2019 Spring RFL Sign-up and Running Thread: 2nd Round

ffs

Diabloii.Net Member
@T72on1 I complained earlier and unsurprisingly it helped. ;)

Done with Barb now, will go through R2 finds at a later stage and just post readouts in results thread. Same as with Sorc, two of my top 5 dropped within last ~1 hour of the set.

Hope to run a few tests for R3 later or tomorrow. @Grape Based on past RFL rounds I assume you'll fire up R3 sign-up thread only after R2 results? (Just because people are discussing R3 already here...)
 

Babyhell

Diabloii.Net Member
@T72on1 I complained earlier and unsurprisingly it helped. ;)

Done with Barb now, will go through R2 finds at a later stage and just post readouts in results thread. Same as with Sorc, two of my top 5 dropped within last ~1 hour of the set.

Hope to run a few tests for R3 later or tomorrow. @Grape Based on past RFL rounds I assume you'll fire up R3 sign-up thread only after R2 results? (Just because people are discussing R3 already here...)
Ok so if it works for everyone, hey ho, I'm complaining! Hey big Z God, you hear that, i'm complaining!

4hrs to go...now will that mean hours or high runes ?

+1 on the R3 thread creation need, i'll be training from tomorrow night or monsday, will try hammerdin (and maybe jzon or blizzard sorc) in CS and AS :)
 

Grape

Diabloii.Net Member
Done, couple hours today, and plan to run many more! Shooting for total of 15 hours (in this batch) before the night. Maybe I'll start the second one if I hit 20 hours before the deadline. Will continue in 1.07 because you never know...

Few finds today:
Screenshot1139.jpg
First Ko rune of the batch. Turned out to be a pretty sweet ring for a min damage oriented char, like Avenger etc. Not useful for questing though.

Screenshot1142.jpg
Put this into use right away. Still have to test whether it's best to have full FRW/add inv though. Could be as I'm running through the docks every game, and FRW doesn't have diminishing returns in 1.07.

Screenshot1143.jpg
Isn't this perfect (for its level req)? Haven't checked yet.

@T72on1 Hope to run a few tests for R3 later or tomorrow. @Grape Based on past RFL rounds I assume you'll fire up R3 sign-up thread only after R2 results? (Just because people are discussing R3 already here...)
Yeah, I think I'll stick to that schedule. No problem for me that there is some R3 discussion already! Heck, feel free to chat about whatever here :p
 

Apila

Diabloii.Net Member
I need to drop out of this. I have a tiny guest this weekend and next week, in addition i have quite a lot of work. Maybe i'll have more time when the next round comes around.

Good luck to everyone else.
 

ffs

Diabloii.Net Member
@Gripphon Ah sorry I misunderstood, I thought your ~225 kills/minute number was in CS. That would have been a bit shocking in comparison to ~145 with Blizzy. In AS that number makes more sense. Nice vid! I don't have the craziest gearing options for Fire Sorc, but will try with her nonetheless, looks like a lot of fun. I have a feeling I'll end up running with Java again though.

I recorded another 10 runs and counted 419 ghosts and 540 regular kills, i.e. a combined 227 kills per run. That number is very close to both your count and what I counted in 2018, so I'm beginning to feel confident that's about right regarding AS density. Will review more batches of runs though. Somehow this one was about 10% slower than my initial tests with the new setup, at 62.7 seconds average, resulting in 217 kills per minute. She managed 55s last year and in initial runs, so that's my benchmark. Maybe I need some additional damage after all, so there's some gear tweaking to be done.

@ffs
Have you considered going with 20 ias for AS javazon? Easy 99 fcr and 100% pierce, unless such pierce is not needed
Yes, tried last year. It works well, but was a little slower than my standard setup.

If anyone is interested – I counted kills for Pit Nova Sorc as well. (Yes, I had an awesome 1-2 hours of running and counting :) gotta love D2.) With her I got an average of ~110 kills per minute, though that was with a focus on boss packs and even including broken LI ones (using CM though). She has a decent ~7.3 boss density and my idea was to have a shot at good S&Us on the side. But the difference in RF efficiency is too big obviously. With focusing only on big groups and ignoring LI bosses/minions, maybe she can get to ~150 per minute or something, but in any case it will be too far behind AS.

EDIT: Pit Java I already tested last year. With my "camps only" approach she got some 160-170 kills/minute there, but that includes Tamoe camp, fewer boss packs than Nova Sorc, and again fewer kills than AS of course.

So yeah, looks like it's going to be an AS fest.
 
Last edited:

CaseyJones

Diabloii.Net Member
Well, I've got my plaguafuryzon to the beginning of act 3 hell, really feeling undergeared now and have died a couple times in the last few hours of play. Think I should focus on some levels and gearing for round 3 and sit round 2 out. Currently level 71. Having to use plague javelin through most of act 3 since so much is light immune.
At least I can run AS for round 3!
 

Gripphon

Diabloii.Net Member
It's not over yet. By checking these numbers from RFL 2017 on my CS sorc:

Runs done: 270
Runes farmed: 159
Avg. time: 69
Runes per run: 0.589
Runes per minute: 0.512

By my math this equals to ~165 creeps per minute killed or ~124 p5 cows. This sorc can be improved though, only question is by how much. Exciting, isn't it. :cool:
It was 2 seal running sorc. Now if I make sorc who can 1 shot ghosts and focus only one seal plus that wing... we shall see. Of course I assume I didn't have super exceptional luck with rune drops.

EDIT
I see NanoMist also post his number of 0.5 runes per minute which is ~161 creeps a minute. Don't know why did the 145 number come up for him. Video analysis?

I don't really believe CS can beat AS, but I don't think it will be far behind either.
 
Last edited:

Gripphon

Diabloii.Net Member
I don't really trust video analysis method. First, it is done over too few runs meaning chance those runs aren't average in density is very high. Secondly, counting itself easily ends in +- 10 creeps/min of error. On the other hand runes were counted on hundreds of runs and there is only a question whether person counting them was lucky and how much. I'm pretty confident rune counting method does end up in +-10 creeps/min confidence interval while for video analysis I think margin of error is easily bigger. In the end both of our data suggest we kill more creeps than what video analysis have shown. I don't think that is just random coincidence that we both ended much too lucky with rune drops.

So, I give much more importance to our runes/minute number than what we counted on the video. I'll apply same method of comparison between CS and AS during round 3.
 

ffs

Diabloii.Net Member
Counting thing never ends. :D

@Luhkoh and @drmalawi concluded in the case of cows that you need >6000 runs to get even near 95% confidence range of +/- 10 kills per minute with drop count method. Far more data than an RFL set, and certainly more short-term samples. With that information I have even more trouble trusting such estimates.

Counting can't overestimate kills (assuming it is done diligently with rewinding when in doubt), while with drop count method that is not the case. It's true you can get a body count that is too low though. I noticed a couple of off-screen kills I didn't catch initially, when briefly teleporting back to check drops. But the drop count method cannot solve that issue either – if you don't catch the off-screen drop, it is not included in your data either.

Re average density, I don't see any signs of huge variance. All counts so far resulted in similar numbers. I plan to count 50 runs worth of kills in total, then I'll have no trouble treating the overall average as sufficiently accurate. From there I will just go with run times.

But I'll again count drops during R3 as well. Counting method is not perfect either and it can't hurt to have an additional number.
 

Gripphon

Diabloii.Net Member
Our goal isn't to hit 95% confidence interval because that's impossible, our goal is to compare two areas in a best way we find suitable with enough data to support our results. I just wanted to say that video counting method across 10 or whatever runs is insignificant to reach such goal. 50 runs on the other hand is much more significant and extrapolating results from it is much more precise. It isn't anywhere near 95% confidence interval, but for our cause it's precise enough. Same as is rune counting method.

From our results we may not precisely say by what % is one area better, but conclusion one area is slightly better, or clearly better, or much better, or similar to other area is good enough.
 

ffs

Diabloii.Net Member
@Gripphon Agreed!

In other news, I experimented with other runs some more. Possible candidate is Javazon running Eldritch/Shenk or Shenk only. I just go in, throw LF once, check drops and move on. Time per kill is in the ~0.5 second range for these runs, so it makes no sense whatsoever to chase leftovers (nor finish off Shenk/Eldritch if they survive). Only big groups warrant engaging again. You usually get the 10+20 minion kills from both bosses pretty reliably, plus there are surrounding mobs frequently, which adds to the body count. What I got so far:

Eldritch/Shenk: About 15-17 second runs and ~35 kills per run (~125-140 kills per minute).
Shenk only is better: 9-10 second runs and ~24 kills per run (~150-160 kills per minute).

It is worse than AS yet again, but not bad for first testing, and Zod can drop there. Decisive question to me at this point seems to be the impact of maps. Haven't played around with rerolling yet – it might be possible to get more density around the bosses (or on the stairs) more reliably by doing so, in which case it could be a good deal more efficient.

Runs are rather boring though. :D

EDIT: Only had some 30 minutes for additional tests today, but there's definitely potential here. Couple of further observations:
  • CtA is not required – Harmony switch can be used to reduce run times further. Not casting BO saves time as well.
  • There are definitely better and worse maps, and I did the first tests on a bad one. Since I wanted to keep the maps on my Javazons for now, I used another char just to see different maps. Only rerolled for 10 minutes, but already got a map with a boss pack spawning with Shenk roughly 80% of the time. Seems entirely possible to get ~200 kills per minute or more with a suitable map – at the very least close to AS level and including Zod chances
  • One big problem I realized though: Picking behaviour. These runs are extremely short – spending 1.8 seconds to pick up one single item means losing a whopping 20% efficiency on that particular run. Definitely bad if planning to get many non-qualifiers. ;)
 
Last edited:

Babyhell

Diabloii.Net Member
One hour left on this second batch. No unique bill for now @DiabloTwoinDC !

Except for a big surprise, first set will be the best, and this time the margin would not be 0.6 points iirc.

Let's drop a crafted Zod !
 

sir_cyclops

Diabloii.Net Member
I was out for a week due to my laptop charger dying. Back in action now but only 11.5 hours in. Might be hard to finish. Also on a 5 hour dry streak -.- #complaining
 

ffs

Diabloii.Net Member
Figured those Java Shenk runs may be suitable to go with the weapon switch Teleport glitch:


Setup:
Griffon's (IAS/MPK)
Highlord's
Titan's
Enigma
3/20 gloves
4-facet Monarch (+18/-18)
Arachnid Mesh
SoJ
SoJ
Silkweave

Since teleporting is based on IAS when using the weapon switch trick, I went for 55 IAS with HLW. Using Arachnid Mesh even though I don't need its FCR, because somehow the glitch didn't work with a Razor setup, for me at least. Ravenfrost is not required either, so double SoJ it is.

Tried this approach for fun mainly, of course any normal setup with FCR works and may even be better – because it's not easy to pull off the glitch consistently. I screw it up often and then teleport slowly and lose time.

At this point I put some effort into it and didn't surpass 155 kills/minute in any batch yet. Run times can be improved, but only so much. Crucial thing is getting consistent additional density near Shenk – I'm not sure if that's possible to an extent to get close to AS numbers.
 
Top