Arcane: AM + IF: Optimal Latch

Mage
I didn't see it explicitly written down anywhere or directly addressed, I thought I would post on this.

Generally speaking, AM damage checks seem to be divided thusly: (1) origination or player buff/debuff calculations are applied upon cast; (2) target buff/debuff calculations are applied per tick.

This means your IF and AC stacks attach on cast to all AM ticks, but your crits, damage mitigation and even your procs, it seems, occur per tick.

So optimal DPS would be to land your casts or instants on the up-tick of IF and throw your AM just before the down-tick.

EDIT: After accounting for an unspecific lag issue, the AM damage does seem to vary with the IF ticks. So if you find yourself in a situation with an AM proc you want to maximize, then try to 'hump' it over the IF up-side.
Those are some very fine hairs you're splitting there. :)
So then what would you do if you're at full charges and missle procs?

AB and murder your mana? Sit there and enjoy the view?
That's not how channeled spells work.

Spells snapshot player stats upon cast completion. Each tick of a channeled spell can be considered a cast in this way, so each tick is going to snapshot your current IF state as it is occurs (leaves your hands).
02/26/2015 10:41 AMPosted by Aenallain
This means your IF and AC stacks attach on cast to all AM ticks, but your crits, damage mitigation and even your procs, it seems, occur per tick.

Not seeing this IF snapshotting effect. AM tick damage fluctuates as IF goes up and down.
It may be wrong, I don't have Blizzards code for it, but my data indicates the IF is snap-shotted. Pay special attention to the last tick vs the first tick damage. Also try casting at max range and on the down-tick; take note that the AM ticks reflect a 5-stack IF when they first land and even though the IF stack has dropped.
Yes, they obviously reflect a 5stack when they land. They snapshot the state at cast completion (or in this case, when the tick completes, aka leaves your hand). This means in the time it takes them to travel, you may jump to another stack state. But that tick snapshot the previous IF state when it was completed. If it snapshot all tick values on the initial cast of AM then ALL tick values would be the same. This is not the case.

It is 100% wrong.
02/26/2015 04:07 PMPosted by Frostedmages
Yes, they obviously reflect a 5stack when they land. They snapshot the state at cast completion (or in this case, when the tick completes, aka leaves your hand). This means in the time it takes them to travel, you may jump to another stack state. But that tick snapshot the previous IF state when it was completed. If it snapshot all tick values on the initial cast of AM then ALL tick values would be the same. This is not the case.

It is 100% wrong.
What numbers and data are you basing that on? I think you are just theory crafting and quoting potentially old sources.Given the statements you've made and the quickness of your response you seem to be just ad lib'ing.

And, no, they would not all automatically be the same given a steady IF (although they often appear that way); there are several other variables involved in AM damage calculation than just the IF level.

I'm more interested in Komma's opinion; I think he based it on data. There are several factors I haven't been able to remove from the equation and they could be skewing what I'm seeing.
I got it from a combatlog.

In game.

That took me all of 20 seconds to get and to verify you hadn't worked with anything in game.

http://i.imgur.com/tSIb2FA.png

The AM ticks are for different values. No trinkets / Ring / Weapon are on, so no procs. No t174pc buff. 100% Mana, had 4AC before beginning the cast, so no possible change in AC/Mastery state. (AM is mana free, so there won't be a dip in mana and then some regen after the first tick messing with results).

If AM Snapshot IF at the beginning of the cast, it would be ticking for the same value. Instead, after a second it starting ticking for more. Which would represent those ticks completing and beginning travel after IF had changed values. Which means that each tick is being snapshot individually upon completion.

http://i.imgur.com/U9BNuEa.png

Similar conditions, no IF. Note all ticks stay the same - because there is no longer IF stepping up or down in between tick completions.
02/26/2015 05:12 PMPosted by Frostedmages
I got it from a combatlog.

...

I see; you didn't understand me. I'll be clearer. I'm not interested in your opinion, Frostedmages. Is that clear enough?
  • (1) Your opinion, by your own admission, is based on 20 seconds of your in-depth research.
  • (2) it is also clear from the combat logs you posted that you aren't looking at all of the data.
  • (3) You keep editing your responses to make your answer look correct and consistent, when we both know that your actual first response was neither correct nor consistent.
  • (4) It is clear from your immediate response times that you did not do any research or even basic fact checking at the time of those responses.
  • (5) The information you keep blurbing without citation is from website summaries that everyone has already read, and not from research; so you are not actually contributing to the discussion.
  • (6) Your sole focus here seems to be to win an internet argument, and not on what is important here which is when to optimally latch AM procs with IF (and honestly that is all I care about); this fact is made obvious by you continually not posting about the topic, but about whether you are right and I am wrong about something.
  • (7) You have never showed any interest in such things previously. You seem to be more motivated by the fact that your butt-hurt over something else I posted.


These are some very good reasons not take you seriously. When put in proper context, it seems you are just Trolling. You are acting very childishly; like a nat buzzing around the actual issue annoying everyone.

...I'll tell you what. If it will make you happy, I will wait till more data is in, and we have a clearer picture of how everything is calculated and when to weave AM; then I will edit my original post and rewrite it to say whatever that is, so that everything you have posted will look foolish. Maybe that will shut you up...?
Just looking at the screenshot though...
02/26/2015 05:12 PMPosted by Frostedmages
http://i.imgur.com/U9BNuEa.png

This is pretty much the same as what I had (but didn't screenshot). AM tick damage in the same cast is fluctuating, and the only culprit is Incanter's Flow. Doesn't this mean it's not snapshotting?
02/26/2015 06:29 PMPosted by Komma
Just looking at the screenshot though...
02/26/2015 05:12 PMPosted by Frostedmages
http://i.imgur.com/U9BNuEa.png

This is pretty much the same as what I had (but didn't screenshot). AM tick damage in the same cast is fluctuating, and the only culprit is Incanter's Flow. Doesn't this mean it's not snapshotting?
It may be the case.

I can't be certain because I'm showing that the damage does track in about 50% of the cases, it stays constant in about 30% of the cases, and it goes counter in 20% of the cases.

Then there is also the fact that the log data is not showing it to be a direct correlation; it shows the damage fluctuation first then some interval later it shows the IF stack change; which doesn't make sense. Why would the damage drop before the IF down-step and rise before the IF up-step? But I can't rule out something like server lag or a log reporting accuracy. So if we adjusted the IF-ticks for something like server or logging lag, then, yes, it could be a direct correlation between the damage adjustment and the IF-tick.

Also, another issue confusing things is that the simulations I run show a slight increase in DPS using the method I described based on the whole AM latching to a single IF. But again, the DPS increase could be for other causes I haven't thought of or factored in; it doesn't directly prove how the AM latches the IF.

...I was hoping you had better data. I'm not sure anymore; there is something else going on that I haven't accounted for. The only way I can think to isolate it is to write out the different algorithmic approaches and try to run simulations against them. But, like Taliq hinted at, does it matter? Is it worth doing all that?; because now we are splitting hairs.

The bottom-line though is, do we hump AM across the max IF, or do we end-cap with AM?
I have not encountered this "lag" you mention. Are you just miinterpreting the cast time (ie. moment AM leaves your hands) versus the impact time of Arcane Missiles? Can you share the log where you think things are looking weird?

Also, what are these simulations you are running that are "based on latching to a single IF"? You seem to be confusing game mechanics mechanic with rotation optimizations. Without showing the details, your description isn't enough to tell us what's going on.

As for gaming IF with AMs, I have trouble believing that a deviation from a mana-optimized rotation is going to yield much if any gain. Incanter's Flow optimization in PVE is a scheduling problem. It's not as simple as "cast things at 20% to gain 20%". You don't stop casting in PVE, which means that for every spell you try to move from low-flow to high-flow, you're moving an equivalent amount of casting from high-flow to low-flow. Is the gain from this differential enough to make up for whatever else you're sacrificing? Whenever you deviate from an "optimal" rotation, you're costing yourself a combination of mana, procs, proc generation, uptime. As many others have concluded, the answer is almost always "no".
To verify what you said about lag, I went and did some dummy testing. As far as I can tell, there is no "lag" or "snapshotting" or any of the things you have been suggesting, even when examined down to near millisecond accuracy.

The following test was done with no weapon (no mastery enchant), no quest ring (no spellpower proc), no spellpower proc trinkets. Only Arcane Barrage, Nether Tempest and Arcane Missiles were casted. Since these are all mana positive spells, they guarantee that all Arcane Missiles are casted at or near 100% mana, which allows us to ignore all effect of mana adept.

https://www.warcraftlogs.com/reports/cTpXfyMWzDRNPhYn/#type=auras&pins=0%24Separate%24%23909049%24casts%240%240.0.0.Any%24124201848.0.0.Mage%24true%240.0.0.Any%24false%247268%5E0%24Separate%24%23909049%24damage%240%240.0.0.Any%24124201848.0.0.Mage%24true%240.0.0.Any%24false%247268&source=1&ability=156150&view=events

00:00:05.734 Komma gains Flowing Thoughts from Komma
00:00:06.659 Komma casts Arcane Missiles
00:00:07.031 Komma casts Arcane Missiles
00:00:07.114 Komma Arcane Missiles Dungeoneer's Training Dummy 2028
00:00:07.395 Komma casts Arcane Missiles
00:00:07.487 Komma Arcane Missiles Dungeoneer's Training Dummy 2099
00:00:07.760 Komma casts Arcane Missiles
00:00:07.865 Komma Arcane Missiles Dungeoneer's Training Dummy 2098
00:00:08.127 Komma casts Arcane Missiles
00:00:08.230 Komma Arcane Missiles Dungeoneer's Training Dummy 2099
00:00:08.593 Komma Arcane Missiles Dungeoneer's Training Dummy 2099
00:00:09.727 Komma's Flowing Thoughts fades from Komma

Flowing Thoughts is applied at 5.734 seconds, when a player reaches 4 stacks of Incanter's Flow. That means a 16% buff is active during 5.734-6.734 and 8.727-9.727. Other periods should have a 5 stack. Look at cast time stamps, we conclude that the first missile should hit with 16% buff, while the remaining 4 should hit with 20% buff. Looking at the damage timestamps in chronological order, that is exactly what we notice: a missile hitting for 2028 and then a whole bunch for 2098-2099. Reverse engineering the damage numbers:
2028 / 1.16 ~= 1748
This means that the base damage was 1748 per hit
1748 * 1.2 ~= 2098
This means the 5 stack hits should have landed for 2098, which is exactly what we saw.

00:00:19.736 Komma's Flowing Thoughts fades from Komma
00:00:20.966 Komma casts Arcane Missiles
00:00:21.339 Komma casts Arcane Missiles
00:00:21.440 Komma Arcane Missiles Dungeoneer's Training Dummy 1889
00:00:21.701 Komma casts Arcane Missiles
00:00:21.802 Komma Arcane Missiles Dungeoneer's Training Dummy 1889
00:00:22.077 Komma casts Arcane Missiles
00:00:22.165 Komma Arcane Missiles Dungeoneer's Training Dummy *3777*
00:00:22.441 Komma casts Arcane Missiles
00:00:22.537 Komma Arcane Missiles Dungeoneer's Training Dummy 1819
00:00:22.905 Komma Arcane Missiles Dungeoneer's Training Dummy 1819
00:00:25.735 Komma gains Flowing Thoughts from Komma

We examine another snippet from 10 seconds later. 19.736 is when Flowing Thoughts fades, which means:
19.736-20.736 is 3 stack
20.736-21.736 is 2 stack
21.736-23.736 is 1 stack
23.736-24.736 is 2 stack
24.736-25.736 is 3 stack
When we examine the timestamps of the casts, we conclude that there should be 3 hits at 2 stack, and 2 hits at 1 stack. Using earlier numbers, we can calculate the damage:
2 stack: 1748 * 1.08 ~= 1888
This closely matches the first two hits, which landed for 1889 each. The third hit is a critical strike for 3777, which also gives 1888.5 when divided by two.
1 stack: 1748 * 1.04 ~= 1818
This matches the remaining two hits, which landed for 1819 each.
02/26/2015 06:25 PMPosted by Aenallain
02/26/2015 05:12 PMPosted by Frostedmages
I got it from a combatlog.

...

...I'll tell you what. If it will make you happy, I will wait till more data is in, and we have a clearer picture of how everything is calculated and when to weave AM; then I will edit my original post and rewrite it to say whatever that is, so that everything you have posted will look foolish. Maybe that will shut you up...?


I have a feeling I'm going to be waiting a very long time.
02/26/2015 09:31 PMPosted by Komma
I have not encountered this "lag" you mention. Are you just miinterpreting the cast time (ie. moment AM leaves your hands) versus the impact time of Arcane Missiles? Can you share the log where you think things are looking weird?
No; obviously not.

02/26/2015 09:31 PMPosted by Komma
Also, what are these simulations you are running that are "based on latching to a single IF"? You seem to be confusing game mechanics mechanic with rotation optimizations. Without showing the details, your description isn't enough to tell us what's going on.
You have to be aware of the theory crafting tools available out there, right? There are several of them such as, SimulationCraft. They publish the simulated rankings for current builds, but they also have the simulation tool available for download and use; it is highly configurable, so you can even define the rotations you want to simulate. You should check it out.

02/26/2015 09:31 PMPosted by Komma
As for gaming IF with AMs, I have trouble believing that a deviation from a mana-optimized rotation is going to yield much if any gain. Incanter's Flow optimization in PVE is a scheduling problem. It's not as simple as "cast things at 20% to gain 20%". You don't stop casting in PVE, which means that for every spell you try to move from low-flow to high-flow, you're moving an equivalent amount of casting from high-flow to low-flow. Is the gain from this differential enough to make up for whatever else you're sacrificing? Whenever you deviate from an "optimal" rotation, you're costing yourself a combination of mana, procs, proc generation, uptime. As many others have concluded, the answer is almost always "no".
You seem to have this backwards. Yes, casting should be on-going. Yes, the spell casts are fungible. However, the optimal rotation is based on lining up your buff windows and timing your highest damage spells to go through them; this is precisely how you maximize DPS. It just so happens because of Mastery: Mana Adept, that your mana optimization also counts as a buff window. I don't know what you mean by "deviation from a mana-optimized rotation", mana-optimization is factored in; there is no deviation here.

02/26/2015 09:54 PMPosted by Komma
To verify what you said about lag, I went and did some dummy testing. As far as I can tell, there is no "lag" or "snapshotting" or any of the things you have been suggesting, even when examined down to near millisecond accuracy.
This makes me sad ...why would you even say something like this? I mean, you have got to know that it is impossible to detect Logging accuracy or lagging issues by looking at the effected log. You can look at it down to the nano-second if you want to ...it's like looking at one side of a coin and saying, "I only see a head; where is this tail you speak of?"

Komma, you did a good job of detailing out step-by-step two instances of an extrapolated log report. However, in a broader sense, this is exactly the same thing that Frostedmages eventually got around to trying to say he was saying; you have just said it in a way that he could not or would not with more precision and detail.

Have you ever opened up the actual WoW Client logs you are uploading and parsed through them manually? That might be a good exercise for you and it may help you understand where I am coming from.
02/27/2015 12:45 PMPosted by Aenallain
You have to be aware of the theory crafting tools available out there, right?

Ahh, that moment when you have no idea who you're talking to.
02/27/2015 12:45 PMPosted by Aenallain
You have to be aware of the theory crafting tools available out there, right?

I'm quite aware of SimulationCraft. In fact, I'm a project member, the author of all the current default action lists, and responsible for maintaining the current Mage module. The reason I ask is because you seem to be unaware of how they function and can be used.

There is no way to override snapshotting mechanics for Arcane Missiles from the user input file. This cannot be done without modifying source code. What you can change through user input is how the simulated player reacts to incanter's flow. You claim to have "done simulations using these methods", when in fact this is not possible unless you are well versed in how to implement snapshotting in SimulationCraft. This leads me to doubt your claim.

We've asked for log evidence showing this "snapshotting" behavior of Arcane Missiles. So far you have provided none, while several logs have been posted that show the contrary. All you have done so far is create misdirections with claims of your unique understanding of logging, server lag and latency. You have also shown that you don't understand how the existing Arcane rotations work in terms of mana optimizations, and why Incanter's Flow optimizations conflict with it. You do not seem to understand the cast-finish snapshotting of spell statistics. So far, you have not said anything in this thread to show that you are actually knowledgeable about existing mechanics.

I gave you the benefit of the doubt, seeing as you are from a reputable guild. Unfortunately, you have made only baseless claims and spread misinformation with no support of your claims. You are doing the community a huge disservice by acting as if you have made a discovery with thorough testing, when in fact you have done neither.

02/27/2015 12:45 PMPosted by Aenallain
Have you ever opened up the actual WoW Client logs you are uploading and parsed through them manually?

I have indeed. My question is - how about you?
Komma, your response is inappropriate. You are taking offense for some reason when you should not. You are lashing out emotionally. You are putting words in my mouth, saying I said things I did not or made claims that I did not. You are exaggerating and reading into things I did say. You are taking hyperbole to an extreme. And, sadly you seem to be arrogant to the point that you are unwilling to learn anything or admit that you have anything new to learn.

I'm sorry that I didn't know your credentials. I tried to be respectful and not talk down to you anyhow. You however are still in the dark about me, and have not shown the same courtesy.

02/27/2015 02:58 PMPosted by Komma
I gave you the benefit of the doubt, seeing as you are from a reputable guild.
Do not try and drag my guild into this. I am not the spokesman for AK. I am not an officer in AK. I do not speak on behalf of or for AK in any capacity. My opinions do not represent the official opinions of my guild or it's officers.

02/27/2015 02:58 PMPosted by Komma
You are doing the community a huge disservice by acting as if you have made a discovery with thorough testing, when in fact you have done neither.

This is an exaggerated emotional and factually incorrect statement. If you read my statements you would see that I said the simulations did not show any significant results, and that I think it actually does not matter. It is you now who have exaggerated the claim and tried to make it seem so great that its incorrectness could be a "huge disservice". Whether it bears out to be true or false, it doesn't actually change anything for anyone, because as I said its impact is not significant. ...talk about blowing things out of proportion!

Furthermore, a discussion about the mechanics of an ability is actually healthy for the community; it is your attitude of trying to trample on other people's points of view that is hurtful to the community. Why would you act like this? After rattling off your credentials, why would you then use them to stifle alternative points of view? What I have said will stand or fall on it's own merits.

02/27/2015 02:58 PMPosted by Komma
You claim to have "done simulations using these methods"

...
02/26/2015 09:01 PMPosted by Aenallain
The only way I can think to isolate it is to write out the different algorithmic approaches and try to run simulations against them.

Taken in context it is clearly implied that I have not been able to run simulations in this way, which is why I feel stuck. ...but instead of you ignore what I said, put your own words in my mouth, and then act outraged on behalf of the community. I am the one who is outraged, Komma ...and, also honestly disappointed in you.

That is just the tip of what needs to be addressed in your last post. I am not going to go through it all at this time because it would take too long and I would probably let my emotions get the better of me. But in a nutshell, your anger driven accusations of error, misdeed, or ignorance are all baseless; when taken in context that is obvious.

...I have 508MB of WoW Client log data to support what I said, plus more from other sources. I think that trumps your two cases from only WoW logs. Where would you like me to stick it? Why don't you bend over and I'll shove it up your !@#, lol ...ok, that wasn't very objective, but I am allowed one in light of your paragraphs of outburst.
Not sure what meds you are on Aenallain but you should up the dosage.

Join the Conversation

Return to Forum