Etched Sigil Proc Mechanics (AS & ICD)

Wizard
1 2 3 5 Next
There was a 2.6.1 PTR bug with Etched Sigil's ICD that was uncovered and fixed:
09/18/2017 05:36 PMPosted by Don Vu
Hey everyone,

Thanks for all the testing. The internal cooldown for the new Etched Sigil was actually scaling with it's damage roll (don't ask), so the internal cooldown was somewhere between 1.25s and 1.5s depending on how your item rolled. We've fixed it for the next PTR patch.

(PTR Only) Fixed an issue with Etched Sigil where the internal cooldown for casting another spender was scaling with the item’s damage roll

After the bug fix, our good Wizard Dolynick tested, and confirmed the following formula below, which is given with some advice:
09/30/2017 03:04 PMPosted by dolynick
Re-tested a few breakpoints and determined the following for Disintegrate (Coef 3).

fpa = fSigil
19f = 76f
18f = 72f
14f = 70f
13f = 65f
12f = 60f
11f = 66f

Which corresponds to Sigil now having an ICD of 59f.

What this boils down to is that you want to be at an FPA that is as close to a divisor of 60 as possible. The max rate of proc is 60f (59f + 1), so FPA 20, 15, 12, 10, 6, 5, & 4 (for AT/Disintegrate) all land on a tick at 60f nicely and are optimal. The less close they are to being a divisor of 60, the slower the proc rate.

The formula for this is:
FPA*(1+ROUNDDOWN(59/FPA, 0))

Note: this formula is the same as:
10/01/2017 12:10 AMPosted by dolynick
FPA*ROUNDUP(60/FPA, 0)

Generally this means you'll want to gear at attack speeds that correspond with Frames/Tick rates that are 60 frame divisors.

Use the charts to help guide you as to what the best breakpoint for your build will be.

Breakpoint tables with corresponding fSigil rates, again courtesy of Dolynick:

Tick/Speed Coefficient 3 (Arcane Torrent, Disintegrate)
Min APS Max APS Tick/s Frames/t fSigil
4.000001 5 15 4 60
3.333334333 4 12 5 60
2.857143857 3.333333333 10 6 60
2.500001 2.857142857 8.571428571 7 63
2.222223222 2.5 7.5 8 64
2.000001 2.222222222 6.666666667 9 63
1.818182818 2 6 10 60
1.666667667 1.818181818 5.454545455 11 66
1.538462538 1.666666667 5 12 60
1.428572429 1.538461538 4.615384615 13 65
1.333334333 1.428571429 4.285714286 14 70
1.250001 1.333333333 4 15 60
1.176471588 1.25 3.75 16 64
1.111112111 1.176470588 3.529411765 17 68
1.052632579 1.111111111 3.333333333 18 72
1.000001 1.052631579 3.157894737 19 76
0.952381952 1 3 20 60
0.909091909 0.952380952 2.857142857 21 63
0.869566217 0.909090909 2.727272727 22 66
0.833334333 0.869565217 2.608695652 23 69
0.800001 0.833333333 2.5 24 72
0.769231769 0.8 2.4 25 75
0.740741741 0.769230769 2.307692308 26 78
0.714286714 0.740740741 2.222222222 27 81
0.689656172 0.714285714 2.142857143 28 84
0.666667667 0.689655172 2.068965517 29 87
0.64516229 0.666666667 2 30 60
0.625001 0.64516129 1.935483871 31 62
0.606061606 0.625 1.875 32 64
0.588236294 0.606060606 1.818181818 33 66
0.571429571 0.588235294 1.764705882 34 68
0.555556556 0.571428571 1.714285714 35 70
0.540541541 0.555555556 1.666666667 36 72
0.526316789 0.540540541 1.621621622 37 74
0.512821513 0.526315789 1.578947368 38 76
0.500001 0.512820513 1.538461538 39 78
0.487805878 0.5 1.5 40 80


Tick/Speed Coefficient 2 (Ray of Frost)
Min APS Max APS Tick/s Frames/t fSigil
4.285715286 5 10 6 60
3.750001 4.285714286 8.571428571 7 63
3.333334333 3.75 7.5 8 64
3.000001 3.333333333 6.666666667 9 63
2.727273727 3 6 10 60
2.500001 2.727272727 5.454545455 11 66
2.307693308 2.5 5 12 60
2.142858143 2.307692308 4.615384615 13 65
2.000001 2.142857143 4.285714286 14 70
1.875001 2 4 15 60
1.764706882 1.875 3.75 16 64
1.666667667 1.764705882 3.529411765 17 68
1.578948368 1.666666667 3.333333333 18 72
1.500001 1.578947368 3.157894737 19 76
1.428572429 1.5 3 20 60
1.363637364 1.428571429 2.857142857 21 63
1.304348826 1.363636364 2.727272727 22 66
1.250001 1.304347826 2.608695652 23 69
1.200001 1.25 2.5 24 72
1.153847154 1.2 2.4 25 75
1.111112111 1.153846154 2.307692308 26 78
1.071429571 1.111111111 2.222222222 27 81
1.034483759 1.071428571 2.142857143 28 84
1.000001 1.034482759 2.068965517 29 87
0.967742935 1 2 30 60
0.937501 0.967741935 1.935483871 31 62
0.909091909 0.9375 1.875 32 64
0.882353941 0.909090909 1.818181818 33 66
0.857143857 0.882352941 1.764705882 34 68
0.833334333 0.857142857 1.714285714 35 70
0.810811811 0.833333333 1.666666667 36 72
0.789474684 0.810810811 1.621621622 37 74
0.769231769 0.789473684 1.578947368 38 76
0.750001 0.769230769 1.538461538 39 78
0.731708317 0.75 1.5 40 80
0.714286714 0.731707317 1.463414634 41 82
0.697675419 0.714285714 1.428571429 42 84
0.681819182 0.697674419 1.395348837 43 86
0.666667667 0.681818182 1.363636364 44 88
0.652174913 0.666666667 1.333333333 45 90
0.638298872 0.652173913 1.304347826 46 92
0.625001 0.638297872 1.276595745 47 94
0.612245898 0.625 1.25 48 96
0.600001 0.612244898 1.224489796 49 98
0.588236294 0.6 1.2 50 100
0.576924077 0.588235294 1.176470588 51 102
0.566038736 0.576923077 1.153846154 52 104
0.555556556 0.566037736 1.132075472 53 106
0.545455545 0.555555556 1.111111111 54 108
0.535715286 0.545454545 1.090909091 55 110
0.526316789 0.535714286 1.071428571 56 112
0.517242379 0.526315789 1.052631579 57 114
0.508475576 0.517241379 1.034482759 58 116
0.500001 0.508474576 1.016949153 59 118
0.491804279 0.5 1 60 60

Thanks again Dolynick for the testing, BDF for the advice, and sVr (just cuz').

-----

Will leave the rest of my original post to preserve post history. Note that some of the below information is incorrect or inaccurate, namely that channeled skills only attempt to proc etched sigil once every so often, this is false. However, Etched Sigil was taking 20-30 extra frames to proc, and that was true during the 2.6.1 PTR.

-----

EDIT (10/1/2017): ORIGINAL POST BELOW (INFO might be false or inaccurate)

In regards to Etched Sigil, channeling skills are only attacking (and attempting to proc it) once every attack cycle. Arcane Torrent and Disintegrate are speed coefficient = 3 spells (ticking 3 times at 1 APS). These will only attempt to proc Etched Sigil once every 3 ticks.

The implications for timing your Etched Sigil properly are worth noting. For full channeling builds, you might have to account for this gap of time (several ticks' worth in duration), and possibly optimize your Attack Speed so as to limit the amount of time wasted.

EDIT: after looking at results, it looks like the 12 Frames / Tick BP is best. This BP happens to also be the easiest to gear for (with a 1.4 AS weapon, only need AS from paragon to meet it). It also happens to be the BP most builds that require this type of full-on channel behavior have been gearing towards for quite some time.

Below will be an analysis of Etched Sigil, and the effects attack speed have on proccing this item in regards to channeling.

NOTE: The scenario described below only applies to full channel (never stopping). It does not account for interweaving manual casts of your spender (such as Meteor) between your channel.

Note that EB can be cast manually without interrupting channeling, which is one of it's strengths, but for EB you also have to account for the CDR reducing your etched sigil EB procs (which is a different calculation altogether).

-----

Other Notes: Times in seconds were taken from the video clips, as such they serve only to mark the difference in time. Times are approximate. My video software was keyframing, so these are a rough estimate.

After some other charting, I figured out the slight variances for the odd numbered Frame intervals. Basically, for 11, 9, and 8 frames / tick tests, there are partial ticks that are being wasted in addition to the listed number of ticks wasted.

For example, on the 11 frames / tick listing, we would expect 22 frames wasted, but there are alternating partial ticks of 6 and 4 frames each. These get added in to the 22 frames, to give 28 and 26, respectively.

-----

Video of Testing (tested on PTR):
  • https://youtu.be/o4KXa0Y6dYU
  • PTR PATCH 2.6.1 - v2.6.1.47251

I've only tested speed co = 3, since I felt those were the most relevant. I can go back and test speed co = 2 (Ray of Frost), if needed. The results below are in relation to Speed co = 3.

20 Frames / Tick | 3 Ticks / s | 0.95239 to 1 AS
Time(s) Time Between Time Wasted Ticks wasted Frames Wasted
0.35 1.633 0.633 2 40
1.983 1.65 0.65 2 40
3.633 1.667 0.667 2 40
5.3 1.683 0.683 2 40
6.983 1.683 0.683 2 40
8.666 2 40

12 Frames / Tick | 5 Ticks / s | 1.53847 to 1.66666 AS
Time(s) Time Between Time Wasted Ticks wasted Frames Wasted
39.733 1.45 0.45 2 24
41.183 1.35 0.35 2 24
42.533 1.4 0.4 2 24
43.933 1.4 0.4 2 24
45.333 1.417 0.417 2 24
46.75 1.4 0.4 2 24
48.15 2 24

11 Frames / Tick | 5.45455 Ticks / s | 1.66667 to 1.81818 AS
Time(s) Time Between Time Wasted Ticks wasted Frames Wasted
20.35 1.416 0.416 2 28
21.766 1.467 0.467 2 26
23.233 1.483 0.483 2 28
24.716 1.467 0.467 2 26
26.183 1.483 0.483 2 28
27.666 2 26


10 Frames / Tick | 6 Ticks / s | 1.81819 to 2 AS
Time(s) Time Between Time Wasted Ticks wasted Frames Wasted
39.55 1.45 0.45 3 30
41 1.5 0.5 3 30
42.5 1.5 0.5 3 30
44 1.516 0.516 3 30
45.516 1.5 0.5 3 30
47.016 1.484 0.484 3 30
48.5 3 30

9 Frames / Tick | 6.66667 Ticks / s | 2.00001 to 2.22222 AS
Time(s) Time Between Time Wasted Ticks wasted Frames Wasted
20.716 1.484 0.484 3 30
22.2 1.516 0.516 3 30
23.716 1.5 0.5 3 30
25.216 1.484 0.484 3 30
26.7 1.5 0.5 3 30
28.2 1.55 0.55 3 30
29.75 3 30

8 Frames / Tick | 7.5 Ticks / s | 2.22223 to 2.5 AS
Time(s) Time Between Time Wasted Ticks wasted Frames Wasted
35.4 1.45 0.45 3 27
36.85 1.45 0.45 3 24
38.3 1.483 0.483 3 27
39.783 1.467 0.467 3 24
41.25 1.466 0.466 3 27
42.716 1.467 0.467 3 24
44.183 3 27
Quoting the below from feedback forums to preserve my discussion.
09/16/2017 02:14 PMPosted by Cratic
Suspicions confirmed. Channeling spells only attempt to proc Etched Sigil once per attack cycle.
09/16/2017 10:00 AMPosted by Cratic
Is this because not every tick of the channeled spell is attempting to proc etched sigil?

I wouldn't expect 100% exact timing, but this does seem a tad slow for a speed co = 3 spell. It ticks at quite a fast rate, even at lower AS.

If the channeled spell of speed co = 3 is only 'attacking' once every 3 ticks, that's an even larger gap of time than I was thinking. Huge damage loss.

Even if this is not considered a bug, it would be nice to have clarification as to what is going on here with mechanics, as this has a huge impact on gearing (say if we should be slotting AS for higher attack rate on channeled builds).

And here's the testing for anyone who is interested:
09/16/2017 02:11 PMPosted by Cratic
So I went ahead and tested for speed co = 3 (Arcane Torrent, Disintegrate):

https://us.battle.net/forums/en/d3/topic/20759186011#post-1

Results show that increases to attack speed generally reduce the time wasted (and allow you to stack twisted sword buff up to 5 twisters).

However, increases to attack speed don't necessarily always reduce time wasted when compared to those breakpoints nearby.

This is interesting because it means there are certain breakpoints where if you're playing a full-on channeling build, you'll want to shoot for those breakpoints over others.
EDIT (10/1/2017): you should verify using the new sigil breakpoints. The below advice was offered based on the OLD (pre-bug fix) sigil breakpoints.

-----

So basically, if I'm reading my results right:

The 12 Frames / Tick | 5 Ticks / s | 1.53847 to 1.66666 AS BP is the one you want to shoot for if you have a full-on channeling build (such as with Firebirds EB).

With a 1.4 AS weapon, that would require only an AS roll from paragon, at a minimum.

You can opt for another AS roll on gear somewhere (weapon or another item) to jump up two Stricken breakpoints, but still stay within the same bracket for Etched Sigil. Interesting.
Why is this news? We've known since Etched Sigil debuted that it only proc'd once per second - irregardless of attack speed.

Given Sigil's nature the short answer to the question of attack speed is that it generally has no real effect. You'll be increasing your LpH, deflection stack gain, proc rate and all of those traditional effects, including the AP/s of casting but you're not going to be increasing the rate of Sigil procs. Your tests suggest that there may be small, non-scaling variance at certain APS but I don't see any conclusive evidence of a signficant, consistent advantage. And to be honest, I'm not even sure what some of your chart entries are meant to represent, most of them don't make any particular sense when discussing tick rates for AT.

If procs happen on actual ticks, then you are going to see a little bit of variance in proc frequency when you don't have a tick rate that isn't a clean factor of 60 (or whatever frame count is Sigil's cooldown of "1 second"). You may have a slight delay for a tick to fall over the 60f interval, which would make sense to account for as a damage loss. For example: at 11 frames per tick you would get ticks at 55f and 66f so you'd basically be 6f longer between procs or 1.1s. That timeframe would shorten with higher breakpoints but would also be non-issues at 30, 20, 15, 12, 10, 6, 5, 3 & 2 frames per tick as well. If Sigil's actual cooldown is some value other than 60 frames, then that changes things as there are going to be very different "clean" divisors.

And more pointedly, even if you can prove a small advantage at certain APS, the tiny % of damage output gain of Sigil procs is mostly likely to be drastically outweighed by stacking Weapon Damage (per hit) stats instead of APS. So once again, I say that unless there is some other mechanic (LpH, deflection, MH procs), APS is mostly irrelevant when you can't meaningfully change the proc rate of Sigil casts. In any situation like that, non-APS damage boosts are the best choice.

-dolynick
Hi Dolynick - You're right. This small of variance would most likely not be worth changing out a significant amount of affixes for. But one? Maybe.

There were a few points to this testing I suppose.

I always assumed channeled spells were attempting to proc etched sigil on every tick - I didn't know that it only attempted to proc etched sigil once per tick set.

Also if results indicated that it was easy to gear for without sacrificing much - why not go for it?

Luckily enough the breakpoint that appears to be best happens to be the easiest to gear for. It also happens to also be what we've been doing pretty much all along.

So no, not much exciting news to be had, but wouldn't you feel better about knowing that the variances exist if you didn't know already?

I know I feel more comfortable recommending this AS breakpoint for any EB channeling build.
09/16/2017 11:19 PMPosted by Cratic
I always assumed channeled spells were attempting to proc etched sigil on every tick - I didn't know that it only attempted to proc etched sigil once per tick set.


AT and (now, since they changed it) Disintegrate behave more or less like cast spam of other skills. To be honest, I haven't stopped to check if RoF was also updated, but afaik it still uses the "old" channelling method. Even so, the old method still had ticks at which it applied updates even if the damage number display didn't reflect that.

09/16/2017 11:19 PMPosted by Cratic
So no, not much exciting news to be had, but wouldn't you feel better about knowing that the variances exist if you didn't know already?


I would prefer to have the actual cooldown of Sigil conclusively proven before I would want to start recommending specific breakpoints. I assume it's 60f as that's generally how the game tracks time and it would be 1s but I've never seen a hard proof of it done.

Generally, I would agree with you. In this case though, I find it irritating to hear people going "Sigil isn't going off every tick???" as this is anything but new info. Sometimes reading this forum feel like watching people rediscover stuff that we already knew over and over. Maybe I'm just getting old and cranky though.

-dolynick
Just saw the updated section of your post.
09/16/2017 02:52 PMPosted by dolynick
You may have a slight delay for a tick to fall over the 60f interval, which would make sense to account for as a damage loss.

That's what I was looking for.

If etched sigil is only allowed to proc once every second (on cooldown or ICD), and the Channel spell is only attempting to attack once every 3 ticks (for a speed co = 3 spell), how many frames are wasted at certain breakpoints in between etched sigil procs if you are holding down a channel?

My method of testing was to look at the test recording, and note times at which Etched Sigil procc'ed via the twisted sword buff icon.
09/16/2017 02:52 PMPosted by dolynick
If Sigil's actual cooldown is some value other than 60 frames, then that changes things as there are going to be very different "clean" divisors.

From the test results, clean divisors actually had some of the highest wasted time. I'm assuming this is due to the ICD still being active when the first potential proc-capable tick was being attempted. If you look at 20 and 10 Frames tests for example, an etched sigil proc didn't occur until a full 40 or 30 frames after the (assumed) 60 frame ES ICD let up.

The odd numbered generally had issues like you suggested, for example, with the 9 frames, there were actually 3 ticks wasted there, but because there was an additional gap of 3 frames due to the odd interval, the full duration of wasted frames was 30, not 27 as you might expect.

This also worked to advantage though in some situations. 12, 11, and 8 frames all show slightly better results.
09/16/2017 11:41 PMPosted by dolynick
In this case though, I find it irritating to hear people going "Sigil isn't going off every tick???" as this is anything but new info. Sometimes reading this forum feel like watching people rediscover stuff that we already knew over and over. Maybe I'm just getting old and cranky though.

Ah, apologies then mate. I was around when etched sigil was implemented, but I wasn't much into testing items in this manner at the time.

Perhaps you could direct me to some of the previous testing that has been performed, if you happen to have that on hand?

Thanks again -

Cratic
the Channel spell is only attempting to attack once every 3 ticks (for a speed co = 3 spell)


This makes no sense as a statement.

You keep talking about this 3 ticks thing when speed coefficients don't really translate like that. Channelling skills (like all skills) do things on tick (cast in the case of other skills, essentially the same thing). What speed coefficient 3 actually means is that the base frames per actions is 20 instead of 60.

09/16/2017 11:44 PMPosted by Cratic
clean divisors actually had some of the highest wasted time


12 frames, which you have chosen as your recommendation is a clean divisor of 60 as well.

"Wasted frames"? There are no wasted frames or ticks in any breakpoint.

09/16/2017 11:44 PMPosted by Cratic
If you look at 20 and 10 Frames tests for example, an etched sigil proc didn't occur until a full 40 or 30 frames after the (assumed) 60 frame ES ICD let up.


Which makes no sense at all for 20f ticks at 60f ICD. It makes even less sense for any X frames cooldown for Sigil at 10f ticks. The former leads me to think that it is not, in fact, a 60f cooldown. The latter leads me to think that there is something flawed in your tests or the entire proc-on-tick method itself. If it is proc'ing on ticks then the maximum delay you should ever see is (tickframes-1)/60 seconds over the 1s cooldown and only when you don't have a clean divisor.

If there is a sound unifying explanation in your tables I'm not seeing it, or your terminology isn't making it clear.

-dolynick
@dolynick

I think what cratic is implying is, the icd is potentially blocking your proc with clean divisor. This is more noticeable with higher latency.

Eg at 10 frames per attack, you expect to get a proc every 6th attack (60 frames /second). But due to icd and latency (say 101ms), then your next proc will be after

60 frames (1s icd) + 11 frames (101ms latency) = 71 frames which in your case becomes on your 8th attack.

In this case, dropping your ias to an odd divisor might work better.

I hope I am not butchering the numbers, but this is what I can glean from my gameplay and cratics number.
09/17/2017 02:50 AMPosted by dejuvenate
@dolynick

I think what cratic is implying is, the icd is potentially blocking your proc with clean divisor. This is more noticeable with higher latency.

Eg at 10 frames per attack, you expect to get a proc every 6th attack (60 frames /second). But due to icd and latency (say 101ms), then your next proc will be after

60 frames (1s icd) + 11 frames (101ms latency) = 71 frames which in your case becomes on your 8th attack.

In this case, dropping your ias to an odd divisor might work better.

I hope I am not butchering the numbers, but this is what I can glean from my gameplay and cratics number.


That's not what he's saying though.

At any rate, I let my self get sucked in and did some testing of my own. I wasn't seeing as much variance in frames per proc as his data but that can possibly be explained by differences in recording and smoothness.

Disintegrate:
1.0 aps, 20f gave procs every 100f
1.01 aps, 19f gave procs every 95f
1.32 aps, 15f, gave procs every 90f
1.54 aps, 12f gave procs every 85f
1.77 aps, 11f gave procs every 88f

Doing a little creative math I found a formula that pretty accurately predicts the observed behavior at all of those data points:

First, the cooldown on Sigil seems to be 80 frames, not 60.
Then apply the tick theory I posted earlier where you have to wait for the next tick over 80f after the first.
The catch is that if it's actually a clean divisor, you have to wait an extra tick.

A formula for all ticks on all channel skills is:
fTick * (1+ROUNDOWN(80/fTick, 0))
where fTick is the frames per tick of your breakpoint.

Let's call our frames per Sigil proc fSigil and our observed proc rate (if I have a test set for that) fObs.

What you get is this:

fTick fSigil fObs
20 100 100
19 95 95
18 90 90
17 85
16 96
15 90 90
14 84
13 91
12 84 85
11 88 88
10 90
9 81
8 88
7 84
6 84
5 85
4 84
3 81


At fTick 12 it's predicting 84 and I observed 85 but that's the only discrepency and it's a single frame only. Also be aware that the tick rates are not 100% consistent in the recordings and I occasionally see a 1 frame variance here an there. What I listed is what I take to be the overall rate.

I also did a RoF test. 1.54 aps, 19f and observed 95 frames per proc. This fits in with the above.

Is it conclusive? No. There are a lot of data points in there that haven't been tested. Feel free to do so yourself and post results here because I'm probably not going to go to all the bother. If you can disprove the theory, alright... but it seems to be holding up on my end so far.

I will say that I am disappointed that it seems that the tooltip for Sigil is off considerably. 80f is 1.33s and it's actually 33.33% slower than stated.

-dolynick
Wow. Thanks for testing Doly!
09/17/2017 01:18 PMPosted by dolynick
That's not what he's saying though.

I was thinking that the extra frames we were seeing above 60, but more than the remainder of whatever frame / tick breakpoint you are at, was being caused by etched sigil only being attempted every 3 ticks, and not on every tick (normalized for speed co = 1, basically).

This is in addition to the impact of the ICD delaying one full tick's worth of frames for evenly divided frame values, as dejuvenate was explaining.

The ICD being higher than 60 was not an assumption I was willing to deviate from apparently, hah! I suppose the fact that my video analysis wasn't perfect down to the frame did not help either.

An 80 frame ICD is a much more plausible explanation in the least. It would explain also why the gap of time (wasted frames as I call them) stay within 20-30 frames when at the 15 frames / tick BP and lower.

With my explanation you might have randomly seen procs that were lower than 20 frames difference (above 60 though).

Overall I'm glad Etched Sigil is being attempted every tick of the channeled spell.

But like you, very sad the ICD is higher than stated on the tooltip.

In the least, we know there is something going on that is causing a delay more than you would expect with having 60 frames in mind as an ICD.

I'll post this on the PTR bug report. Hopefully we can see a bug fix at some point soon.
nice work gents.

Just curious. Did anyone try this with different spenders being cast by the sigil? I highly doubt it matters, but I would be curious to see if different spenders (perhaps with different break points) behaved the same.

cheers,

aloc
09/17/2017 04:37 PMPosted by Cratic
Wow. Thanks for testing Doly!


Well, it's a theory that I can't actually say is 100% right. It fits and it's been confirmed at the data points I have actually tested at but that could be a lucky coincidence. The more BP's tested and confirmed, the more likely it is that I am correct but I only did the 8 easier ones I could without too much trouble.

That's why I said if you can present test data that disproves it, go ahead. For now, I think the 80f cooldown appears to be right.

Also, if this is correct, then the same default recommendation of 1.54 - 1.667 APS for AT/Disintegrate is definitely not correct for RoF. For RoF the 21f 1.3634 - 1.4286 breakpoint is probably the most practical and strongest BP. Hitting that with a Deathwish equipped would actually require not assigning all plevels into the IAS option. Alternatively, the 17f 1.66668 - 1.765 is the option if you want to add IAS.

^^I should test the above RoF stuff to verify because if it does bear out, then the theory is probably correct and there is a fairly big swing between 21f and 19f or 20f (up to ~15% DPS).

09/17/2017 05:16 PMPosted by aloc
Just curious. Did anyone try this with different spenders being cast by the sigil?


No. I tested with Arcane Orb: Obliteration because I figured it would be the easiest to to see. I was using the visible cast aura as my marker for proc instances. I don't believe there is going to be any difference by spender being cast. I can't see any reason for that to be the case. Meteor is going to have a bit longer delay between cast and damage but that has nothing to do with actual cast times.

I don't see breakpoints of the Sigil spender being relevant because, even if they were somehow relevant, the Sigil ICD is well over the fpa of any spender even at 1 APS. If someone does want to test it though, I don't see any harm. It just might be a bit of wasted time.

-dolynick
Also, if this is correct, then the same default recommendation of 1.54 - 1.667 APS for AT/Disintegrate is definitely not correct for RoF. For RoF the 21f 1.3634 - 1.4286 breakpoint is probably the most practical and strongest BP. Hitting that with a Deathwish equipped would actually require not assigning all plevels into the IAS option. Alternatively, the 17f 1.66668 - 1.765 is the option if you want to add IAS.

^^I should test the above RoF stuff to verify because if it does bear out, then the theory is probably correct and there is a fairly big swing between 21f and 19f or 20f (up to ~15% DPS).


Ok. I went and tested RoF at 21f, 20f & 17f. The results were exactly as predicted. I'm fairly confident that we're on the right track as far as predicting what to shoot for. If someone wants to test some high APS breakpoints that would probably remove any remaining doubts.

-dolynick
Here are full tables of predicted Sigil proc rates:


Tick/Speed Coefficient 3 (Arcane Torrent, Disintegrate)
Min APS Max APS Frames/t fSigil
4.000001 5 4 84
3.333334333 4 5 85
2.857143857 3.333333333 6 84
2.500001 2.857142857 7 84
2.222223222 2.5 8 88
2.000001 2.222222222 9 81
1.818182818 2 10 90
1.666667667 1.818181818 11 88 *Tested & Verified*
1.538462538 1.666666667 12 84 *Tested & Verified*
1.428572429 1.538461538 13 91
1.333334333 1.428571429 14 84
1.250001 1.333333333 15 90 *Tested & Verified*
1.176471588 1.25 16 96
1.111112111 1.176470588 17 85
1.052632579 1.111111111 18 90 *Tested & Verified*
1.000001 1.052631579 19 95 *Tested & Verified*
0.952381952 1 20 100 *Tested & Verified*
0.909091909 0.952380952 21 84
0.869566217 0.909090909 22 88
0.833334333 0.869565217 23 92
0.800001 0.833333333 24 96
0.769231769 0.8 25 100
0.740741741 0.769230769 26 104
0.714286714 0.740740741 27 81
0.689656172 0.714285714 28 84
0.666667667 0.689655172 29 87
0.64516229 0.666666667 30 90
0.625001 0.64516129 31 93
0.606061606 0.625 32 96
0.588236294 0.606060606 33 99
0.571429571 0.588235294 34 102
0.555556556 0.571428571 35 105
0.540541541 0.555555556 36 108
0.526316789 0.540540541 37 111
0.512821513 0.526315789 38 114
0.500001 0.512820513 39 117
0.487805878 0.5 40 120

Tick/Speed Coefficient 2 (Electrocute, Ray of Frost)
Min APS Max APS Frames/t fSigil
4.285715286 5 6 84
3.750001 4.285714286 7 84
3.333334333 3.75 8 88
3.000001 3.333333333 9 81
2.727273727 3 10 90
2.500001 2.727272727 11 88
2.307693308 2.5 12 84
2.142858143 2.307692308 13 91
2.000001 2.142857143 14 84
1.875001 2 15 90
1.764706882 1.875 16 96
1.666667667 1.764705882 17 85 *Tested & Verified*
1.578948368 1.666666667 18 90
1.500001 1.578947368 19 95 *Tested & Verified*
1.428572429 1.5 20 100 *Tested & Verified*
1.363637364 1.428571429 21 84 *Tested & Verified*
1.304348826 1.363636364 22 88
1.250001 1.304347826 23 92
1.200001 1.25 24 96
1.153847154 1.2 25 100
1.111112111 1.153846154 26 104
1.071429571 1.111111111 27 81
1.034483759 1.071428571 28 84
1.000001 1.034482759 29 87
0.967742935 1 30 90
0.937501 0.967741935 31 93
0.909091909 0.9375 32 96
0.882353941 0.909090909 33 99
0.857143857 0.882352941 34 102
0.833334333 0.857142857 35 105
0.810811811 0.833333333 36 108
0.789474684 0.810810811 37 111
0.769231769 0.789473684 38 114
0.750001 0.769230769 39 117
0.731708317 0.75 40 120
0.714286714 0.731707317 41 82
0.697675419 0.714285714 42 84
0.681819182 0.697674419 43 86
0.666667667 0.681818182 44 88
0.652174913 0.666666667 45 90
0.638298872 0.652173913 46 92
0.625001 0.638297872 47 94
0.612245898 0.625 48 96
0.600001 0.612244898 49 98
0.588236294 0.6 50 100
0.576924077 0.588235294 51 102
0.566038736 0.576923077 52 104
0.555556556 0.566037736 53 106
0.545455545 0.555555556 54 108
0.535715286 0.545454545 55 110
0.526316789 0.535714286 56 112
0.517242379 0.526315789 57 114
0.508475576 0.517241379 58 116
0.500001 0.508474576 59 118
0.491804279 0.5 60 120


-dolynick
Yeah, I highly doubted the spenders fps mattered, but having a 80 fps seems too similar to 60fps + ~2X. This made me wonder if the sigil could trigger a spender which requires the built in fps (the +~20). I really don't know the mechanics well enough to be honest. I'm on my mobile so it's hard to check, but what is the the fps for AO obliteration? The table shows 1 APS as 20 for the Disintegrate skill I believe. I wonder if their coding causes it to default attack at 20 fps every 60 seconds. So it actually could be something like:

60 fps -> cast at 1AS (20 fps), start 60 fps ICD, repeat. Producing the in-game ICD of 80fps

If this is the case, we would have no way of knowing I guess unless it varies with the spenders fps. I think it would fall in line with what you have worked out.

The only reason I thought this is that they might believe it to be programmed with a 60second ICD, but the implementation mechanics make it as you have described.

* sorry for multiple edits. I'm not so good on the smart phone.
09/17/2017 09:41 PMPosted by aloc
what is the the fps for AO obliteration?


The base FPA for Arcane Orb is 57.8571 frames.

If it were using the base FPA of the spender, then my tables wouldn't be hitting the mark spot on anywhere. I'd expect them all to be 2 frames off since the math is using a round 80 and not 77.8571.

I will admit that I have not tested using another spender though.

-dolynick
Yeah, it could be defaulting to the channeled skills fpa, which is 20. Its hard to imagine that they would put in a pure and simple wrong number, but we all make goof ups.

Impressive testing gents.
A newbie question on Etched Sigil and Arcane Orb (I am a casual wizard player so it takes time for me even to decipher all those abbreviations).

I'd like to play Frozen Orb in next patch. I wonder if it makes sense to use Etched Sigil and low arc power coster - like Ray of Frost, and shoot Frozen Orbs for free? As I understand, I will shoot them much slower than if I shooted them directly?

And another question: will such Etched Sigil Frozen Orb shot trigger Unstable Scepter additional orb? As far as I understand - not: Such an arcane orb is considered an item proc and it will not trigger another proc. Is it correct?

Join the Conversation

Return to Forum