R6 generator testing / discussion

Monk
03/01/2017 07:54 PMPosted by Mazarini
Big problem with the game design atm is paragon points redistribution during rift.
To be competetive u will have to do what most players at the
top of LeaderBoards do - Go all out DPS build : skills and passives ( no Harmony) and gear, start the rift with 2-3mil hp and at RG switch to 600k hp (or more in case of dangerous RG). Either u do this or u loose 1min+ on GR 117 and higher.

This of course is currently being done by different players in different ways:
-by hands ( when u pull last pack of mobs go to empty corner or previous floor).
-mouse or keyboard licensed macro drivers.
-3rd party macro programm.
-and what is worst in current situation - bot programm , writing directly to memory , fail free.
Not only this provokes to unhealthy competition but also takes away a big part of planning your character's stats balance. This also is being used for solo pushes. Needs to be fixed asap.


Wow, lame way to "cheat" the LB.
03/01/2017 03:10 PMPosted by BaesedGod
I still don't understand how the discussion now is just between {WFF}, Q, SG gen or {Q}, WFF, SG gen as the best RG killers. Based on Davlok's STDPS charts, isn't TC the highest dps - if so what would be the combination of skills you would use with TC? Of course I'm going off the assumption we're only talking about RG fights since this is what gen monks are in the meta for.

Your total R6 damage is a function of your spirit level (Shenlong 2pc bonus), stricken stacking (gen desync affects this), and RNGing an RG without a single add.

TC is terrible > 1 target, FoT in general looks like the worst stricken stacker, and TC has much worse spirit generatation compared to Quickening... and we can only pick one rune per generator. But if someone can find a way and show up on the leaderboard, I'm sure everyone else will copy it! ^_^

03/01/2017 07:54 PMPosted by Mazarini
Big problem with the game design atm is paragon points redistribution during rift.

Yea, not sure why they allow Paragon changing in GR if equipment is locked.

03/01/2017 07:54 PMPosted by Mazarini
I also have one question-
Situation: we have channeling pylon for RG. As the dashing strike animation speed (velocity or whatever, call it as u wish) afaik is higher than any generators - may it be possible to stack stricken with it faster by just using DS during channeling?

Interesting question, I was bored enough to test it out on the PTR:

https://gfycat.com/CostlyGrotesqueHeron

If you look real fast, you will see DS deal the same damage 3 times in a row before increasing 1x Stricken stack. (12772x3, 12863x3) Oddly each DS was slightly different frames in between, so might be a distance delay somewhere along the way as well. So it looks like no, it isn't worth it to spam DS even if the cast speed is faster than a generator.
The stricken ICD for DS is probably based on your sheet APS or something.
TC is terrible > 1 target, FoT in general looks like the worst stricken stacker, and TC has much worse spirit generatation compared to Quickening... and we can only pick one rune per generator. But if someone can find a way and show up on the leaderboard, I'm sure everyone else will copy it! ^_^


Yeah I figured, you're pretty much banking on the RG to RNG no adds. However, there are a couple people on NA leaderboards using TC: Sc hlyy(TC, Foresight, BF) & Sc fanatic(TC, WFF, SG),

Fun fact, the top group just pushed 119 this morning running a physical monk with Q, SG, WFF.
TC is terrible > 1 target


It is definitely worse but not as terrible as one could think judging from tooltip : first 2 strikes are single target anyway and AoE of third strike is
very very small.
And of course once RG summons minions monk should stop CCing them and barbarian needs to rageflip them away.
03/02/2017 03:41 AMPosted by Davlok
https://gfycat.com/CostlyGrotesqueHeron


Thank you for testing.
03/01/2017 02:24 PMPosted by nand
I'll try reporting a bug.

A lot of flying dragon testing later, here are my conclusions:

1. The simulator on http://freakinsweetapps.com/flying-dragon-up-time-calculator/ is more or less accurate. (It breaks down at very high APS because it doesn't simulate APS breakpoint rounding)

2. d3planner's FD uptime is way, way, way too high. Even at 1 APS it reports 40%, but reality is more like 25%.

I've tracked down a couple of possibilities as to what the bug might be, but there are still some tests that are open. Here's what I learned so far:

1. Proc coefficient of the ability does not matter at all. Quickening and Static Charge have the exact same FD uptime, but quickening has twice the proc coefficient of static charges.
2. Hitting a mob vs hitting the air does not seem to matter.
3. The duration is 7s, and the ICD is 5s. (This confirms what we already know)
4. Flying dragon proc rate scales with your gen APS buffs (R2, Alacrity) and sheet APS
5. Flying dragon proc rate seems to be normalized to APS. With APS 1.0, even though you're punching about 1.34 times per second on average (due to the innate attack speed buff that generators get), you're still proccing flying dragon as if you had exactly one attack per second.

What I haven't been able to test yet is whether the FD uptime depends on the ability (e.g. fists of thunder vs crippling wave), since they have different innate APS buffs and also different frame timings. Further tests would be welcomed.

At any rate, I posted everything I was able to figure out over at https://github.com/d07RiV/d3planner/pull/190
I also figured out why crippling wave stacks stricken with every hit, while the other generators do not:

The crippling wave damage is near the beginning of the animation, rather than near the end. This means that the third hit does damage, procs the ICD, and *then* plays the long animation; rather than playing the long animation first and then dealing the damage + proccing the ICD.

So the delays from hit to hit, in practice, are:

1st hit -> 2nd hit = short (~260ms)
2nd hit -> 3rd hit = short (~260ms)
3rd hit -> 1st hit = long (~380ms)

For other abilities, the damage is done closer to the end of the animation, which means that the delays are

1st hit -> 2nd hit = short
2nd hit -> 3rd hit = long
3rd hit -> 1st hit = short

But since 3rd hit -> 1st hit has the longer ICD, this means the 1st hit gets skipped and doesn't apply stricken.
I tested a bunch more Flying Dragon mechanics, and also more or less rooted out the source of the d3planner FD uptime bug.

1. Testing FD uptimes is a real, real pain. Even over a 10 minute test, your results can vary anywhere from 60% uptime to 80% uptime depending on how lucky or unlucky you got.

2. FD uptime seems to depend on the generator. For FoT and WotHF, I consistently got something around ~70% uptime (at 2.55 sheet APS), but for DR and CW it was lower (58% for DR, 52% for CW).

While 58% for DR sort of falls into the “eh, could have been bad luck” category, 52% FD uptime for CW over a 10 minute test is far lower than I would pin solely on luck.

So in other words, CW-based specs are sort of automatically worse than other specs solely based on poor flying dragon uptimes. This effect is not captured or simulated on d3planner, and I'm not sure where it would even come from.

3. d3planner has such insanely high FD uptmes because they use a “averaged out PRNG”. Basically, on d3planner, if you have a 10% crit chance, every 10th hit will be a crit - guaranteed. So for FD, with a 0.04% proc chance, every 25th hit will proc FD - guaranteed. In other words, if you can fit 25 hits into the 2-frame FD reproc window (~5 FpA average) then you would get 100% FD uptime on d3planner, but not in reality - in reality you only have a 63% chance of reproccing FD before it runs out under those same circumstances.

Editing the d3planner source code to make it use a real PRNG (like Math.random) makes the anomalous FD uptime immediately go away.
03/03/2017 07:38 PMPosted by nand
Editing the d3planner source code to make it use a real PRNG (like Math.random) makes the anomalous FD uptime immediately go away.

With this edit in place, I simulated a bunch of trial runs. The results were actually very interesting on a purely visual level. Basically, you can divide specs into high variance (high risk - high reward) and low variance.

The high variance specs were those with FitL instead of BoH; the low variance specs were those with BoH or {Q}. This is because they can be consistently full spirit heading into the main element.

On average, {WFF} performed better than {Q}, by like 1-2% at most. In terms of consistency, {Q}+BoH was the most consistent, followed by {WFF}+BoH.

Here are the raw numbers:

{WFF} + FitL:
61.059
55.229
57.039
56.314
61.684
55.542
56.116
58.538
52.185

{WFF} + BoH:
57.559
58.869
56.207
58.500
58.115
58.599
54.913
58.045
55.306

{WFF} duogen:
58.196
52.602
57.911
56.169
59.818
55.408
59.076
57.351
58.727

{Q} + BoH:
56.200
57.654
55.552
55.971
55.554
56.517
55.691
55.562
57.385

{Q} + FitL:
54.287
58.094
55.000
58.239
58.824
56.806
55.176
57.919
55.983
54.048

Summary:
{WFF} + FitL: 57.078 ± 2.967
{WFF} + BoH: 57.346 ± 1.488
{WFF} duogen: 57.251 ± 2.225
{Q} + BoH: 56.232 ± 0.804
{Q} + FitL: 56.438 ± 1.770

So this pretty much confirms what we already know, namely that:

1. In general, {WFF} vs {Q} is so close that it's basically a question of your gear, nothing else
2. BoH is more consistent than FitL, but FitL has higher potential if you get godly FD procs. It has the lowest individual score and the highest individual score.
3. {Q} is easier to play than {WFF}+BoH, so that's a factor speaking for it

FD uptime was usually in the ~75% range, not the absurd 93% reported by d3planner. Fixing that PRNG bug really helped.

Now that we've got this sorted out, I'm ready to re-evaluate {TC}.
Not a surprise at all. One of the first things I noticed when I switched from Mangle to CW solo in the last era with Inna gen was that my FD proc rate went way up. It was overall a much more stable build. Maybe that might make cold gen more viable solo as well as with groups.
03/03/2017 08:07 PMPosted by nand
Now that we've got this sorted out, I'm ready to re-evaluate {TC}.


First of all, I fixed the bug where TC applied its extra area damage during the first two hits for single target, leaving only the third hit in place.

{TC} / BF / CW / BoH:
56.403
56.461
54.777
53.568
54.455
56.500
55.651
55.871
54.617
56.484
51.779
52.849
57.097
53.981
56.164
-> 55.111 ± 1.556

Note: This spec was pretty bad, as I sort of suspected based on the fact that we're missing out on the stupid high weapon damage scaling on WFF and SG. The 15% IAS buff did not make up for it, nor did the 10% additive buff.

It's worth mentioning that I was using unrealistically perfect gameplay conditions: I never interrupted a 3-hit combo to recast BW or BF, I never applied it more than necessary (stopped casting BF as soon as it crit), and I applied it exactly on the cooldown / duration. Slacking on those made the spec even worse. It's the most impossible spec to play out of the lot, for sure.

Another huge downside of the spec is that you lose a lot from the lightning, physical and fire being so far apart. The Q/{WFF}/SG/BoH combo works so exceptionally well because it has great synergy: You can BoH either during cold or after cold, and doing it after cold gets you free BoH for the fire element as well, which is no joke. You also get a Q phase right before, so you're always going into cold with full spirit. It's incredibly consistent, and my personal favorite by far.

{TC} / WFF / SG / BoH:
55.537
61.115
58.745
55.227
56.598
59.897
54.017
58.265
59.666
56.057
59.949
56.841
57.167
60.703
58.627
56.262
59.816
56.427
-> 57.829 ± 2.084

With the bug fixed, this spec wasn't performing nearly as well as people gave it credit. It was pretty similar to {WFF} + FitL; in that you just didn't get any good consistency out of it. The lack of quickening really hurts.

That said, the gameplay I used for this was fairly simple: BoH shortly before lightning element (if spirit low or FD not active), or during lightning element (if spirit low or FD runs out). Basically, just make sure BoH is active during lightning if necessary. Nothing fancy.

Further caveat: I haven't incorporated any tweaks to bane of the stricken. It's possible that the faster stricken of crippling wave pushes that spec slightly higher. It's also possible that the optimal way to play these specs is to switch between generators every frame at high boss HP. This would in principle prevent the third attack from ever landing, which means more stricken stacks per second. That said, the loss of damage is so big that I'm not sure if it's worth it.
Btw, in case these numbers don't mean much to you, it's useful to convert them back to time. For example, at 20qd HP, a DPS figure of 55t translates to 363 seconds = 6:03, and 60t translates to 333 seconds = 5:33.

So basically, the difference between the best and worse of these specs is something like ~20s clear time, for 5-6 min RG fights.
I must preface my question with a compliment. WOW! nand, you have done one hell of a job explaining the mechanics of proper R6 play. I am very, very appreciative of your effort and information shared.

If you could, please take a look at my R6 monk,

https://us.battle.net/d3/en/profile/Bulvai-1755/hero/54224110

I use it for 4 man group play exclusively at this point. My questions are thus:

1)is it worth dropping Serenity for something else such as Searing Grasp if tri-gen is definitely the way to go?

and more importantly:

2)What would be an optimal rotation during RG fights vis a vis popping DS, BoH, and WFF vs Quickening during the various CoE elemental cycles?

I read your rotation on the previous page but I was hoping for clarification if I'm missing something.

Any advice you could provide would be greatly appreciated and once again, outstanding job!
1)is it worth dropping Serenity for something else?

I mean that's up to you. Serenity is a defensive choice, and you could pick up an offensive choice instead. Simply picking up Blinding Flash as a third DPS spell boosts your damage, and swapping that out for Searing Grasp instead of blinding flash boosts your damage even further.

So if you want to deal more damage, then you need to drop serenity for another offensive skill. That said, the game isn't entirely about damage - it depends on what you need. Spending 10s dead during an RG fight is worse than losing out the third generator.

I read your rotation on the previous page but I think it was for a tri-gen setup so I decided to post this interrogative separately.

For duogen I would just always pop BoH during the mid-way point of the physical cycle, and otherwise play it out as I suggested. That way you always get BoH during cold and the majority of physical. If you want to get a more specific answer I would recommend using d3planner to simulate some skill rotations (the skill condition system is pretty specific and allows you to target individual CoE buffs).
Thanks for the prompt reply!! I look forward to reading more of your extremely informative posts.
Hi , nand. Could you plz test this setup : WoHF: HoL, Q, CW:BW ? And also this http://www.diablofans.com/builds/88221-new-s10-generator-monk-boss-killer-out-with-cold
03/19/2017 01:43 PMPosted by Garfunkel
Hi , nand. Could you plz test this setup : WoHF: HoL, Q, CW:BW ? And also this http://www.diablofans.com/builds/88221-new-s10-generator-monk-boss-killer-out-with-cold


The second spec cannot fit Serenity (tri-gen and no Quickening requires BoH for spirit) and IMO REQUIRES paragon swap at RG (full Vit to full Dex). This is especially true on Season where survival is low due to worse gear and less paragon.

Tclap combo gen needs eval, but I'd not recommend the second spec for anything except extreme push with paragon reassign. The fact that the dfans guide does not mention the paragon reassign makes it a newb trap (more than newb trap actually, since even seasoned players are not likely aware of this retarded mechanic).

A duo gen version with just Tclap and Blazing (+ Serenity) is more realistic. That's probably what I'd test and just figure the trigen version w CW debuff does about 7% more edps with bonus of about 3% more WD edps.
I've always been aware of paragon reassignment throughout the grift, just never really thought of using it in a 4p party like that. Just mainly noticed it because of that damn plus sign in the middle that I kept clicking on.
03/20/2017 02:53 PMPosted by StoleOwnCar
I've always been aware of paragon reassignment throughout the grift, just never really thought of using it in a 4p party like that. Just mainly noticed it because of that damn plus sign in the middle that I kept clicking on.


It's only a thing in the current meta for fragile RG killer like Gen monk. They don't do any dps during rift, and in fact don't want to kill trash the WD will killl. The need to be alive to buff WD and alive to pull. The paragon reassign literally gives them a 600-1000% EHP buff during GR clear portion and lets them be lazy. The removed dex is literally at 0 cost.

Incidentally, since I think most gen monks run with at least some vitality to be of more use during clear, the reassigner also gains edps during the RG fight, where only minimal EHP is required. (They can move all vit to dex whereas the monk that doesn't reassign, probably has some extra vitality they are stuck with at RG).

Join the Conversation

Return to Forum