Inserter throughput

Place to discuss the game balance, recipes, health, enemies mining etc.
User avatar
disentius
Filter Inserter
Filter Inserter
Posts: 694
Joined: Fri May 12, 2017 3:17 pm
Contact:

Re: Inserter throughput

Post by disentius »

@quyxkh:
Proof of concept: you can get max troughput with belts [EDIT] and chests. :mrgreen:
pic

quyxkh
Smart Inserter
Smart Inserter
Posts: 1027
Joined: Sun May 08, 2016 9:01 am
Contact:

Re: Inserter throughput

Post by quyxkh »

haha awesome. my compactness fetish is a blinkin disease, I left one whole side of the assembler unused … for what?

Did I get the math that wrong too? What's that 2240? 8/.5*1.4*60, cs8, .5s recipe, 1.4 productivity, 60 seconds, 1344 items/min? ,,, ah, the math works for a 100-sec interval, is that it?

User avatar
disentius
Filter Inserter
Filter Inserter
Posts: 694
Joined: Fri May 12, 2017 3:17 pm
Contact:

Re: Inserter throughput

Post by disentius »

Oops sorry. is i/s X 100 so 22.40 i/s
Your math is spot on: 1344/60 = 22.4
Vanattenvelds running averager works well this way)

nafira
Fast Inserter
Fast Inserter
Posts: 107
Joined: Fri Mar 16, 2018 12:20 pm
Contact:

Re: Inserter throughput

Post by nafira »

disentius wrote:
Wed Feb 05, 2020 7:40 pm
Pickup speeds from belt in 0.18.3:

Image
meh... edit, I understood :

1 is known to be slow (swing time and lane speed does not match, it's like getting item 1 by 1)
2 is slower than 3 because it's feeding only on one lane (22.5i/s but with less stacking)
3 is a bit faster because items are stacked at maximum letting no other choice like in the second one (22.5i/s)
4 is very fast because it "virtually" is a 45i/s lane, allowing for a full fetch.

quyxkh
Smart Inserter
Smart Inserter
Posts: 1027
Joined: Sun May 08, 2016 9:01 am
Contact:

Re: Inserter throughput

Post by quyxkh »

Yeah, arm extension is surprisingly slow, it's one of Factorio's remaining puzzles. Most people never even notice it at all.

User avatar
disentius
Filter Inserter
Filter Inserter
Posts: 694
Joined: Fri May 12, 2017 3:17 pm
Contact:

Re: Inserter throughput

Post by disentius »

nafira wrote:
Thu Feb 06, 2020 6:37 pm

1 is known to be slow (swing time and lane speed does not match, it's like getting item 1 by 1)
2 is slower than 3 because it's feeding only on one lane (22.5i/s but with less stacking)
3 is a bit faster because items are stacked at maximum letting no other choice like in the second one (22.5i/s)
4 is very fast because it "virtually" is a 45i/s lane, allowing for a full fetch.
I would put it slightly different:
1 can not profit from inserter chasing items, all the others can.
2 is slower than 3 because 2 goes: retract/pickup/turn/extend/drop, and 3 goes: pickup/turn/drop. and extend/retract is relatively slow
4 is faster than 3 because it takes full advantage of the chasing mechanism over two lanes, and three can only use chasing on one lane.

nafira
Fast Inserter
Fast Inserter
Posts: 107
Joined: Fri Mar 16, 2018 12:20 pm
Contact:

Re: Inserter throughput

Post by nafira »

disentius wrote:
Thu Feb 06, 2020 8:26 pm
nafira wrote:
Thu Feb 06, 2020 6:37 pm

1 is known to be slow (swing time and lane speed does not match, it's like getting item 1 by 1)
2 is slower than 3 because it's feeding only on one lane (22.5i/s but with less stacking)
3 is a bit faster because items are stacked at maximum letting no other choice like in the second one (22.5i/s)
4 is very fast because it "virtually" is a 45i/s lane, allowing for a full fetch.
I would put it slightly different:
1 can not profit from inserter chasing items, all the others can.
2 is slower than 3 because 2 goes: retract/pickup/turn/extend/drop, and 3 goes: pickup/turn/drop. and extend/retract is relatively slow
4 is faster than 3 because it takes full advantage of the chasing mechanism over two lanes, and three can only use chasing on one lane.
Thx for putting it right !

So, the difference between 2 and 3 is pure animation ? It shouldn't impact the throughput since it can't empty a lane.

User avatar
disentius
Filter Inserter
Filter Inserter
Posts: 694
Joined: Fri May 12, 2017 3:17 pm
Contact:

Re: Inserter throughput

Post by disentius »

@Nafira:
No, it is a real game mechanic.
More info on the wiki:
https://wiki.factorio.com/Inserters

nafira
Fast Inserter
Fast Inserter
Posts: 107
Joined: Fri Mar 16, 2018 12:20 pm
Contact:

Re: Inserter throughput

Post by nafira »

disentius wrote:
Thu Feb 06, 2020 9:22 pm
@Nafira:
No, it is a real game mechanic.
More info on the wiki:
https://wiki.factorio.com/Inserters
I know by fact that it is, but why is it that way ? I don't think it's legitimate, and that's why people are pointing that's it's very slow.
I know it's about precision, but this going too far in certain situation ...

quyxkh
Smart Inserter
Smart Inserter
Posts: 1027
Joined: Sun May 08, 2016 9:01 am
Contact:

Re: Inserter throughput

Post by quyxkh »

Did I fail to communicate there, or what?

I like it that the game poses puzzles. I like it a lot. The puzzles are the point, everything else is there either to motivate the puzzles or just be rewards and distractions because those are worth loving too. Subtleties in behavior that you don't even notice until you try to get absolute peak performance, then there's tradeoffs everywhere you turn and you can never have it all, but if you try some time, you get what you need, that's what makes this game so good.

nafira
Fast Inserter
Fast Inserter
Posts: 107
Joined: Fri Mar 16, 2018 12:20 pm
Contact:

Re: Inserter throughput

Post by nafira »

quyxkh wrote:
Fri Feb 07, 2020 6:39 pm
Did I fail to communicate there, or what?

I like it that the game poses puzzles. I like it a lot. The puzzles are the point, everything else is there either to motivate the puzzles or just be rewards and distractions because those are worth loving too. Subtleties in behavior that you don't even notice until you try to get absolute peak performance, then there's tradeoffs everywhere you turn and you can never have it all, but if you try some time, you get what you need, that's what makes this game so good.
I agree on the baseline. Getting the most of the system is very rewarding and it shouldn't be easy.

My point is : where is the threshold between "knowing well the game and building optimized stuff" and "running a math simulation".
Because I can absorb :
Inserters not grabbing items because item chasing them, no problem at all, I never do that
Inserters picking on one lane is limited
Inserters picking on a virtual "45 i/s single lane"

What I can't handle is chasing a needle in haystack like "since the inserter natural extension is the opposite lane, you have a difference of 4% between picking up in front of it and the opposite lane".
This is insanely absurd.


I repeat, I have absolutely no problem with the "45 i/s single lane" I just learn about, and I'm glad I'm still learning.
But this 4% is just useless since it's not meant to reach a peak. Don't you agree ?

kitters
Inserter
Inserter
Posts: 34
Joined: Tue Jul 23, 2019 4:48 pm
Contact:

Re: Inserter throughput

Post by kitters »

I've read everything you wrote.
O, the problem isn't new. Sadly, that devs decided against simplification of inserters.
Throughput reaaly depands on a lot of stuff. I've learned a lot about mechanics and a lot of ways to load from belts, especially "Vegemeister unload trick", thanks, disentius. (i mentioned 6 beacons instead of 5, but thanks).
The problem of these methods you mentioned is that they take a lot of additional space (it's limited because of beacons' range) I was taking just from undeground belt, and it has really low throughput and two inserters are not enought. It's like a need to accept this stuff with inserters and find a compromis.
But adding another one inserter (3 in sum) must rectify the lack, but it didn't. Inserters just begin to stay inactive most of a time.
It's another problem. Maybe it's because of lack of buffer of items inserter can put into assembler. The buffer is The ingredients for 1 craft in addition to the ingredients for the number of crafts that can complete during one full inserter swing; but at least the ingredients for 2 crafts and at most the ingredients for 100 crafts.
But inserter inserter is not only swinging but collect items from the belt most of the time. And buffer have enough time to be exhausted.
What do you think about this?

User avatar
disentius
Filter Inserter
Filter Inserter
Posts: 694
Joined: Fri May 12, 2017 3:17 pm
Contact:

Re: Inserter throughput

Post by disentius »

6 beaconed and 4 prod moduled green circuit factories on expensive need:

INPUT:
- 34 Iron plates/sec -> 34/45 = 0.756 blue belts, 2 assemblers
- 136 copper coils/sec -> 136/45 = 3.022 blue belts, 5.714 assemblers
- 24.285 copper plates/sec -> 24.285/45 = 1.079 blue belts

OUTPUT:
- 23.8 green circuits/sec


With creative belt spaghetti It is certainly possible (and ugly:)
I did not find a good way to expand 2 copper belts to 3 copper belts without losing troughput: all copper coil belts need to be fully saturated for this setup to work.
PoC- GC expensive 6 beacons belts.gif
PoC- GC expensive 6 beacons belts.gif (4.92 MiB) Viewed 4831 times

kitters
Inserter
Inserter
Posts: 34
Joined: Tue Jul 23, 2019 4:48 pm
Contact:

Re: Inserter throughput

Post by kitters »

By the way, disentius, did you find a way to load 4 belts per wagon with balance on chests?
viewtopic.php?p=456201#p456201
I couldn't find the answer.

kitters
Inserter
Inserter
Posts: 34
Joined: Tue Jul 23, 2019 4:48 pm
Contact:

Re: Inserter throughput

Post by kitters »

I didn't ask you to try to simulate my factory or try to find a solution of supplying assemblers.

I just pointed on theoretical situaton, when inserter stops while working despite not supplying an assembler.
I even pointed on possible reason - the buffer is calculated using rotating time, less than one period. It's stops refusing to overload a buffer, but a buffer is exhausted before its next refilling.
And I faced this situation.
Do you want to repeat the situation or what?

User avatar
disentius
Filter Inserter
Filter Inserter
Posts: 694
Joined: Fri May 12, 2017 3:17 pm
Contact:

Re: Inserter throughput

Post by disentius »

Aha. I missed the "theoretical" bit.
Yes, it is possible to empty a buffer before an inserter has picked up from a belt, and refilled the buffer. Ful bonus stack inserters use 37 ticks to deliver the next load.
And yes, Vegemeister did find a balanced solution for 4 belt per wagon loading, here is the blueprint:

kitters
Inserter
Inserter
Posts: 34
Joined: Tue Jul 23, 2019 4:48 pm
Contact:

Re: Inserter throughput

Post by kitters »

For some reason i can't copy that blueprint.
But can copy other prints in this threadю

User avatar
disentius
Filter Inserter
Filter Inserter
Posts: 694
Joined: Fri May 12, 2017 3:17 pm
Contact:

Re: Inserter throughput

Post by disentius »


kitters
Inserter
Inserter
Posts: 34
Joined: Tue Jul 23, 2019 4:48 pm
Contact:

Re: Inserter throughput

Post by kitters »

thanks a lot.
God, it's genius.

RinDiddy
Inserter
Inserter
Posts: 24
Joined: Sat Sep 22, 2018 1:13 pm
Contact:

Re: Inserter throughput

Post by RinDiddy »

Kitters,

From my experience stack inserters are not the answer when you need fast input into a machine, especially one that makes circuits. I've done something similar to that ( my concept was to create as many free circuits as possible in the fastest way ).

The main problem is stack size. The inserter will attempt to "grab" up to it's limit before placing items into a machine. When speed is of the essence, moreso with gobs of items on fast belts, this is not the answer. You can resolve this by going into the inserter and setting its stack limit to 1. You're probably better off with just a standard fast inserter for what you are trying to accomplish.

If you want to maximize the potential of half-second assembly machines adding robots with logistics chests can work far better than belts.

Post Reply

Return to “Balancing”