increase. RNG exists in our rotation, but doesn't change the fact that obliterate hits harder than frost strike, and it certainly does not change the fact that killing machine is better used on obliterate than on frost strike.
The entire first section on the OP is all about not wasting resources, and notes to adjust for RNG. Assuming you follow that, no rng in the world will stop waiting at the proper times to be a damage increase.

#19 irongnome

irongnome

    Glass Joe

  • Members
  • 24 posts

Posted 06 November 2012 - 02:47 PM

I think the entirety of this thread can be summed up as follows (no offense intended, just for clarity's sake):

Get a swing timer addon.
Hold off for a few milliseconds at the end of a white swing and NOT spam frost strike.
Game KM procs as much as possible on Oblits in that manner.
Otherwise just follow the same priority list in the main frost thread (ie. not capping runes or runic power)

If i missed something, please expand upon it.

#20 Italiandk

Italiandk

    Glass Joe

  • Members
  • 11 posts

Posted 06 November 2012 - 03:25 PM

I think the entirety of this thread can be summed up as follows (no offense intended, just for clarity's sake):

Get a swing timer addon.
Hold off for a few milliseconds at the end of a white swing and NOT spam frost strike.
Game KM procs as much as possible on Oblits in that manner.
Otherwise just follow the same priority list in the main frost thread (ie. not capping runes or runic power)

If i missed something, please expand upon it.


I'm tracking Swing Timer with Weak Auras and it's actually more than "hold off a few milliseconds" to see if KM occurs or not. It's actually like 0,5 second (I waited "few milliseconds" after the swing occurred, I saw no KM procced, I OBed and KM procced).




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users