Diablo® III

Mechanics And Game Information Compendium

You can get +100%cc with Ray of Frost, but it requires:
*High critical chance - like 60%. So you'd use things like Nat's Ring+Boots with the ring having some intrinsic cc)
*Pickup some gear with %Crit Chance for Ray of Frost skill (ie Wizard Hat, Maras, and Skull Grasp)
*Use Energy Armor (Pinpoint Barrier)
*Trigger the +15%cc from Frost Nova (Deep Freeze) - This one is the issue I think, since the bonus doesn't last longer than 12s.
*3%cc bonus from Scoundrel.
People have done 100%cc before with Energy Twister, so it should be basically the same principle (albeit still a PITA). :) I believe you can get ~82%cc for Ray of Frost just from gear. So that's about 90% without resorting to Frost Nova (Deep Freeze).


There´s a couple of problems with that if you want to collect samples with different APS
- the cost of certain RoF items you suggest, not to mention their accessability
- I´d need two of the same items which are already difficult to get - to test different APS
- for example Nat´s 2 set bonus with either the ring and pants or helm and pants, the first option means you´ll always have at least 8% IAS the second would mean no RoF crit chance, no APoC
- you´re right about Deep Freeze which is out of the question as you need to have stable conditions for a CM test and refreshing Frost Nova stops the channeling (unless you want to also count the frames between the channels) plus having 5 targets around you is kind of tricky for survival with your research gear
- channeling high APS is difficult to sustain so you need APoC, which again puts stress on your gear choices
etc.

But OK, let´s not be negative and check what´s possible:

base 5 helm 6(13) bracers 6 ring 6 ring 6(13) amulet 10 gloves 10 source 10 pants 1 belt 1 scoundrel 3 pinpoint 5 = 75% cap

Works, I could get this without any IAS with a Nat´s Helm and a belt instead of Inna´s Pants. Unless someone lends me their max crit chance Mara or Skull Grasp, I can´t get more because I wouldn´t be able to test different APS values with the same crit chance value. But 75% should be better than the coin flip 50% that I tested.

The maximum aps with this setup would be either an 11% dagger (1.665) or chantodo´s force (1.65), IAS on amulet, 2 rings, chest, bracers, pants, gloves, belt, chanto´s force. All with perfect rolls should put me at 3.01365 aps with the dagger, I´d need one more item for chanto´s will to break 3 aps and I don´t have a 6% crit mempo (neither a 6% crit chance reflection) :/

I´ll see what I can do.
Reply Quote
Absolutely, it's not easy at all to do this. In fact, it's a PITA. I was merely suggesting that it's possible if you gear heavily towards it. ^_^
Reply Quote
Wow Nubtro... Yes, that's a LOT of frames to click through.
Thanks. I won't get to it before the weekend, but I'll try and look at some more APS values, especially comparing setups with a larger APS difference.

By the way, what do you use for video playback? Usually I'm very happy with VLC. But for video analysis it really bothers me that elapsed time is shown in minutes and seconds only. I wasn't able to find a way to display frames or at least milliseconds.
Reply Quote
06/20/2013 02:12 AMPosted by Nubtro
base 5 helm 6(13) bracers 6 ring 6 ring 6(13) amulet 10 gloves 10 source 10 pants 1 belt 1 scoundrel 3 pinpoint 5 = 75% cap


Gear:
Helm: %6 + %8 Ray of Frost (rare wiz hats only)
Neck: %10 + %7 Ray of Frost (Mara's)
Chest: None
Shoulder: None
Gloves: %10 (any source)
Belt: %1 (Inna's or Tal Rasha's)
Bracers: %6 (any pair really)
Legs: %1 (Inna's)
Ring1: %6+%6 Ray of Frost (Skull Grasp)
Ring2: %6 (Nat's Ring)
Boots: 0% (Nat's Boots 2-set bonus)
Source: %10
Weapon: %1 (Kill)

%78(gear) + %7 (nat's bonus) + %5 (Base) + %3 (scoundrel) + %5 (Pinpoint) = %98 CC without deepfreeze
Edited by BDF#1838 on 6/24/2013 5:57 AM PDT
Reply Quote
Posts: 3,301
View profile
You can get +100%cc with Ray of Frost, but it requires:
*High critical chance - like 60%. So you'd use things like Nat's Ring+Boots with the ring having some intrinsic cc)
*Pickup some gear with %Crit Chance for Ray of Frost skill (ie Wizard Hat, Maras, and Skull Grasp)
*Use Energy Armor (Pinpoint Barrier)
*Trigger the +15%cc from Frost Nova (Deep Freeze) - This one is the issue I think, since the bonus doesn't last longer than 12s.
*3%cc bonus from Scoundrel.
People have done 100%cc before with Energy Twister, so it should be basically the same principle (albeit still a PITA). :) I believe you can get ~82%cc for Ray of Frost just from gear. So that's about 90% without resorting to Frost Nova (Deep Freeze).


There´s a couple of problems with that if you want to collect samples with different APS
- the cost of certain RoF items you suggest, not to mention their accessability
- I´d need two of the same items which are already difficult to get - to test different APS
- for example Nat´s 2 set bonus with either the ring and pants or helm and pants, the first option means you´ll always have at least 8% IAS the second would mean no RoF crit chance, no APoC
- you´re right about Deep Freeze which is out of the question as you need to have stable conditions for a CM test and refreshing Frost Nova stops the channeling (unless you want to also count the frames between the channels) plus having 5 targets around you is kind of tricky for survival with your research gear
- channeling high APS is difficult to sustain so you need APoC, which again puts stress on your gear choices
etc.

But OK, let´s not be negative and check what´s possible:

base 5 helm 6(13) bracers 6 ring 6 ring 6(13) amulet 10 gloves 10 source 10 pants 1 belt 1 scoundrel 3 pinpoint 5 = 75% cap

Works, I could get this without any IAS with a Nat´s Helm and a belt instead of Inna´s Pants. Unless someone lends me their max crit chance Mara or Skull Grasp, I can´t get more because I wouldn´t be able to test different APS values with the same crit chance value. But 75% should be better than the coin flip 50% that I tested.

The maximum aps with this setup would be either an 11% dagger (1.665) or chantodo´s force (1.65), IAS on amulet, 2 rings, chest, bracers, pants, gloves, belt, chanto´s force. All with perfect rolls should put me at 3.01365 aps with the dagger, I´d need one more item for chanto´s will to break 3 aps and I don´t have a 6% crit mempo (neither a 6% crit chance reflection) :/

I´ll see what I can do.


I have both a RoF SOJ and a SG you could use if it helps.
Reply Quote
I spent like 20m to get the gear, oh well my curiosity is bigger...

Test 1: 75% CC 3.0096 aps (1.76 * 1.71) RoF Cold Blood 0.333 proc coefficient

3 sec 5 frames (52 frames)
25 sec 52 frames 22 sec 47 frames (1367)
1 sec 53 frames (70 frames)
23 sec 37 frames 21 sec 44 frames (1304)
1 sec 27 frames (42 frames)
22 sec 4 frames 20 sec 37 frames (1237)
1 sec 26 frames (36 frames)
24 sec 12 frames 22 sec 46 frames (1366)
2 sec 56 frames (39 frames)
23 sec 51 frames 20 sec 55 frames (1255)
_____________________________________________________________
30000 CD frames 239 frames outside of RoF 6529 RoF frames (1305.8 avg)
23232 CM frames = 387.2 procs / 108.81667 sec 77.44 procs / 21.7633 sec
3.5583/sec

Test 2: 75% CC 1.20 aps RoF Cold Blood 0.333 proc coefficient

1 sec 27 frames (42 frames)
25 sec 37 frames 24 sec 10 frames (1450)
2 sec 25 frames (68 frames)
27 sec 47 frames 25 sec 22 frames (1522)
1 sec 50 frames (57 frames)
22 sec 10 frames 20 sec 20 frames (1220)
1 sec 54 frames (64 frames)
27 sec 13 frames 25 sec 19 frames (1519)
1 sec 41 frames (62 frames)
24 sec 31 frames 22 sec 50 frames (1370)
4 sec 2 frames (48 frames)
27 sec 2 frames 23 sec 0 frames (1380)
_____________________________________________________________
36000 CD frames 341 frames outside of RoF 8461 RoF frames (1410.16667 avg)
27198 CM frames = 453.3 procs / 141.01667 75.55 procs / 23.502778 sec
3.2145/sec

Not enough tests done (I´ll do more when I have more time) but there seems to be a slight aps scaling of CM, but between 1.20 and 3.01 aps I only got a 10% CM increase in procs/sec.

Any kind person willing to reverse calc the average number of frames between the "checks" for CM procs in each case? I have to leave now.

EDIT: Adding most recent test data.

___________________________________________________________________________________
75% CC 1.50 aps
14 sec 1 frame (51)
39 sec 32 frames 25 sec 31 frames (1531)
44 sec 25 frames (84)
66 sec 27 frames 22 sec 2 frames (1322)
9 sec 15 frames (72)
31 sec 2 frames 21 sec 47 frames (1307)
34 sec 34 frames (56)
60 sec 8 frames 25 sec 34 frames (1534)
4 sec 7 frames (61)
29 sec 25 sec 25 sec 18 frames (1518)
30000 -324 -7212 (1442.4)
22464 = 374.4 / 120.2 74.88 procs / 24.04 sec 3.1148/sec
___________________________________________________________________________________
___________________________________________________________________________________
75% CC 1.512 aps
2 sec 17 frames (58)
25 sec 44 frames 23 sec 27 frames (1407)
28 sec 38 frames (60)
52 sec 3 frames 23 sec 25 frames (1405)
29 sec 40 frames (42)
52 sec 15 frames 22 sec 35 frames (1355)
21 sec 50 frames (62)
43 sec 14 frames 21 sec 24 frames (1284)
46 sec 11 frames (54)
69 sec 16 frames 23 sec 5 frames (1385)
23 sec 22 frames (76)
45 sec 30 frames 22 sec 8 frames (1328)
48 sec 48 frames (47)
75 sec 20 frames 26 sec 32 frames (1592)
42000 -399 -9756 (1393.7)
31845 = 530.75 / 162.6 75.82 procs / 23.229 sec 3.2641/sec
___________________________________________________________________________________
___________________________________________________________________________________
75% CC 1.582 aps
1 sec 48 frames (42)
32 sec 42 frames 30 sec 54 frames (1854)
35 sec 14 frames (57)
64 sec 23 frames 29 sec 9 frames (1749)
30 sec 4 frames (39)
51 sec 39 frames 21 sec 35 frames (1295)
54 sec 1 frame (44)
78 sec 53 frames 24 sec 52 frames (1492)
43 sec 2 frames (81)
71 sec 42 frames 28 sec 40 frames (1720)
30000 -263 -8110
21627 = 360.45 / 135.16667 72.09 procs / 27.0333 sec 2.6667/sec
___________________________________________________________________________________
___________________________________________________________________________________
75% CC 1.668 aps
2 sec 0 frames (59)
21 sec 35 frames 19 sec 35 frames (1175)
27 sec 56 frames (61)
50 sec 56 frames 23 sec 0 frames (1380)
54 sec 0 frames (79)
76 sec 21 frames 22 sec 21 frames (1341)
18 sec 52 frames (49)
41 sec 19 frames 22 sec 27 frames (1347)
44 sec 26 frames (49)
67 sec 1 frame 22 sec 35 frames (1355)
30000 -297 -6598 (1319.6)
23105 = 385.08333 / 109.96667 77.01667 procs / 21.99333 sec 3.5018/sec
___________________________________________________________________________________
___________________________________________________________________________________
75% CC 1.77 aps
2 sec 31 frames (41)
23 sec 13 frames 20 sec 42 frames (1242)
53 sec 5 frames (31)
75 sec 0 frames 21 sec 55 frames (1315)
20 sec 51 frames (85)
43 sec 17 frames 22 sec 26 frames (1346)
48 sec 27 frames (81)
71 sec 26 frames 22 sec 59 frames (1379)
19 sec 10 frames (69)
47 sec 6 frames 27 sec 56 frames (1676)
30000 -307 -6958 (1391.6)
22735 = 378.91667 / 115.9667 75.78333 procs / 23.1933 sec 3.2675/sec
___________________________________________________________________________________
___________________________________________________________________________________
75% CC 1.876 aps
6 sec 32 frames (34)
31 sec 57 frames 25 sec 25 frames (1525)
8 sec 24 frames (37)
32 sec 11 frames 23 sec 47 frames (1427)
35 sec 1 frame (25)
57 sec 18 frames 22 sec 17 frames (1337)
2 sec 16 frames (47)
28 sec 0 frames 25 sec 44 frames (1544)
49 sec 27 frames (41)
76 sec 21 frames 26 sec 54 frames (1614)
30000 -184 -7447 (1489.4)
22369 = 372.81667 / 124.11667 74.5633 procs / 24.8233 sec 3.0038/sec
___________________________________________________________________________________
___________________________________________________________________________________
75% CC 2.002 aps
2 sec 16 frames (36)
23 sec 7 frames 20 sec 51 frames (1251)
39 sec 2 frames (34)
64 sec 41 frames 25 sec 39 frames*(1539)
1 sec 32 frames (44)
27 sec 11 frames 25 sec 39 frames (1539)
4 sec 29 frames (39)
24 sec 43 frames 20 sec 14 frames (1214)
1 sec 45 frames (57)
22 sec 37 frames 20 sec 52 frames (1252)
30000 -210 -6795 (1359 avg)
22995 = 383.25 / 113.25 76.65 procs / 22.65 sec 3.3841/sec
___________________________________________________________________________________
___________________________________________________________________________________
75% CC 2.145 aps
1 sec 31 frames (40)
22 sec 37 frames 21 sec 6 frames (1266)
31 sec 53 frames (38)
51 sec 40 frames 19 sec 47 frames (1187)
22 sec 36 frames (40)
44 sec 30 frames 21 sec 54 frames (1314)
59 sec 35 frames (53)
83 sec 10 frames 23 sec 35 frames (1415)
37 sec 40 frames (33)
59 sec 21 frames 21 sec 41 frames (1301)
30000 -204 -6483 (1296.6 avg)
23313 = 388.55 / 108.05 77.71 procs / 21.7 sec 3.5811/sec
___________________________________________________________________________________
___________________________________________________________________________________
75% CC 2.31 aps
1 sec 29 frames (49)
21 sec 4 frames 19 sec 35 frames (1175)
1 sec 41 frames (59)
30 sec 5 frames 28 sec 24 frames (1704)
1 sec 21 frames (43)
29 sec 38 frames 28 sec 17 frames (1697)
1 sec 19 frames (44)
30 sec 0 frames 28 sec 41 frames (1721)
1 sec 22 frames (47)
27 sec 54 frames 26 sec 32 frames (1592)
1 sec 30 frames (47)
26 sec 6 frames 24 sec 36 frames (1476)
1 sec 31 frames (45)
28 sec 16 frames 26 sec 45 frames (1605)
42000 -334 -10970 (1567 avg)
30696 = 511.6 / 182.833 73.0857 procs / 26.119 sec 2.7982/sec
1 sec 48 frames (40)
31 sec 57 frames 30 sec 9 frames (1809)
12 sec 4 frames (40)
34 sec 50 frames 22 sec 46 frames (1366)
54000 -414 -14145 (1571.7 avg)
39441 = 657.35 / 235.75 73.0389 procs / 26.1944 sec 2.7883/sec
___________________________________________________________________________________
___________________________________________________________________________________
75% CC 2.50155 aps
1 sec 27 frames (46)
22 sec 51 frames 21 sec 24 frames (1284)
2 sec 1 frame (80)
22 sec 1 frame 20 sec 0 frames (1200)
2 sec 16 frames (47)
23 sec 10 frames 20 sec 54 frames (1254)
3 sec 9 frames (46)
24 sec 16 frames 21 sec 7 frames (1267)
1 sec 41 frames (64)
24 sec 5 frames 22 sec 24 frames (1344)
30000 -283 -6349 (1270 avg)
23368 = 389.4667 / 105.81667 77.8933 procs / 21.1633 3.6806/sec
___________________________________________________________________________________
___________________________________________________________________________________
75% CC 2.73105 aps
1 sec 26 frames (49 frames)
22 sec 4 frames 20 sec 38 frames (1238)
1 sec 26 frames (50+20 frames)
23 sec 52 frames 22 sec 26 frames (1346)
1 sec 43 frames (63 frames)
24 sec 33 frames 22 sec 50 frames (1370)
2 sec 8 frames (91 frames)
23 sec 5 frames 20 sec 57 frames (1257)
4 sec 3 frames (198 frames)
24 sec 32 frames 20 sec 29 frames (1229)
30000 -471 -6440 23089 = 384.81667 / 107.33 76.963 procs / 21.4667 3.5852/sec
Edited by Nubtro#2147 on 6/23/2013 7:44 AM PDT
Reply Quote
06/21/2013 08:37 AMPosted by Nubtro
there seems to be a slight aps scaling of CM, but between 1.20 and 3.01 aps I only got a 10% CM increase in procs/sec.

astounding results actually...looks like only CC affects CM considerably...AS doesn`t matter much for beam skills in terms of number of procs/sec...go figure
Reply Quote
Yeah, I think Nubtro has just broken CM Wizards o.O
We knew there's something fishy going on, but that's insane. I don't really know what to think of it yet.

But I took a closer look at Spectral Blade - Deep Cuts. It's kind of the budget version of 100% crit chance. The whole DoT is either crit or non-crit. So when it does crit, we got 3 full seconds of crit frames.
Now what I found from my video analysis is very similar to Nubtro's results with RoF. But at least with the data I have, it looks more like a 6 frames check interval for CM procs, instead of 3. No matter how often I try, I have never seen procs with less than 6 frames between them.
Could be bad luck. But on the other hand, 3 frames means 60 potential procs over 3 seconds. The LoH proc coefficient is 0.14. So 60*0.14? That does sound like a bit too much.

The 75% crit tests also seem a bit off.
In the 3APS test there was an average of 1305.8 frames of RoF channeling.
Let's assume 75% of those were crit frames -> 979.35 frames.
Now if checks are performed every 3 frames that leaves 326.45 proc frames.
With a proc coefficient of 0.3333 that should be around 108.8 procs.
But there were only 77.44 procs.

Am I screwing up the math here? Or are there still some pieces of the puzzle missing?

PS: I'm still working on the update for the compendium. Lots things to add and rework.
Reply Quote
Yeah, I think Nubtro has just broken CM Wizards o.O
We knew there's something fishy going on, but that's insane. I don't really know what to think of it yet.

PS: I'm still working on the update for the compendium. Lots things to add and rework.

Not sure what you mean by that but it´s not over until we have the exact mechanics, so don´t update that part just yet. I´ve done more tests the past couple of days at most breakpoint values. I´ll put the exact test results in my previous post, here´s an overview:

The skill that I tested was Ray of Frost - Cold Blood because of its reduced cost and relatively high proc coefficient, which allowed me to channel it even at very high aps long enough to bring the Archon Pure Power cooldown to zero.

Note that I collected 5 samples of each because it would take ages to get more and I needed a basic comparison table - even 5 samples showed differences between certain breakpoints.

crit chance <-> breakpoint <-> tested aps <-> CM procs/sec <-> LoH check length <-> CM checks/sec <-> CM check length
75% >30/10 3.00960 aps 3.5583 CM procs 9 14.2332 4.2155
75% >30/11 2.73105 aps 3.5852 CM procs 10 14.3408 4.1839
75% >30/12 2.50155 aps 3.6806 CM procs 11 14.7224 4.0754
75% >30/13 2.31000 aps 2.7883 CM procs 12 11.1532 5.3796
75% >30/14 2.14500 aps 3.5811 CM procs 13 14.3244 4.1887
75% >30/15 2.00200 aps 3.3841 CM procs 14 13.5364 4.4325
75% >30/16 1.87600 aps 3.0038 CM procs 15 12.0152 4.9937
75% >30/17 1.77000 aps 3.2675 CM procs 16 13.0700 4.5907
75% >30/18 1.66800 aps 3.5018 CM procs 17 14.0320 4.2760
75% >30/19 1.58200 aps 2.6667 CM procs 18 10.6668 5.6249
75% >30/20 1.51200 aps 3.2641 CM procs 19 13.0564 4.5954
75% >30/21 1.50000 aps 3.1148 CM procs 20 12.4592 4.8157
75% >30/26 1.20000 aps 3.2145 CM procs 25 12.8580 4.6664

Note that the last two columns are just calculated (assumed) values derived from the actual CM procs/sec values (first one by dividing the CM procs/sec value by crit chance/100 and by the proc coefficient and the second value by dividing 60 frames by the previous column value) and they should´ve helped me figure out what´s going on (didn´t help lol). You might as well ignore those two columns and just focus on the underlined CM procs/sec values that I got in the game in my tests that you can find here
http://us.battle.net/d3/en/forum/topic/8770117237?page=7#126

So there are indeed CM breakpoints among LoH/APoC breakpoints. At the end of the day, I know that people will only care about "tell me which breakpoints gives me the most CM procs" and the above table tells you that sort of info...but I´m researcher and it´ll bug me until I figure out why.

As you can see, certain breakpoints are less desirable because they give you less CM procs - the exact reason is yet to be figured out. The ones that gave me the least CM procs were 12 and 18 frame long LoH/APoC "checks" (when the game checks for a crit and all subsequent frames after this check are the same - crit or no crit until the next check). If you divide 12 and 18 by 6 you get a whole number...nice pattern, right? lol...

Anyway. I need to do a few more tests to confirm that the CM breakpoints depend on the "check" frame length and not on proc coefficients so I´ll see if Sleet Storm gives me 0.1875/0.333 = 56.306% of the CM procs that Cold Blood gave me at the same aps value that I test.
Edited by Nubtro#2147 on 6/23/2013 11:09 AM PDT
Reply Quote
06/23/2013 07:39 AMPosted by Nubtro
Not sure what you mean by that but it´s not over until we have the exact mechanics, so don´t update that part just yet.

I meant this is big news. People are chasing higher breakpoints in part because they expect a significant increase in CM and Shocking Aspect procs. From what you've found out so far, the only benefit of a higher breakpoint might be fast Energy Twister cast rate. And of course more LoH/APoC. but after all, most want to optimize for maximum DPS, not for sustain and AP management.

So what are you thinking right now? CM might be completely independent from proc coefficients and just working off a (variable?) CM check length, somehow interacting (but not really scaling up) with the LoH tick frame lengths?
Reply Quote
Tested Sleet Storm

1.51200 aps 75% CC
2 sec 44 frames (101)
35 sec 28 frames 32 sec 44 frames (1964)
46 sec 26 frames (91+(212))
86 sec 22 frames 39:56 - 3:32 = 36 sec 24 frames (2184)
3 sec 59 frames (69)
39 sec 45 frames 35 sec 46 frames (2146)
1 sec 30 frames (88)
39 sec 25 frames 37 sec 55 frames (2275)
3 sec 11 frames (59)
35 sec 35 frames 32 sec 24 frames (1944)
29 seec 15 frames (82)
65 sec 40 frames 36 sec 25 frames (2185)
26 sec 29 frames (80+(60+20+8))
67 sec 36 frames 41:07 - 1:22 = 39 sec 45 frames (2385)
34 sec 30 frames (94)
73 sec 32 frames 39 sec 2 frames (2342)
48000 -664 (83 avg) -17425 (2178.125 avg)
29911 = 498.51667 / 290.41667 62.3145833 procs / 36.3020833 sec 1.7166/sec
_________________________________________________________
1.58200 aps 75% CC
4 sec 47 frames (43)
43 sec 53 frames 39 sec 6 frames (2346)
2 sec 12 frames (57)
46 sec 13 frames 44 sec 1 frame (2641)
2 sec 34 frames (74)
43 sec 43 frames 41 sec 9 frames (2469)
4 sec 6 frames (86)
45 sec 4 frames 40 sec 58 frames (2458)
2 sec 35 frames (73)
43 sec 44 frames 41 sec 9 frames (2469)
30000 -333 (66.6 avg) -12383 (2476.6 avg)
17284 = 288.0667 / 206.38333 57.61333 procs / 41.27667 sec 1.3958/sec
crit chance <-> breakpoint <-> tested aps <-> CM procs/sec <-> LoH check length <-> CM checks/sec <-> CM check length
75% >30/19 1.58200 aps 1.3958 CM procs 18 9.9257 6.0449
75% >30/20 1.51200 aps 1.7166 CM procs 19 12.2069 4.9153

compare with Cold Blood
75% >30/19 1.58200 aps 2.6667 CM procs 18 10.6668 5.6249
75% >30/20 1.51200 3.2641 CM procs 19 13.0564 4.5954

I got slightly less procs than I predicted.
0.1875/0.333 = 0.56306
2.6667 * 0.56306 = 1.5017; I got 1.3958 (as if the SS proc coefficient was 0.1743)
3.2641 * 0.56306 = 1.8379; I got 1.7166 (as if the SS proc coefficient was 0.1751)

Not sure why, but it´s clear that the less desirable CM breakpoints also apply to Sleet Storm. People running Ray of Frost should stay away from the following aps breakpoints

0.96775-1.00000 aps (needs test confirmation, based on theory)
1.20001-1.25000 aps (needs test confirmation, based on theory)
1.57895-1.66666 aps (lowest number of CM procs according to my tests, considerably less than other breakpoints)
2.30770-2.50000 aps (second lowest number of CM procs according to my tests, considerably less than other breakpoints)

because these provide considerably less CM procs than the rest. More tests need to be done to figure out which breakpoints give you the most CM procs. The table in my previous post
http://us.battle.net/d3/en/forum/topic/8770117237?page=7#129
should give you a general idea which breakpoint is better and which on is worse.


I meant this is big news. People are chasing higher breakpoints in part because they expect a significant increase in CM and Shocking Aspect procs. From what you've found out so far, the only benefit of a higher breakpoint might be fast Energy Twister cast rate. And of course more LoH/APoC. but after all, most want to optimize for maximum DPS, not for sustain and AP management.

So what are you thinking right now? CM might be completely independent from proc coefficients and just working off a (variable?) CM check length, somehow interacting (but not really scaling up) with the LoH tick frame lengths?

Yeah it´s kind of big news. According to my results there is no CM scaling with aps like with LoH/APoC (more aps = more procs) when it comes to Ray of Frost (also assuming the same applies to Disintegrate and Archon Disintegration Wave). But there seem to be better and worse breakpoints when it comes to CM procs (my tests showed up to 38% more procs between the best and worst one).

I haven´t tested Wicked Wind or stuff like Blizzard so I´m not sure if they follow the same pattern. Those who want to maximize those skills will have figure it out themselves. These CM tests are really time consuming. But seeing as I´ve already tested a lot of breakpoints and WW has the same breakpoints as RoF, it should be an easier job if you compare your results with mine.
Edited by Nubtro#2147 on 6/23/2013 11:18 AM PDT
Reply Quote
@Nubtro: If true, these are VERY interesting results!

You wouldn't happen to have the 5 measurements for each attack speed breakpoint available? I'm curious on the amount of variance in each averaged measurement you made per attack speed breakpoints for the CM/procs. I think it's worth ruling out that these "less desirable" breakpoints do not simply arise from large measurement variability. For example, would the entire dataset pass an ANOVA test for significant differences? And, if so, would we find that the "less desirable" attack speed breakpoints are significantly different from other breakpoints (ie with a 2 sample t-Test)?

I'm a little leery of trusting mean measurements, especially if some of the data is behaving in a seemingly abberant way. And this is where stats come in handy. If the data can't pass a significant test that means we probably don't have enough data to actually say that some breakpoints are in fact undesirable (or it's random). Either way, we probably need to rule out the null hypothesis before we can really say anything for sure.
Edited by TekkZero#1963 on 6/23/2013 11:29 AM PDT
Reply Quote
@Nubtro: If true, these are VERY interesting results!
You wouldn't happen to have the 5 measurements for each attack speed breakpoint available?

If you had read it thoroughly ;)
You might as well ignore those two columns and just focus on the underlined CM procs/sec values that I got in the game in my tests that you can find here
http://us.battle.net/d3/en/forum/topic/8770117237?page=7#126

It´s on this page, just a few posts above.

The more crit chance, the more reliable the results are. If only I could test more than 75%, but I had to borrow a reflection from a friend for the Sleet Storm tests already (needed 3 APoC sources to be able to channel it against one oponent). Another option would be to run the base 120 sec Archon cooldown. I think the gap between the better and the worse breakpoints would be more distinct.

But yeah, feel free to help me with the tests. They require a lot of patience and clicking hehe. I have to manually count the frames from Archon activation until I start channeling RoF (first damage frame on the target), then click until the next second starts to figure out the exact frame my RoF started (seconds and frames) and afterwards the moment the CD is reduced to zero. I use a freezing/blinding Scoundrel and Diamond Skin to survive the channeling without stopping it. Not an easy task with research gear.
Reply Quote
Updated a few chapters. Added everything from Nubtro's suggestions and placed warning labels where the mechanics are a bit fuzzy right now.

@Nubtro: Why not starters area in A1? No survival issues there. Or does your research gear have too much DPS for the zombies?

I also wrote a small LUA plugin for VLC that displays the frame count on screen. The first 5-10 frames after video start are a bit weird, but after that it's 1 frame per click. And 60 frames is exactly at 1s, 120=2s, etc. so it syncs up well. So finally I can skip around without losing track of frames, yay.
Let me know if you're interested.
Reply Quote
06/20/2013 12:45 PMPosted by BDF
base 5 helm 6(13) bracers 6 ring 6 ring 6(13) amulet 10 gloves 10 source 10 pants 1 belt 1 scoundrel 3 pinpoint 5 = 75% cap


Gear:
Helm: %6 + %8 Ray of Frost (rare wiz hats only)
Neck: %10 + %7 Ray of Frost (Mara's)
Chest: None
Shoulder: None
Gloves: %10 (any source)
Belt: %1 (Inna's or Tal Rasha's)
Legs: %1 (Inna's)
Ring1: %6+%6 Ray of Frost (Skull Grasp)
Ring2: %6 (Nat's Ring)
Boots: 0% (Nat's Boots 2-set bonus)
Source: %10

%71(gear) + %7 (nat's bonus) + %5 (Base) + %3 (scoundrel) + %5 (Pinpoint) = %91 CC without deepfreeze


You missed Bracers for 6%, bringing it to 97% without DF. Isn't there a weapon that adds 1% to CC?

As for the rest of the recent posts, that looks like a lot of interesting data. I only had time to skim over the results and unfotunatelly don't really have any time to help contribute since I'm in the midst of running code for my research and I'm not going to have much, if any, spare time for quite some time. I do really appreciate all the time and effort put into gathering and analyzing the data though, so big thanks to all who are doing that.

If you end up with a consensus of the CM/SA behavior, try to catch me in game or send me an IM at diablofans and I'll try to find a couple hours to update my CMWW simulator, not that anyone really seems to use it, heh. So far it seems like you have pretty conclusive results for RoF at least. Since I've only glanced over it I'm not sure if the testing was conclusive or still needs a few extra tests for WW, or how directly the results apply to WW. i.e., is 1 WW cast the same as 6s channel of RoF?
Reply Quote
06/23/2013 02:27 PMPosted by Loroese
You missed Bracers for 6%, bringing it to 97% without DF. Isn't there a weapon that adds 1% to CC?


Corrected stupid mistake.

06/23/2013 02:27 PMPosted by Loroese
If you end up with a consensus of the CM/SA behavior, try to catch me in game or send me an IM at diablofans and I'll try to find a couple hours to update my CMWW simulator, not that anyone really seems to use it, heh. So far it seems like you have pretty conclusive results for RoF at least. Since I've only glanced over it I'm not sure if the testing was conclusive or still needs a few extra tests for WW, or how directly the results apply to WW. i.e., is 1 WW cast the same as 6s channel of RoF?


I'm interested as well, and there are SOME people who still use matlab =)
Reply Quote
@Nubtro: If true, these are VERY interesting results!
You wouldn't happen to have the 5 measurements for each attack speed breakpoint available?

If you had read it thoroughly ;)
You might as well ignore those two columns and just focus on the underlined CM procs/sec values that I got in the game in my tests that you can find here
http://us.battle.net/d3/en/forum/topic/8770117237?page=7#126

It´s on this page, just a few posts above.

The more crit chance, the more reliable the results are. If only I could test more than 75%, but I had to borrow a reflection from a friend for the Sleet Storm tests already (needed 3 APoC sources to be able to channel it against one oponent). Another option would be to run the base 120 sec Archon cooldown. I think the gap between the better and the worse breakpoints would be more distinct.

But yeah, feel free to help me with the tests. They require a lot of patience and clicking hehe. I have to manually count the frames from Archon activation until I start channeling RoF (first damage frame on the target), then click until the next second starts to figure out the exact frame my RoF started (seconds and frames) and afterwards the moment the CD is reduced to zero. I use a freezing/blinding Scoundrel and Diamond Skin to survive the channeling without stopping it. Not an easy task with research gear.
Yep, I had missed you're earlier explanation where you explained what each row meant. I think I now understand what the data you were showing above, so I'll try to apply some stats to the data and see if the differences are significant. ^_^
Reply Quote
TekkZero here you go, more samples. I´ve changed the formatting to make it simpler (hope it is). 15 samples of 1.582 aps and 2.50155 aps (chose the breakpoints that gave me the largest difference before).

75% crit chance Cold Blood 1.58200 aps 120 sec Archon cooldown
(Archon until RoF) - RoF until zero CD start - finish - time - frames = CM procs/sec
(50) 01:38 34:06 32:28 1948 = 2.6704 procs/sec
(41) 00:50 28:42 27:52 1672 = 3.2817 procs/sec
(40) 08:23 45:26 37:03 2223 = 2.2209 procs/sec
(47) 42:01 73:08 31:07 1867 = 2.8313 procs/sec
(61) 17:03 49:50 32:47 1967 = 2.6294 procs/sec
(63) 06:05 33:22 27:17 1637 = 3.3598 procs/sec
(66) 38:07 73:27 35:20 2120 = 2.3651 procs/sec
(48) 41:42 76:13 34:31 2071 = 2.4534 procs/sec
(55) 23:23 55:04 31:41 1901 = 2.7585 procs/sec
(56) 16:50 46:53 30:03 1803 = 2.9623 procs/sec
(527) ------ avg 32:01 19209 (1920.9 avg)
72000 -527 -19209 = 52264 / 19209 = 2.7208 procs/sec avg

75% crit chance Cold Blood 1.58200 aps 100 sec Archon cooldown
(42) 01:48 32:42 30:54 1854 = 2.2136 procs/sec <- lowest
(57) 35:14 64:23 29:09 1749 = 2.3979 procs/sec
(39) 30:04 51:39 21:35 1295 = 3.6031 procs/sec <- highest
(44) 54:01 78:53 24:52 1492 = 2.9920 procs/sec
(81) 43:02 71:42 28:40 1720 = 2.4413 procs/sec
(263) ------ avg 27:02 8110 (1622.0 avg)
30000 -263 -8110 = 21627 / 8110 = 2.6667 procs/sec avg

TOTAL 102000 -790 -27319 = 73891 / 27319 = 2.7047 procs/sec avg

75% crit chance Cold Blood 2.50155 aps 120 sec Archon cooldown
(Archon until RoF) - RoF until zero CD start - finish - time - frames = CM procs/sec
(47) 02:12 27:20 25:08 1508 = 3.7434 procs/sec
(50) 54:20 81:31 27:11 1631 = 3.3838 procs/sec
(54) 59:45 89:08 29:23 1763 = 3.0533 procs/sec <- lowest
(55) 58:54 82:19 23:25 1405 = 4.0854 procs/sec <- highest
(61) 42:38 67:51 25:13 1513 = 3.7184 procs/sec
(75) 53:48 79:59 26:11 1571 = 3.5353 procs/sec
(53) 05:53 30:05 24:12 1452 = 3.9222 procs/sec
(53) 58:07 82:51 24:44 1484 = 3.8160 procs/sec
(55) 10:29 34:54 24:25 1465 = 3.8771 procs/sec
(62) 04:45 31:01 26:16 1576 = 3.5292 procs/sec
(59) 05:41 32:01 26:20 1580 = 3.5196 procs/sec
(624) ------ avg 25:41 16948 (1540.7 avg)
79200 -624 -16948 = 61628 / 16948 = 3.6363 procs/sec avg

75% crit chance Cold Blood 2.50155 aps 100 sec Archon cooldown
(46) 01:27 22:51 21:24 1284 = 3.6371 procs/sec
(80) 02:01 22:01 20:00 1200 = 3.9333 procs/sec
(47) 02:16 23:10 20:54 1254 = 3.7472 procs/sec
(46) 03:09 24:16 21:07 1267 = 3.6993 procs/sec
(64) 01:41 24:05 22:24 1344 = 3.4167 procs/sec
(283) ------ avg 21:10 6349 (1269.8 avg)
30000 -283 -6349 = 23368 / 6349 = 3.6806 procs/sec avg

TOTAL 109200 -907 -23297 = 84996 / 23297 = 3.6484 procs/sec avg

I think the 35% difference between those two breakpoints is quite conclusive.
Edited by Nubtro#2147 on 6/24/2013 2:07 AM PDT
Reply Quote
06/23/2013 02:27 PMPosted by Loroese
You missed Bracers for 6%, bringing it to 97% without DF. Isn't there a weapon that adds 1% to CC?


Sever
Reply Quote
TekkZero here you go, more samples. I´ve changed the formatting to make it simpler (hope it is). 15 samples of 1.582 aps and 2.50155 aps (chose the breakpoints that gave me the largest difference before).

75% crit chance Cold Blood 1.58200 aps 120 sec Archon cooldown
(Archon until RoF) - RoF until zero CD start - finish - time - frames = CM procs/sec
(50) 01:38 34:06 32:28 1948 = 2.6704 procs/sec
(41) 00:50 28:42 27:52 1672 = 3.2817 procs/sec
(40) 08:23 45:26 37:03 2223 = 2.2209 procs/sec
(47) 42:01 73:08 31:07 1867 = 2.8313 procs/sec
(61) 17:03 49:50 32:47 1967 = 2.6294 procs/sec
(63) 06:05 33:22 27:17 1637 = 3.3598 procs/sec
(66) 38:07 73:27 35:20 2120 = 2.3651 procs/sec
(48) 41:42 76:13 34:31 2071 = 2.4534 procs/sec
(55) 23:23 55:04 31:41 1901 = 2.7585 procs/sec
(56) 16:50 46:53 30:03 1803 = 2.9623 procs/sec
(527) ------ avg 32:01 19209 (1920.9 avg)
72000 -527 -19209 = 52264 / 19209 = 2.7208 procs/sec avg

75% crit chance Cold Blood 1.58200 aps 100 sec Archon cooldown
(42) 01:48 32:42 30:54 1854 = 2.2136 procs/sec <- lowest
(57) 35:14 64:23 29:09 1749 = 2.3979 procs/sec
(39) 30:04 51:39 21:35 1295 = 3.6031 procs/sec <- highest
(44) 54:01 78:53 24:52 1492 = 2.9920 procs/sec
(81) 43:02 71:42 28:40 1720 = 2.4413 procs/sec
(263) ------ avg 27:02 8110 (1622.0 avg)
30000 -263 -8110 = 21627 / 8110 = 2.6667 procs/sec avg

TOTAL 102000 -790 -27319 = 73891 / 27319 = 2.7047 procs/sec avg

75% crit chance Cold Blood 2.50155 aps 120 sec Archon cooldown
(Archon until RoF) - RoF until zero CD start - finish - time - frames = CM procs/sec
(47) 02:12 27:20 25:08 1508 = 3.7434 procs/sec
(50) 54:20 81:31 27:11 1631 = 3.3838 procs/sec
(54) 59:45 89:08 29:23 1763 = 3.0533 procs/sec <- lowest
(55) 58:54 82:19 23:25 1405 = 4.0854 procs/sec <- highest
(61) 42:38 67:51 25:13 1513 = 3.7184 procs/sec
(75) 53:48 79:59 26:11 1571 = 3.5353 procs/sec
(53) 05:53 30:05 24:12 1452 = 3.9222 procs/sec
(53) 58:07 82:51 24:44 1484 = 3.8160 procs/sec
(55) 10:29 34:54 24:25 1465 = 3.8771 procs/sec
(62) 04:45 31:01 26:16 1576 = 3.5292 procs/sec
(59) 05:41 32:01 26:20 1580 = 3.5196 procs/sec
(624) ------ avg 25:41 16948 (1540.7 avg)
79200 -624 -16948 = 61628 / 16948 = 3.6363 procs/sec avg

75% crit chance Cold Blood 2.50155 aps 100 sec Archon cooldown
(46) 01:27 22:51 21:24 1284 = 3.6371 procs/sec
(80) 02:01 22:01 20:00 1200 = 3.9333 procs/sec
(47) 02:16 23:10 20:54 1254 = 3.7472 procs/sec
(46) 03:09 24:16 21:07 1267 = 3.6993 procs/sec
(64) 01:41 24:05 22:24 1344 = 3.4167 procs/sec
(283) ------ avg 21:10 6349 (1269.8 avg)
30000 -283 -6349 = 23368 / 6349 = 3.6806 procs/sec avg

TOTAL 109200 -907 -23297 = 84996 / 23297 = 3.6484 procs/sec avg

I think the 35% difference between those two breakpoints is quite conclusive.
Thanks for putting the data (and in an easy to read format)! ^_^ So I ran a few quick stats on the data you just posted. If I can later, I'll run it on your other data too.

The results agree that the difference in the CM procs per second between the two breakpoints is statistically speaking significant. And thus cannot be simply explained by large variance. Here's what I did:
*Step 1: I used a 1-way ANOVA to test for significant differences in the four groups. I found there was significance: p = 2.3551e-006 (groups: 1.582aps with 100s and 120s Archon CD, and 2.501aps with 100s and 120s Archon CD). Note: This test merely tells us something is significant and doesn't tell us, which group is significant
*Step 2: Next, I double checked that the 100s and 120s Archon CD were not fundamentally different from each other with a 2-sample t-Test. The results were not significant: p = 0.9234 (for 1.582aps) and p = 0.8158 (for 2.501aps).
*Step 3: As a final check, I pooled together the 100s and 120s Archon CD data together for each breakpoint, and then compared the breakpoints with a 2-sample t-Test. The result was significant at p=4.3084e-008.

P.S. For those that might be less familiar with stats, a p-value of 0.05 means we have a 5% error of being wrong in concluding there is a difference, while a p-value of 0.01 means we have a 1% error of being wrong. Thus, the p-values I'm reporting here are very significant. The chance of the stats being wrong in concluding there is a difference is VERY low (at least statistically speaking).
Edited by TekkZero#1963 on 6/24/2013 7:49 AM PDT
Reply Quote

Please report any Code of Conduct violations, including:

Threats of violence. We take these seriously and will alert the proper authorities.

Posts containing personal information about other players. This includes physical addresses, e-mail addresses, phone numbers, and inappropriate photos and/or videos.

Harassing or discriminatory language. This will not be tolerated.

Forums Code of Conduct

Report Post # written by

Reason
Explain (256 characters max)
Submit Cancel

Reported!

[Close]