Sushi thread

[sullen gale]

lol

Ok so now that I'm only measuring the contents of the belt, I don't really need to "reset" it or anything (because I accidentally F picked up some stuff), because it just... balances itself. Ok, this is neat.

I think I now see the appeal of sushi belts, and their utility.

Why did you get me into this!!

[plain hare]

Even cooler is you can use those with zero combinators

[sullen gale]

hmm?

[plain hare]

So you can make early game blueprints with that

With this sushi belt concept

[sullen gale]

OH

[plain hare]

Since wires are free in blueprints zero tech needed

[sullen gale]

OH

ok

that's neat

damn

[plain hare]

Naturally I had some fun with this idea

https://media.discordapp.net/attachments/198104144391700490/984193363194568795/unknown.png

[sullen gale]

ok that's.. intense

[plain hare]

I think it could be so much better with red belt sushi belts

And designed with that in mind

[sullen gale]

I've often been building bigger/dumber. Spent the last week pushing out artillery outposts beyond my cloud a ways, in preparation for expanding production on Nauvis so I can reliably send supplies to other planets.

but now I need to get into the nitty gritty of some of these things with the limited logistics requesters I have and sushi belts look like the best option.

e.g. for building cargo rocket parts, etc.

[plain hare]

It definitely is a good option, and even with requester doesn’t cause bot attrition

https://cdn.discordapp.com/attachments/419527276875481111/1068511369760096346/image.png

I just got production science started

Desperately low on oil (11 refineries from 8 oil fields)

So I went for coal liquefaction

[sullen gale]

I'm playing SE, and coal liquifaction is locked behind a science type I dont have yet 😦

[plain hare]

Yeaaa. Same here

That’s the production science I just got

I only have launched 6 cargo rockets

7th headed for cryo

[sullen gale]

Oh, I've messed something up, I didn't remove my memory cell stuff, now some of my inserters (that pull subcomponents out and put them back on the sushi belt) think that there are more items than actually exist. I should just wire them directly to the belt.. hmm.

[plain hare]

Yep that’s what I usually do

That or the output Chest at the end of the sushi belt

The belt doesn’t have to buffer

I think in that base up there the science Inserters at the top are wired all the way to the belt at science

[sullen gale]

Ok. I think its working now. This is really cool lol

[tacit cedar]

I had an idea
basically, when an assembler takes an item, it leaves an empty spot on a belt, and you could put the output back on the same belt to make some kind of sushi
it's very hard to keep it working though, at least without limiting the input

I guess this design in particular can't handle productivity modules

oh wait, it really was just the prod module's fault
it works fine now

[hardy fable]

ah yeah I made something like that a long time ago

the idea was to replace a full barrel with an empty one and vice-versa

a bit like trains but on a belt

[tacit cedar]

ah I have discovered a flaw
I backed up the output (filled the chests) so the machines stopped working
and now the entire belt is full of gears and it can't place the output anymore

[hardy fable]

yup, you need to make sure the flow is continuous

[tacit cedar]

this fixes it, it just recombines the two sushi halves if the red science output is full

[hardy fable]

https://twitter.com/cobalt_kura/status/1622297106492067840

what if we are the sushi

[gritty mango]

i was wtf then read it was a render

[normal gulch]

@hushed atlas the blue is how items get returned to the sushi mall

[steep cloak]

fireball don't I have examples somewhere

[normal gulch]

red is output that is headed out

I think so

[hushed atlas]

i see. this doesn't clarify much, sadly.

[normal gulch]

can you rephrase ur question?

[hushed atlas]

do you have an example that satisfies all of these criteria:

[naive rivet]

Looping back is never a problem because the looped back items just get priority for going back on the belt

[hushed atlas]

• sushi belt.
• has a nontrivial amount of types of items on it.
• clear example, without tons of other clutter.
• image has more than 3 pixels.
• all relevant mechanisms are shown and not offscreen.
• has the loop back thing.

[steep cloak]

here's an example image

[hushed atlas]

cool, thanks.

surely this won't scale particularly well for 40+ item types and such, however.

[normal gulch]

it does scale well

as seen in that cluttered image I sent in #vanilla-chat of the mall

[steep cloak]

you can have an arbitrary number of items

[hushed atlas]

the input mechanism alone would be huge. and you'd have to repeat it multiple times if you wanted to segment it.

[naive rivet]

You can just use bots for input

[hushed atlas]

???.

[normal gulch]

The limiters are compact now

[steep cloak]

yea I posted the other one cause it was simpler to look at

[normal gulch]

This is extremely compact compared to old iterations of using splitters to divide belts up

plus more precise

[plain hare]

https://cdn.discordapp.com/attachments/139677590393716737/884198439083184178/unknown.png

[steep cloak]

is that the old design

[normal gulch]

I think so

im trying to load my megabase where I placed the old design

but oh well, its a 100x100 block while you can get that down to 1/4 or even less now

[shut current]

it looks needlessly complicated until you try picking apart what each bit does.

The fact that the feedback is routed in near each input mixer makes it look more complex than it need be.

Also one wonders why you'd have all the post splitting before the final join and you realize later its so you can ratio limit how much goes in.
Like, 1/2 steel + 1/2 copper + 1/8 stone block + 1/4 red chips, etc.

so essentially the final ratio looks like 1:1:2:4

wait, 4:4:2:1

[normal gulch]

you may be able to make it more compact but when circuit controlled sushi is so much better why try?

[shut current]

Not necessarily. This system would be very robust, and require absolutely no counting. Its based on fractions.

Essentially, you can route in stuff you need in certain ratios on a single belt, say if something needs 4 steel, 4 copper, 2 red chips and 1 stone block, it will always have that ratio.

With a combinator box and counting, you'd sometimes have uncompressed bits of belt because belts have a tendency to have either 6 or 8 items (3 or 4 per lane), which hampers counting.

[normal gulch]

#bot-stuff message

codegreen is a genius and made it so you dont need a loop

[steep cloak]

that's literally what my circuit does

there's no counting involved, just outputting a ratio

[normal gulch]

and if ur worried about power just slap down a single and an

[foggy junco]

how robust is it?

[normal gulch]

it can survive a missing input, but if you accidentally pick up some of the items off the belt itll probably jam

[strange oak]

it ain't revolutionary, but I finally (mostly) figured out the basic concept for memory cell counting

Tomorrow will be fun. But for now I gotta sleep.
(yeah, I know it's miscounting at the moment, I was figuring out how to reset in case of error )

[normal gulch]

Hmm for science? I don’t know if I would do it this way. The main downside is if something messes up you have to restart and empty the loop

[strange oak]

I used it for my science setup, but it was mostly just a small test setup to learn how to use memory cells in this game

It's something I've always kinda struggled with (and uses for pulses)

I intend on keeping loops on the smaller side for now as well as "just enough" production for what it needs, so the cleanup isn't too awful.

that said, thanks again for making me learn this, it was a lot of fun to do something new

[normal gulch]

It would be easier and less likely to break if you use active measurement sushi

try this Sushi 101: https://cdn.discordapp.com/attachments/419527276875481111/1004819025441865818/unknown.png If you want to replicate take red or green circuit wire, choose 1, wire every belt to its neighbors. For the belt settings use measure belt contents, hold. Now you have a giant rotating measured chest. Take your wire and connect it to belts / inserter loading the belt. Enable them when (specific color of) science < X (X being desired count on the belt) .
One belt / inserter per science loading the loop

[strange oak]

Ah, that was how I did it in the past. I wanted to use this as an exercise to learn counters because at some point I think active reading, while much simpler, becomes a bit goofy to construct at larger scales

Putin's chest taught me that one, and it's the ultimate sushi beginner exercise for sure.

Counters are just abstracting an actively read belt (my view of things)

[timid cosmos]

Items/time is a good way to do it for large systems. My sushi-only factory used 20 reading stations spaced around the factory which read a single nearby belt piece over 10 seconds. All of those single belt's contents were combined on a wire and then divided by 20 to get an average across the whole factory

It's not 100% accurate but in my system there was enough room for fudging

and it should never deadlock if set up correctly

and you can do as much belt-running as you want picking up all the items and the system repairs itself

[strange oak]

Now that's a good idea

Kind of hits on what I was asking Putin just now

[timid cosmos]

that technique is how this factory ran, you might've seen it before the Sushi Village

[normal gulch]

yes the sushi village is exactly what I would recomend for you

maybe not exactly like this with the many small loops though

[strange oak]

Hmmmm. Perhaps I'll give that a spin for some subfacrories

[strange oak]

so I'm guessing this is built off of that same or a very similar circuit to what codegreen posted in the faq entry

[steep cloak]

nope

that's all memory cell pretty sure

my circuit is rate limiting

[normal gulch]

Its not straight up memory cell, its a fancier version where you read 15? belts around the factory that read the belts and that then is what is compared to against what you want on the belt

so this sushi heals itself

[strange oak]

yeah, I'm wondering how to approach using that rate limiting idea on a bigger scale than just like, feed in ingredients for one recipe

[steep cloak]

add up modules into a bigger ratio that applies to a single belt

[normal gulch]

May be interested in how me and putin did it for the sushi mall. Once the mall output is full, we stopped the items needed for that to go on the sushi belt

[strange oak]

oh i'm starting to get an idea yeah

[normal gulch]

So the ratio constantly changes as each assemblers output is full

[strange oak]

so like a constant combinator output that gets shut off and subtracted out of the 'total rate' put onto the one belt when its desired product is done

say you're putting like, 3 iron per second, 3 gears per second, and 3 circuits per second onto a belt, and you produce off that belt some theoretical item that takes 1 of each per second
when that's done you "shut it off" by subtracting that out so you're now only doing 2/s of each on the belt and so on

that's a neat idea

[normal gulch]

not quite

[normal gulch]

Using this example, lets say all the Green circuit needs are fulfilled. The circuit then has less input so it stilll outputs a full belt but its then half iron / half copper.

so the ratio constantly adjusts based on what assemblers need stuff

[strange oak]

interesting

[normal gulch]

And then I added support so the belt isnt more than 1/X of the way full with 1 item as the compact rate limiters cant handle much throughput

[strange oak]

I'll have to really start understanding some of the circuitry for how this filling to an input ratio works

that's why i thought codegreen's circuit earlier had some say here, seemed like one application of that

[normal gulch]

I think theres 3 main parts.

enabling/disabling of the constant combinators

smaller part making sure the belt isnt more than 1/x of the way full of 1 item

codegreen's circuit

[strange oak]

yea

the limiter for putting things on a given rate
the constant combinators for deciding what the rate is
and the last edge case solution thing?

[normal gulch]

edited it so its in order

[strange oak]

okay yea that makes sense to me

cool, thanks 🙂 that is quite enlightening

[normal gulch]

Id share an image but I dont know how enlightening that would be, its combinator hell

[strange oak]

is it the explanation of the circuitry going on in codegreen's circuit?

that one makes a lot more sense to me now given this application of it

[normal gulch]

+faq codesushi

[strange oak]

yep yep

[normal gulch]

That is part of it yes

That one is modified so it doesnt have a return though

if you want later today (5-6 hours) when I can get on factorio I can get you blueprints of how I did it

code's circuit though is on here somewhere

[strange oak]

sure, feel free to ping be though i'll be a bit busy for sometime this evening with D&D

i do have the codegreen circuit bp already

[plain hare]

try this Sushi 101: https://cdn.discordapp.com/attachments/419527276875481111/1004819025441865818/unknown.png If you want to replicate take red or green circuit wire, choose 1, wire every belt to its neighbors. For the belt settings use measure belt contents, hold. Now you have a giant rotating measured chest. Take your wire and connect it to belts / inserter loading the belt. Enable them when (specific color of) science < X (X being desired count on the belt) .
One belt / inserter per science loading the loop

[foggy junco]

@mellow panther mainly 3 ways of doing sushi:

• active measurement (what pc linked above) where you measure how much is on the belt and only add more once it's below a threshold
• flow limiting sushi, you limit the input flow of the items (e.g. limit it so only quarter of a belt is input) and then priority input the sushi belt back into the outputs
• memory cell sushi, you make a memory cell to remember what is on the belt and then proceed similar to active measurement

[strange oak]

Memory cell sushi:

[gritty mango]

aren't these a pain to fix if something goes wrong

[normal gulch]

yes

have to clear the system

[gritty mango]

☠️

[strange oak]

question with this sushi circuit: how exactly do you come up with the combination of rate/ratio?

i wanted to think the ratio was simply the recipe on the belt - in this example the blue science just consumes 3:2:1 red circuit, engine, sulfur, so the ratio val in the constant combinator made sense, but i'm not sure where I = 9 comes from

application: this horrible setup

[steep cloak]

I calculated the items per second that the build would consume

or items per unit of time

I and S are the numerator and denominator of the fraction of the belt you want full

you could make S the speed of the belt and I the number of items you want on the belt

as long as both numbers are in the same units of time it will function the way you want it to

[strange oak]

I guess what threw me off was this:

I couldn't find some combo that gave me 9/60 but it seemed to work for the build, so I assumed this wasn't the correct way of going about this

Ah, I think I understand now. For engines at .5/s it asks for .5 steel, .5 gears and 1 pipe/s, so a total of 2 items / 15, if it's some decimal (say, 2.1) then the fractional form (2.1/15 = 21/150, etc)

i broke the emergency stop wire but yeah, i think i understand what's going on here and now to use it. man, this is incredible, thank you for responding so quick

[steep cloak]

@strange oak the 9/60 comes from the lowest integer multiple of the ratio total and belt speed
1.125 + 0.75 + 0.375 = 2.25
belt input fraction = 2.25 / 15
= (9/4) / (15*1)
multiply both sides of the fraction by 4 and you get (9*1) / (15*4) or 9 / 60

gah stupid italics

[strange oak]

Yeah, I figured that out midway through I think, thanks for the confirmation and running the original example back

[steep cloak]

technically it's not the lowest integer multiple because it could be 3/20 but that's unnecessary confusion

you can avoid a lot of that math by plugging in the ratio normally and seeing what the T signal is on the output of the each + 0 : T combinator and dividing by belt speed for the decimal percentage, then divide 1 by that decimal and you get the number you need to multiply by

that's not quite right

[strange oak]

to be honest, traditional math is easier for me to do with rate calc -> fraction

[steep cloak]

missing the machine input factor

[strange oak]

about to fire up a test run

[steep cloak]

good luck

I can't believe how versatile my design is, not to be bragging or anything but it's really cool seeing the variety of applications of a relatively simple thing I made

[strange oak]

it is arguably one of the most amazing things i've ever seen in this game, with respect to how much it could allow me to completely transform how i play the game

as far as i can tell, it works!

i set the shutoff to be engines > 50 in the chest, but it has a lot of spillover room of course

The visualizer for this is great. Okay, I'm pretty dang happy with this, it's a great spot to pause the night on

[steep cloak]

looks awesome, glad you're finding that much utility out of it

[normal gulch]

dont quote me but you may be the first person to use that circuit outside of a testing world!

[strange oak]

if you can afford the circuitry room elsewhere, you can then compress down to just one input belt for anything as long as the total demand is less than the belt's max throughput (this is a bit of an obvious statement) but the implications for spaghetti are... quite interesting

[steep cloak]

you could use one of the factorissimo but for circuits mods to make it smaller

there's also a lot of applications for the circuit that don't use all of its features and therefore could have some combinators removed

the simplest version is just a set 1:1 ratio for everything, and that's just a clock

[normal gulch]

great for science that one is

[steep cloak]

I agree, yoda

[foggy junco]

Thus active measurement and flow limiting are more common

[hardy fable]

reject combinators, embrace splitter sushi

[foggy junco]

Embrace bloodbus

Or what was that one called again

That used certain items on the belt as activate condition

[gritty mango]

Items?

You mean hormones

[strange oak]

active reading my beloved

[foggy junco]

huh

[pulsar lance]

I think it was hormones on the bloodbus or sth. like that?
There doesn't seem to be an agreement, I found a discussion here:
#vanilla-chat message

[foggy junco]

It was mentioned in the old sushi FAQ

But it was changed at some point it'd seem

[normal gulch]

Blood bus used to use hermones and now it’s usually just active measurement

[plain hare]

Blood bus 1 was just active measurement. Blood bus 2 was the cool and unique ‘item as a hormone’ concept.

[strange oak]

blood bus... 2?

[plain hare]

When the last faq was written I decided blood bus #2 was way more blood like.

2 creators, years ago, one for each blood bus thing

[strange oak]

what exactly do you guys mean by 'item as a hormone'

[normal gulch]

Pistol to request another item or Group of items

[plain hare]

Like… for instance pistol could be your iron plate control ‘hormone.’ Measure some belts at the hormone factory. If iron on belts < X, add some hormone. If iron on belts > Y, remove some hormone. At the iron plate factory, every tine hormone pulses a belt, add 10 plates.

[normal gulch]

I was thinking for the modern sushi rails we could also use a hormone to control nuclear power construction

[strange oak]

oh that's interesting

i think i might try that for train sushi then

[plain hare]

Broadly speaking, it’s using an item itself as information transfer

[strange oak]

yeah, that makes sense, the moment you said 'pistol'

[normal gulch]

When you put it like that it’s useless on the sushi rails

[strange oak]

it's funny though

[plain hare]

Yeah the hard part for it’s practical application is finding a time when it’s useful to use an item for information transfer

[normal gulch]

If you have a belt you have electricity which has wires

[strange oak]

i am not known for doing things because they are practical

either way, hormone sushi is something i'll have to incorporate into this world to go along the other systems

[plain hare]

That’s alright im guessing we will have to try it out a few times inefficiently to find a good practical use case

And by we I mean… this sushi thread. To my knowledge no one has tried a hormonal sushi belt since it’s inception

[normal gulch]

Just seems needlessly complicated

[plain hare]

Oh I got an idea

I could switch out the circuits on my reactor to output a fuel cell for every empty fuel cell picked up

It’s simple but it technically counts xD

Seems like it would be more vulnerable though

[normal gulch]

Yeah way more vulnerable

You can cut the combinators in half already if you wanted to

[strange oak]

so here's an idea i had today:
send a train back and forth on a scheduled timer + inactivity to finish loads/unloads
load in the message request items and send the train out to the main dispatch, where it unloads and depending on the items that came through it'll load up and then go back out to refill

i think this will be a decent way to start getting into the mindset of using these items though this is undoubtedly just a bad way of doing things

i think with certain filtering on belts to ensure machines get only the amount they need for the recipes you can abstract out things like 1 pistol = ingredients for 250 green circuits or something yadda yadda.

the next step is gonna be to hook this into one big sushi belt much later on for a totally organic (but not in the way we usually use that term here) approach

[strange oak]

Ok, coming back to this. Question / mod idea- could anyone think of interesting uses of an item that abstracts multiple items? Think a mix between deadlock crating and deadlock stacking beltboxes.

I guess technically it's just an extra crafting step but I wonder if there's any interesting value to compressing things in that way

[steep cloak]

like compressing different items into one conglomeritem?

I guess compression in general but mixing adds extra complexity

[strange oak]

yeah, exactly that

I guess it might make it a bit easier to physically supply complex ingredient machines though (in modded cases)

[hardy fable]

you mean adding unfungible items to factorio?

I've been working a bit on that for the last two weeks

[strange oak]

I'm about to sleep, but I am curious as to what your ideas were

[hardy fable]

my idea was to make all assembling machines into furnaces then add a UI just like regular assembling machines to select the recipe group

my goal is to be able to use different alternate materials

for example, making a car with a titanium engine, aluminum plates and steel gears

putting all ingredient combination would be an absolute mess

in this example, I need the car item to contain info about its material content, for recycling purpose

[strange oak]

interesting

[halcyon wing]

What's a good way to increase sushi bus throughput? If I want to add more belts, how do I ensure everything goes to where it's needed? Or is it generally much easier to just spray and pray?

[hardy fable]

from my experience with multibelt sushi: do not mix them

keeping the mixer and recycler dedicated to the same belt is way easier

[strange oak]

ok, good morning
@hardy fable depending on what you mean by nonfungible potentially yeah

my idea was more like the output of deadlock stacking belt boxes but with more than one item

[halcyon wing]

What do you mean? Make each sushi belt it's own isolated demands/requests?

[hardy fable]

by nonfungible I mean that in factorio, any copper plate is the exact same as any other copper plate, they are thus fungible
in another hand, I would like a "metal plate" item that is either steel, aluminum or titanium, that can be used the same way (making metal gears from metal plate for example) except when recycling (recyling a car would thus return various amount of steel scrap, aluminum scrap or titanium scrap depending on what would have been used in the ingredients)

[strange oak]

that is interesting but markedly different

[hardy fable]

in your case you wouldn't use the item for a recipe, right?

I mean other than uncompressing it

but at the same time it runs into the exact same issue as my project, if you try to do it following the usual factorio rules you're gonna end up making a compression recipe for every single possible item combination there is

so yeah we have different goals but achieving nonfungible items would solve both

[strange oak]

Yeah, I figured using the item as a recipe would pretty much just add an extra crafting step. Like... if you combine wires and iron to make electronic circuit packs you've just... made electronic circuits.... so I think the only thing you can do is uncrate them first

the use case for this to me immediately seems to be dealing with machines liking to buffer more than they really should or need to but that usually ends up not being much of an issue

i'm not sure, the more i think about it the less useful i really think it is

re: compression for every single possible combo

yeah, that's also a worry, you basically just end up creating at least n(recipes in game you want to meaningfully stack which are most). okay, i think i've gathered enough to not really continue much with this idea, but if any sushi brains can think of another reason or use, let me know

[normal gulch]

What’s your usecase for this sushi belt?

[halcyon wing]

Vanilla base, everything but iron copper and green chips. Or space exploration science

[normal gulch]

what I would do is divide it up as much as needed.. so a belt for science, a belt for some intermediates, and you can use inserters to take from one sushi belt to another

[half phoenix]

@mellow panther

[strange oak]

i'm particularly fond of how this little modules and red circuit factory came out

the switch between feeding the modules and chest of modules was fun to wire in

[strange oak]

Right, so another rock and poor foresight proves yet again this circuit is invaluable.

according to my math I'll need 4.938 items/s and this will be super fun to figure out how that slots back in
I want to say the lowest common denominator I can use is 500

[desert tide]

yeah, 2469/500

why that number?

[strange oak]

4.63, .154 (x2)

I still need to figure out how to properly input this for the circuitry (see codegreen's discussion w/ me prevously above for context) though I think I might have the wrong procedure in mind. have to think of it later, though

[steep cloak]

it gets real complicated with prod mods

not because of the productivity, but because of the speed differences

you need to know exactly how much goes on the belt

[strange oak]

yeah, I'm considering not even bothering where possible, but the red circuits will have something similar going on, I think

might just not sushi it though I reallly wanna or just do it a different way with loopbacks and dreaded memcell

[strange oak]

it's dangerous to leave fireball and i unattended in your world

[strange oak]

[steep cloak]

I feel like I've sold weapons to a terrorist group

[foggy junco]

Haha

[gritty mango]

What happens if you don't filter the blocked part of the splitter?

Does the universe implode?

[steep cloak]

no, it's just free

[glass needle]

Wait, I just realized using inserters you can make circuit sushi that has access to every item on the belt

For example, replace all splitters and undergrounds with inserters and everything will be in-network

Is this... useful at all?

[hardy fable]

wat

[glass needle]

Circuit-only underground

So you just connect everything with a red wire, and you can have sushi everywhere

[naive rivet]

It'd probably be easier to make some circuit that stores the amount of items on a certain interval as values, which can then be passed on to a more general system :P

[glass needle]

Fair enough

[half phoenix]

https://www.youtube.com/watch?v=CCdpXXfiBFM which one of you did this?

[strange oak]

i have some questions about the editing

practical concept of sushi though, i think i'll do something similar in my k2 world

[eternal olive]

sushi!

[tender sluice]

im going to do it! im going to make the sushi without looping thingo!!

im going to try at least

[tender sluice]

this is so cool

i hope i didnt do this in a flawed way, it seems like sinewave's screenshots have more combinators

oh wait i can make it show the combinator stuff in alt mode

ok yeah it does seem to be flawed in some way, when i ran it for a while then disconnected supply and added it back it got clogged (after removing the loader in the top of the middle lane)

ah i got it, nvm!

yeah this is really cool

oh this is called code sushi

ah looks like the extra combinators are the emergency stop, so this implementation probably works (just missing that feature)

[steep cloak]

yeah its for when there's not enough input or when output backs up it will freeze the clock

[normal gulch]

this is cool, I never thought of doing electric furnaces with this sushi belt

[north perch]

@woeful pulsar

+faq codesushi

[woeful pulsar]

Codesushi

#bot-stuff message

[strange oak]

ay, codegreen moderator now, congrats (i've been outta town and just got home, perhaps the slightly wrong place)

[steep cloak]

thank

[normal gulch]

thank you for the weapons

[steep cloak]

you are not welcome

[jaunty shadow]

how far does a wall need to be from a flamethrower turret so spiters wont shoot over?

[plucky isle]

Why is this in the sushi thread?

[jaunty shadow]

oops

wrong channel

[normal gulch]

@signal obsidian #quick-questions message

so this is a bunch of loops, its what id call an advanced form of memory cell. it uses reading stations spread throughout the base to see how many items are going through some belts and then averaging it to enable/disable inserters

[signal obsidian]

he even has ore going onto the belt lmao

but he tracks it all through the factory instead of just at the entry?

[normal gulch]

he tracks it at certain intervals, theres a lot of entries iirc

I shouldve asked for a save at some point to look closer

because he has some spots where trains dump entire trainloads of ore onto it and it balances out

[signal obsidian]

damn thats cool

[normal gulch]

yeah

I may be insane but i want to make a better version at some point

endgame

so 9 machines surrounding 1 beacon

[signal obsidian]

if i was able to make the 2 belts balance properly i think i would be ok for now

[normal gulch]

balance properly how?

[signal obsidian]

im only at green science and i think i could definitely do all that as is

i mean just have them spread everything equally

so all the ore and the runoff is sent equally to 1 of the 4 belts going back through the factory

[normal gulch]

yeah.. I think the problem is running 4 belts like that isnt good as you want/need much more throughput than the 4 belts offer

[signal obsidian]

i probably will later

but for now i think im ok

[normal gulch]

for now yeah, when you realize its later its probably too late

[signal obsidian]

how many belts of ore input would i need do you think?

obviously 1 belt of iron ore wont be enough late game

[normal gulch]

im not sure how big are you going?

the base I showed you above averagaed 4.5 spm

[signal obsidian]

just wanna launch the rocket eventually

doesnt need to be the most efficent but dont want to wait hours per science

[normal gulch]

your 4 belt system assuming later you upgrade to blue should be able to last you, granted you will wait a while

[signal obsidian]

ill take it step by step

4 belts should be fine for now

[normal gulch]

yeah

[signal obsidian]

thanks for the help btw

[normal gulch]

your welcome, I love sushi and ive never really experiemented with multi belt identical sushi so wanted to give it a shot no matter what!

[signal obsidian]

back to messing with designs for my 4 belt sushi balancer

what we thinking?

[normal gulch]

hmm I think it should work

except for one thing: have input priority on the line coming from the left

[signal obsidian]

what do you mean

[normal gulch]

on the splitters, use the input priority feature on the belt thats coming from the left so that it doesnt jam up

[signal obsidian]

oh right yeah

so priority on the belt that goes through and not the dead end one

wait no thats output

output priority wont matter because 1 end doesnt go anywhere

[normal gulch]

yeah im talking about input

[signal obsidian]

ok i think i got it

still have a couple things to set up before i set it in motion tho

[signal obsidian]

my balancer didnt work at all oops

[normal gulch]

thats why i had that big balancer

[signal obsidian]

yeah i might need to use that

[signal obsidian]

omega brain balancer idea

[normal gulch]

I dont trust that at all haha

[signal obsidian]

can add more splitters until you do lmao

[normal gulch]

is infinity splitters an option?

[signal obsidian]

not quite but we can get a lot

i feel like it shouldnt work but i dont know why not

its too easy and stupid looking to work

[normal gulch]

same

[signal obsidian]

its not 100% balanced but every splitter gets you closer to it

[signal obsidian]

added 4 more belts so 8 belt sushi balancer is next

i have a plan for it so we will see

[mystic heath]

are you gonna make an entire factory out of a single sushi?

[normal gulch]

That’s the plan

Have you seen doshdoshingtons video on it? It can be greatly improved

@mystic heath

[steep cloak]

dosh did the second to worst way of doing sushi for that scale

the worst way being no circuitry whatsoever and just random inserting

[normal gulch]

100% agreed

[signal obsidian]

What's the best way

[signal obsidian]

Yes

Tore down my old base and am setting up a more ideal one, got 8 belts of input and set up over 100 steel furnaces

I don't want to just filter insert and filter splitter spam tho

I want the beautiful belt with iron ore, green circuits, red science, steel and everything else all happily co exisiting

I know I'll need more than 8 belts later on and probably a train system but I'll deal with that when I need it, I think I should currently be fine for all green science and probably blue too

[normal gulch]

I havent delved into multi belt sushi setups much, but perhaps something like tangled in bluegrass's sushi base where its a bunch of small loops and 17? belts spread throughout the factory that get read. #quick-questions message
The other option could be rate limited sushi.. but not sure on that

[signal obsidian]

Looks interesting, might incorporate aspects of it

I don't foresee any challenges until I reach purple science at the moment, I think I should be fine until then

Might need some redesigning at that point

[mystic heath]

🫡 good luck DakMillan!

[signal obsidian]

isnt it beautiful

[mystic heath]

quite :)

[normal gulch]

thats cool

[steep cloak]

do the red belts actually help mix things?

[signal obsidian]

it mixes it all pretty evenly

making them red stops backup a bit

i think anyway

i saw someone say it once

it seemed like its doing something

stuff cant be put onto the belt because its so full

need to adjust combinators a bit

[mystic heath]

I think dosh solved this one by having the assemblers output onto a side belt and then priority merging it into the sushi

[normal gulch]

so the main sushi belt backs up?

[mystic heath]

🤔 I suppose it would. Then you just need more assemblers eating the belts and compressing them into products

[signal obsidian]

its about ratios

you want less than you can handle

but as close to what you can exactly handle as possible

i set up more furnaces than necessary because if ore gets through it wont do anything until it loops all the way back around

so its vital to not send too much ore in

[hardy fable]

people have been requesting it all the time, here you go folks
circuit-less and deadlock-free 7 science packs sushi belt

I was able to make it 1 tile thinner

[halcyon wing]

What happens if you're not consuming military science?

[hardy fable]

it will be recycled back to the sushi belt

[halcyon wing]

Ooo, nice.

[hardy fable]

see the lil yellow arrow?

that's an input priority

[halcyon wing]

I need a method for expanding sushi to more than one lane.

[hardy fable]

if there's military science on the belt it will use this first

[halcyon wing]

I spent way too long fiddling with sushi in my Space Exploration run.

Only to ultimately have serious throughput problems with my design.

[hardy fable]

make multiple ones

I have tried for a very long time to make a multi belt sushi with no sucess

[halcyon wing]

I ended up having a separate sushi for biology and for material sciences.

DoshDoshington did an 6 lane omnisushi base.

Basically it had 2 machines pulling from 3 belts each.

Rather than trying to steer stuff.

[hardy fable]

you may find a way to implement them in a compact manner even if they are not mixed together

[balmy willow]

sushi train

[normal gulch]

thats complicated

[balmy willow]

the beginnings

just made basic smelting and green chips

def didnt take like 10hours of fiddling to make the computer part and like 30 hrs to make it stop breaking every few minutes

[normal gulch]

interesting.. I dont really understand / tried to comprehend them

[balmy willow]

i basically just built logistic train network mod but with vanilla circuits and for only one train

[normal gulch]

interestnig

so its only 1 item at a time still though?

[balmy willow]

on the train yeah

but it trainsports a whole trainfull at a time

[normal gulch]

cool concept, is it working now?

[balmy willow]

it seems to be

it hasnt broken for a while

[normal gulch]

awesome!

im now thinking of the cursed possibility of 1 hugeee train

[balmy willow]

yeah im only using 8 wagons but you could theoretically do it with infinity wagons

[hardy fable]

please define this term

[glass needle]

Well, it's just like a train bus, but you do it with sushi style

I wonder whether megabase in a single sitting is possible? Like, maybe a 24-hour marathon thing, sure, but would that be enough?

[hardy fable]

I swear I don't understand 😅

you have one train track with trains containing different items? that's just a normal train system

[balmy willow]

no, the whole base only has one train and that one train transports everything

so the one train will transport some copper ore then some iron ore then some iron plates etc

but its all the same train

[normal gulch]

I actually have thought about it before and think it is possible with enough blueprints!

[hardy fable]

ah, thank you 👍

[glass needle]

oh whoops this was the sushi thread so sorry about the off topic didn't realize

But ye I think I'll try 🙂

[normal gulch]

let me know how it goes! I think I want to try as some point as well

[steep cloak]

aren't there speedruns for it

[glass needle]

for... a megabase?

[steep cloak]

yeah

[glass needle]

well, for 100%

I'm not sure what scale that's at. It's certainly large, but mostly for making green chips

[normal gulch]

I dont see it on speedrun.com

[steep cloak]

https://www.reddit.com/r/factorio/comments/i91l61/segmented_railworld_megabase_speedrun_1k_spm_in/

found it

I knew I had seen one before

[normal gulch]

thats cool.. I bet i could do it faster though

ill have to try it out at some point

[glass needle]

well, that's segmented, so not one run

Also, calling it a speedrun to 1k SPM is a bit misinformed because SPM is not an easily measured end state like launching a rcoket. Speedruns ought to be verifiable easily.

I see they made up some rules for themselves, but it's easy to see how to cheese them, not to mention that it implicitly sets a 10 hour minimum time limit on the run and chooses no default settings, plus only measures space science, and well, this just seems like not a speedrun lol

Nvm I'm being too harsh ignore me

[dusky thistle]

nah i kinda agree. only space science isnt exactly an 'spm' amount by any usual imagination. That'd be sspm.

[balmy willow]

maybe the timer stops when the base is complete then the base has to be left afk for 1-2 hours to ensure its actually 1kspm?

[glass needle]

ye that's what I imagined. One issue is that this can (even unintentionally) be cheesed with sufficient buffers, which is why the 10h thing is a good idea.

[balmy willow]

only potential issue with 10h is that the mining outposts could run out

so youd have to make sure you build alot of them

[glass needle]

Good point. It’s really just a messy thing. I think the person in the post made a good judgement call in requiring half SPM before then

I usually go for a certain number of science packs myself

Actually, using their ruleset and the fact they essentially enforced 500SPM over most of 20 hours, a simple clean rule of 10k science packs in the shortest time might be good, since that would essentially require 1k SPM for most of it in order to be reasonably finished

Oh wait I’m multiplying SPM by hours lol

Sorry for all the messages, but “1 million science packs” sounds close and also has a nice ring to it

[balmy willow]

could also be a speedrun to a certain infinite research level

[eternal olive]

I am not an expert on. cutting edge sushi design. With current suchi techniques, will productivity researches mess with sushi design? Is YOUR favorite sushi design ruined by producitivy research?

[plain hare]

No

[steep cloak]

it's just more items/less machines

[normal gulch]

it will become slightly less efficient because you will have some machines not running due to full output belts. But the extra productivity will be really nice to have and it wont effect much besides a bit more upfront cost in form of an extra assembler or two.

[hardy fable]

short answer: no
but I am not sure I'll still use sushi belts in the endgame

I might get to direct insertion

for throughput reasons only

[steep cloak]

context: https://discord.com/channels/139677590393716737/1152752514073104434

the reason I avoid memory cells for state is because they are easily invalidated
less so if you're only keeping state of what items have passed through for a ratio, but that's basically the same thing as controlling the output rate for each thing of the ratio

[neat idol]

only time when I'm ok with memory cell designs is when there's decay built in e.g. rolling average counters, since those can be self calibrating: https://katiska.cc/n/temp/factorio/sushi/sushi-average.mp4

[steep cloak]

the circuit in the question thread linked above uses a memory cell to keep track of the state of the ratio cycle, not the state of the belt

which is a better way of doing it, but comes with a price

you can't use it for things where there is an ideal ratio, not a required ratio, because it's dependent on always fulfilling the item amount per cycle

@cosmic arch

(if you wanted to continue that is, no pressure)

[cosmic arch]

Not really, I was mostly looking to help the issue in the question thread. I will however come back here once Im finished with my abomination of an IR3 sushi mall

[iron moss]

SUSHI😤 🍣

[distant gorge]

!s warn @iron moss Please don't spam random nonsense.

[twilit shard]

@.janpeter. warned

Reason: Please don't spam random nonsense.
Duration: 30 days, 10 hours and 30 minutes

[distant gorge]

I ignored the first case

[iron moss]

im just trying to get convo going somewhere

[distant gorge]

You're really bad at it. Post something interesting and conversation stimulating instead of stuff that's just noise.

Actually try doing some sushi build instead of just saying "SUSHI"

[iron moss]

thats why i found myself in offtopic
and tbh i thought this was a honest literal sushi appraisal topic

im not as advanced in factorio to know such slang words

[steep cloak]

scroll up

the topic of sushi comes up often enough in the main channels that it deserved it's own place, so it didn't clog them up and also was centralized

this thread is about having many different items on the same belt, and the designs from doing so

[iron moss]

bro i watched that video up there now you said it and ballz yeah thats cool!

i sorta tried the same with ammo but it had to be perfectly calculated to work otherwise one or the other would start dominating the lane

[limber dirge]

Oh wow, I have finally found the gold mine of sushi minded engineers.

This thread is endless, and it will take me forever to go through the content.
This deserves to be a whole separate discord dedicaated to Sushi.

@hoary trail Would it be possible for you to share BPs to these builds?
#930147588160782436 message
#930147588160782436 message
#930147588160782436 message

This is genius. I have a some sushi ideas/builds too. I will share tomorrow.
Some of the ideas need some help, and I think this may be the best place for me to brainstorm them.

@steep cloak I am thinking this shouldn't clog even when there is low power.
Would it be possible for you to share a BP of this one?

I have a bigger version of this that doesn't use decider or combinators, but may not be as robust as yours.

[steep cloak]

+faq codesushi

[woeful pulsar]

Codesushi

#bot-stuff message

[steep cloak]

grr, that faq needs improvements

@limber dirge that links to the blueprint

it's pushed to it's limits, I didn't design the circuit for loopless sushi, but it can technically be used for that as long as you calculate the material ratio and input amount/time ratio

[limber dirge]

Thanks I will study it later tonight.

[clear wedge]

Blueprint String

Blueprint 1

[foggy junco]

@plain hare
So i'm making a request-based sushi mall where the throughput for the different materials will depend on what's being crafted at the moment - iirc you did something similar at some point, how did you manage the signals that request the input materials? Did you do constant combinator + arithmetic combinator for every assembler, or is there a better way?

[plain hare]

Constant and arithmetic combinator for each recipe yeah.

It’s a fair bit of effort, not sure how worth it will be

[foggy junco]

ight, was hoping there was a smarter way to do that

[foggy junco]

well, i'm doing a seablock run right now

[plain hare]

Oh ok

Maybe a fair bit then idk

[foggy junco]

i'm not even past green science yet

and the state of my "mall" is horrendous

[foggy junco]

dear god the things you can do with bob's adjustable inserters....

[white stag]

I made this right after watching dosh's video

[clear wedge]

Blueprint String

900 SPM

(Modded features are shown as question marks)

[tacit cedar]

wonder how I should get lds out
I'm thinking of just making room on the sushi belt (maybe doubling the space steel takes, but only putting it on one side, and using the other to deposit lds)
and then taking it out of the left side after recycling everything else
but I'm pretty sure that could cause a deadlock
maybe I wire the buffer chests and turn off output inserters if it gets too full?

[charred bolt]

Does anyone know how to properly wire sushi so when I attempt to let the inserters see the memory cell count it doesn't go off the fritz?

[distant gorge]

What scheme are you attempting? There’s multiple different methods for sushi

[charred bolt]

Sushi belt mall. I can send you a photo of my idea in a couple of minutes

[charred bolt]

That is the "brains" and is where I am running into my issue

[distant gorge]

I mean more like, are you keeping track of inputs and outputs? Are you keeping track of what’s on the belt directly? That high level methodology

[charred bolt]

Oh my bad

[charred bolt]

I use read belt contents to keep track of what I add onto the belts, then by using a memory cell and a combinator that multiples everything by -1, I attatch wires to inserters in order to keep track of what I put on and off on the belts. It's not 100% accurate, but it's close enough for my idea

[distant gorge]

So the ones placing down need to both read and write, and the ones picking up only need to write

If the ones placing down are just pulsing, there shouldn’t be an issue with them spazzing out

[charred bolt]

I think that's where my issue is coming from. I have all inserters trying to read and write. However when I try to connect the memory cell data to the power poles containing the data, my memory cell has a stroke

[charred bolt]

Well I half figured it out, now what I want to know is why whenever I place down the red wire, the cell freaks, but when I use the green wire, it's just fine

Eh, I'll make it work. Thanks for the help

I'm really sorry, would you be able to send a picture or smth? Now I have the issue of the red and green wire adding up to cause my inserters to not function properly

[distant gorge]

Why are you using both?

[charred bolt]

Because I use one wire for adding items to the belt, and one for off the belt (Sorry for shitty timing, was in class and getting food)

[distant gorge]

That seems like part of the problem

I can mock something up in an hour or two

[charred bolt]

If you wouldn't mind, that would be greatly appriciated

[clear wedge]

Blueprint String

Blueprint 1

(Modded features are shown as question marks)

[distant gorge]

That's the most basic way, if you don't want to have a central control point for the thresholds

The inserter on the left works whenever there's less than one speed module

[charred bolt]

Thank you

[distant gorge]

If you want to control the count on the belt from a central place, you can do something like this

[clear wedge]

Blueprint String

Blueprint 1

[distant gorge]

the amount requested on the belt is negative the amount that's in the constant combinator, assuming the inserters are comparing against 0

[charred bolt]

Thank you very much

[distant gorge]

Then there's another way that's a little trickier to conceptualize but I like as it's not reliant on the memory cell having a perfectly exact count of what's on the belt always

It measures the amount going by and "smooths" the average out in time to measure the overall rate of stuff, then inserters activate depending on whether the rate of their item is above or below what it should be

[clear wedge]

Blueprint String

EMA

[distant gorge]

This is the measuring circuit, tuned to the average throughput in items/second over the last ten seconds

[charred bolt]

Imma act like I know what that is

[charred bolt]

So just to make sure I have this right, The output of the memory cell is only attatched to the wires putting stuff on the bus?

[distant gorge]

yerp

[charred bolt]

Thank you

[charred bolt]

Where would I put inputs and outputs for these?

[distant gorge]

Input is the one combinator that doesn’t have an input, and vice versa

[half phoenix]

someone school me on how to do this safely pls

[clear wedge]

Blueprint String

Blueprint 1

[tacit cedar]

non-return sushi has always been rough for me
this should work as long as the inputs never run dry

[half phoenix]

yeahhh it's the "as long as" part I'm trying to deal with

I guess that's a lost cause?

maybe I could do like this and add a circuit that stops the gears if there's no steel, or vice versa idk

still wouldn't be perfect

hmmmmmm

it occurs to me that I could do 4 ingredient sushi and instead of having two input belts (top and bottom) I could use one as a return

these inserters seem to be doing the business
except still got one stuck machine on the end when gears got interrupted

[tacit cedar]

behold, my certificate of disqualification

[half phoenix]

did I say 4 ingredient? I meant 5. shit

[clear wedge]

Blueprint String

Codegreen's Rate Limiter

[tacit cedar]

simple circuit sushi machine

put items in the desired ratio into the constant combinator and it magically works

[half phoenix]

👀

ratios will be fun to calculate with productivity bonuses

[half phoenix]

what am I doing wrong here

oh the belts have filter conditions on them

[white stag]

How does the top assembler get pipes?

[half phoenix]

🤞🏻

lol wrong side fuck

alright that's some sushi lol. kinda defeats the original purpose of a compact design, but amusing to be sure

thanks @tacit cedar

[tacit cedar]

it asks nicely

[tacit cedar]

combinators stolen directly from codegreen

[half phoenix]

I think I can explain how it works if you care xD

[distant gorge]

“Enable if everything == 8”

[half phoenix]

huh, this seems to work with == 7

[tacit cedar]

since it's only half a belt, maybe only ==4?

or I guess, <=4

[half phoenix]

curved belts have 7 items I guess

[half phoenix]

this seems to be resistant to deadlocking

[clear wedge]

Blueprint String

Blueprint 1

[distant gorge]

You’d put it on the straight piece before that

I think curved pieces have different item counts?

[distant gorge]

Only if you’re putting the circuit after the first sideload but before you merge them

[tacit cedar]

no, before the sideload
if it's <=4, then it can work even if only half a belt of throughput is coming, right?
or now that I think about it, it might break by activating too early if a full belt is coming in

I guess == 8 would also work at half throughput tbh

[distant gorge]

<= shouldn’t work right anyway

And 4 vs 8 depends on whether you’re sideloading beforehand

And you can’t do less than the maximum because then the belt will start moving before stuff is backed up enough which means it might release stuff at different rates

[steep cloak]

+faq codesushi

[woeful pulsar]

Codesushi

#bot-stuff message

[steep cloak]

@half phoenix

there's another example

no engine machines, but you can get it loopless and no memory cell

[half phoenix]

two combinators are unpowered 🤓

[half phoenix]

okay no cap, when you do a sushi loop like this, why not put the packs on the same belt as the ingredients? you can peel them off anywhere with a filter splitter

[clear wedge]

Blueprint String

Blueprint 1

[clear wedge]

Blueprint String

Blueprint 1

[half phoenix]

🤔

[tacit cedar]

Does it break at all if gets backed up?

[tacit cedar]

splitter sushi engines

[half phoenix]

in the first one yes, it will slowly deadlock, although it takes a surprisingly long time.
in second one I added a check that stops the belt if there's blue science to be unloaded

[half phoenix]

@plain hare I can't believe you're not in here dunking on me for finally giving sushi a try

[plain hare]

And now in the SE discord I’m known to be obsessed with direct insertion :p

[tacit cedar]

i am cooking some sushi rn

[tacit cedar]

was fun getting the ratios

[tacit cedar]

@prisma sequoia you can look through here

[prisma sequoia]

thanks

[tacit cedar]

behold

[tacit cedar]

more splitter sushi

[hardy fable]

ah, glad to see somebody enlightened by the only rightful way of the holy splitter

[dusty hamlet]

this is accepted sushi

[red obsidian]

rocket tech card sushi is warming up, gotta pause that to go work on oil (no heavy set up -> no lube -> no silo )

[tacit cedar]

anyone know why this sometimes backs up a bit?
like just before the final splitter there's a clump

happens with most splitter sushi I build

[plain hare]

Well you mixed that area with the primary input belt

[tacit cedar]

wym

only one half of the belt is sushi

here's another one, there's like two items clumped up at the final splitter (top right)

[plain hare]

I think ima need you to circle the clump xD

[tacit cedar]

it's not a very big one
just the just the inputs to that splitter on the right lane

oh wow engines has a big clump of steel

and pipes

it doesn't really seem to be breaking anything though

[tacit cedar]

I guess I could save a spot, like this limits 3 items with a 25% rate limiter each, which essentially leaves an empty spot on the sushi belt

I could also leave a smaller spot, like if I gave each 30%, leaving only 10% empty instead of 25%

[bleak heath]

You can put the copy tool in a filter slot?!

[tacit cedar]

idk if you can do it from in-game

I stole the bp from codegreen

the bp of a single splitter

[plucky isle]

It used to be possible, now it needs a console command

[red obsidian]

It's very handy, because if you use the typical deconstruction planner instead, you might accidentally lose your stockpile of deconstruction planners when trying to automate recursiveblueprints

[eternal olive]

hate when that happens

[limber dirge]

What are these black icons on splitters?
Looks like a copy to clipboard icon from most GUI standards.

[distant gorge]

Yep

It is

[limber dirge]

Are those in vanilla? Never seen those black copy thingy to be used as splitter filter.

[plucky isle]

They are the ingame copy tool icons. You can't use them on splitters (anymore) without console commands

[clear wedge]

Blueprint String

Force Right - Blue

[limber dirge]

Thanks, can you also comment on what effect that does on splitter?

[plucky isle]

It filters all your copy tools to one side

And all the other items to the other

[normal gulch]

its used so 2 items arent wasted and doing nothing at the end of a splitter

[limber dirge]

Oh don't people use decon planners for that?

[plucky isle]

Most people do, but what if you happen to have your decon planners on a belt? They'll just get stuck at the splitters!

[normal gulch]

why not use the most cursed item?

[limber dirge]

Its a very corner case issue, why would there be any decon planners on a belt?

[plucky isle]

Do I really need to specify that that is sarcasm?

[limber dirge]

oh, well so if that was a sarcasm..... what is the real answer for "why not use decon planner as filter on splitters"?

Or, is it just a meme?

Gotcha

[steep cloak]

decon planner still has a minuscule chance of getting on a belt, so I'd rather use a more failsafe version @limber dirge

I also find the copy tool less distracting to look at in world

[limber dirge]

Hey Mitru, What is the eventual ratio of ingredients on the sushi belt in this picture?

[red obsidian]

At the beginning of the belt, 1:1:1.

The end of the belt is completely drained and fed back into the inputs; the screenshot is already at equilibrium (well... as long as science is getting researched and there's enough input material). The splitter mechanisms each inject 1/4 of a yellow belt of each ingredient into the sushi belt.

[steep cloak]

@bleak steeple here is the sushi thread

[bleak steeple]

Nice.

[naive rivet]

An integral part of the server

[mystic heath]

it ties everything together

[eternal olive]

and wraps it up tightly

[normal gulch]

@exotic oar can you share a screenshot of the entire area in map view?

[normal gulch]

wait now im even more confused of the goal trains -> sushi belts back to trains for no reason?

[exotic oar]

reason is more trains

[normal gulch]

okay lol

then what Id do is take 2 belts instead of 6 from each train then that should work?

[exotic oar]

everything works

i just need 18 to 6 balancer

[normal gulch]

2 9-3 balancers might work and then a 6-6 on the end

[exotic oar]

hm

9-3 i dont have

[woeful pulsar]

Balancer

Balancer Compilation
Balancer Mechanics

[clear wedge]

Blueprint String

9_3_balancer

[exotic oar]

no i need 18-6

[normal gulch]

well atleast currently that doesnt exist

you have loaders right? technically you can use them to make a balancer with a big chest..

[exotic oar]

big chest

is causing trains to unevenly unload

this is not may to happen

[normal gulch]

hmm im not sure

[exotic oar]

i tried it

[normal gulch]

what about taking less belts off a train as I said before

half the belts off trains, then a singular 9-3 balancer

[exotic oar]

then it takes time to get empty

welllllllllll its ok t wolrs as it does now

[normal gulch]

more time to empty = more trains :P thats kindof what you wanted

[exotic oar]

got solution

[normal gulch]

interesting

this would be great for changing train sizes I bet

[hallow sonnet]

im trying to make a sushi belt with the filter stack inserters changing their filter when there are 20 packs on the belt but it doesnt seem to work..
I wired the cc into the subtractor, going into the memory cell which will send the signal to the inserters to change packs
the green inserters should send a signal for how many packs they pick up
(2nd pic is showing -8 instead of 20 idk)

[hallow sonnet]

i can vc if any1 wants

[foggy junco]

how exactly are things configured?

or drop a bp perhaps

[hallow sonnet]

its not letting me render it

here it is

[clear wedge]

Blueprint String

Blueprint 1

(Modded features are shown as question marks)

[plucky isle]

neat

[foggy junco]

it doesn't seem to have all of the connections in there that i see in your screenshots when i place said bp

[clear wedge]

Blueprint String

Blueprint 1

(Modded features are shown as question marks)

[hallow sonnet]

how about now

i got it the inserters to set a filter but they wont change when they pick up packs

[eternal olive]

im trying an all sushi base. i dont know what im doing. its fun.

i need to decide how I want to procede with oil processing

if i want to do it normally or put it on the sushi loop with barrels

i suspect the spm is going to be very low. i cant imagine you can get even 5 spm on a single belt

i need to think about techniques to increasing sushi spm. maybe i can go up to 4 belts -- 2 on each side of each assembler

[mystic heath]

if you just go with raw numbers you'd need ~18.6 items per second for 1 SPM all sciences. This includes all liquids barreled, all smelting done off of the sushi, all intermediates hit the sushi and aren't directly inserted into consumers, and does not account for items removed from the sushi to make intermediates. The assumption is that there is always enough of everything on the belt in the proper ratios to make 1 SPM.

• Removing barrels drops it to ~18.01 ips
• then removing raw copper/iron ore drops it further to ~12.8 ips (yay now 1 spm fits on 1 yellow belt!)
• then removing the stone that's turned into bricks drops it to ~12.52 ips
• removing iron plate used to make steel drops it to ~11.3 ips
  So without barrels can say 11.3 - 12 items per second on the belt per SPM. With 4 yellow belts you have 60 items per second of throughput so 5 SPM is about the limit. With 4 red belts you could get up to 10 SPM without doing any direct insertion.

Used factoriolab to figure out the number of items required. productivity 3s used in the rocket silo because it saves a ton of resources. No prod 3s in any other assemblers, so there is room to further reduce resource requirements and increase SPM for the same belt item density

[steep cloak]

I would like to point out that it's not like you're inputting all 18.6 items/s going around all the time and then taking them all as output
in a concept similar to cache locality, if you're putting items on the sushi belt in a specific order, you can sort of use the belt as temporary higher throughput
for example, I could put some copper cable on the belt right before green circuits, and they will get mostly used up for the green circuits, leaving the belt as a whole with less items moving forward past the green circuits
so if you design the base correctly, you can actually fit much more on a single belt than you'd think, simply due to ordering your factories around the belt

you can do this even further by keeping track of if an item is needed, like if you have enough green circuits on the belt you can disable the output for the copper cable machines right before the green circuit machines as well

[eternal olive]

behold: the future of sushi: four lane sushi

[eternal olive]

good numbers and it gives me an idea of the magnitude of SPM to expect. thanks! I'm gonna do smelting off sushi because it just feels wrong to do it on the sushi, I dont know why

productivity 3s used in the rocket silo
I assume you are doing these numbers with space science? for me I think I'll just launch the rocket -- no space science needed, unless inspiration sparks me. that might eek out some extra spm.

[eternal olive]

some cool tricks named here and I'd love to see them in place but right now I think the aestetics of sushi I'm interested in is the items being equally distributed, so I'll not attempt to consciously implement these tricks, or count on them

[normal gulch]

j5 have you seen dosh's video on sushi? its close to how you are making your base

[eternal olive]

no i have not

[steep cloak]

here's an example where it really mattered, this was for a challenge getting blue science above I think 15 i/s in only 6 tiles of height and the input locations could not be changed only using red + green tech
I had the correct number of machines for each recipe, but it was getting bottlenecked from too many items being on the belt
I had to tweak the item counts and recipe locations for everything manually for quite some time, and it was getting closer and closer to working until it finally did

[normal gulch]

that is cursed

[steep cloak]

I had a bunch of input on the belt all the way at the start of the loop and I needed to consume as much of it as possible, while also making sure that the stuff I was producing wouldn't fill up the belt either, and allowing enough space for the science final products to pass through

[eternal olive]

my goal with this base is to get familiar with the benifits and limits of extremely basic sushi so I can use it judiciously in a normal playthough or even in SA

here's a zoom in on a single block. its extremely busy but the idea is simply that each assembler can reach to the 4 belts surrounding it and can place an item on the close and far belt

again, no clue what I'm doing so I'm sure there's a multiude of ways that it can be made better

one thing im worried about is how I add items to the array, it seems like it will run into thoughput items soon. but for now it works fine. and ive' been delaying dealing with barrels, which is going to be suffering i think

[normal gulch]

first major difference between yours and dosh's base is the form of sushi. Yours is active memory while his is memory cell

[eternal olive]

ive heard many of the sushi masters here sing tales of memory cell woes so I am avoiding it

[normal gulch]

yup

[eternal olive]

though i hate the aesthetics of every belt being wired up

[normal gulch]

same

there is a mod

[eternal olive]

oh?

[empty compass]

Clean Sushi

Cleans up visuals for circuited sushi belts by removing the heavy belt frames.

Owner

berggen

Downloads

38

[eternal olive]

woah

I'm actually doing this with no mods but I might add this one

seems worth

[normal gulch]

it isnt perfect as you still see the wires but its still great

[mystic heath]

ah yeah, it's with space science included. space science uses tons of resources. Without space science it looks like the items per second on the belt drops considerably, down to ~6.27 items/s/SPM with no prod 3s (~9.5 SPM with 4 yellow belts). CodeGreen is right about being able to get more out of your belts with a good design

[eternal olive]

we have blue science

the last hurdle to cross is cracking and its all downhill

[eternal olive]

you may not like it but this is what peak base looks like

[wary horizon]

why not use a memory cell?

[normal gulch]

theyre super annoying to re set

[eternal olive]

ive already made about 500 'errors' which would require me to reset the memory cell (forgot a wire on an inserter, picked up items from the belt, deconstructed a belt etc)

so not using a memory cell has saved my butt a lot

[foggy junco]

No assemblers mk2?

[foggy junco]

Anything that interacts with the belt in some way, be it you building / changing designs and picking up items as a result, or manually picking up / inserting items (be it by accident or on purpose), or biters damaging parts of it, or simply some part of the circuit not working, will not be stored in the memory cell.

So if an error happens, you usually have to completely reset everything to fix it, as in, pick everything up from the belts and machines, and reset the memory cell

On top of that, memory cell sushi is extremely difficult to troubleshoot. I've just helped out another dude recently who had a circuit which worked most of the time, but sometimes didnt register the inputs fed into the circuit. He ended up with like 50k copper ore in a system that should only have held 3k.

Memory cell sushi only tells you what is stored in the memory cell, and nothing else. So it's really intransparent and difficult to troubleshoot.

Throughput limiting sushi instead, especially if combined with a small buffer chest, can usually survive getting blasted to bits by a nuke / ran over with biters, and when rebuilt by bots they can fix itself with no user input required. It's also really easy to troubleshoot.

Active measurement sushi can't necessarily fix itself, but since you never get any false readings (like e.g. with memory cell sushi), it's really easy to locate and fix errors.

Which us why those two are usually preferred

[foggy junco]

by the way @plain hare , in your sushi kitchen design, you only do flow-limiting, right? Or do you also measure e.g. the resources that return back into the kitchen and then increase the throughput of the items that get consumed at a higher rate, e.g. if you feed 50 belts / min and 50 energy poles in the input, but 0 belts / min return while 40 energy poles / min return (so all of the input belts get consumed but the energy poles don't because e.g. you're currently building more of one type of block), does your circuit take that into consideration and feed more belts into the system?

I was thinking about whether or not it might be worth measuring the items that return (and e.g. average them over 1 minute), compare that to the requested inputs, and then increase resources that get consumed more than others by some configurable percentage

If the sushi belt gets really long (like 1km) there's obviously a pretty big delay, so in that case e.g. it might make sense to measure the contents of once per block or something and then divide by the amount of blocks

[plain hare]

self stabilizing :D.

nah I never went quite that far

[foggy junco]

because depending on what you use the sushi castle for, it might make sense

e.g. in my solar one, when i'm expanding the sushi belt itself obviously it will need more belts / undergrounds / big electric poles, while when i'm expanding the solar it will need more roboports / substations

or in case with walled sushi castles, you might want to feed more ammo if the sushi belt detects that it's low on ammo

the circuit should already be able to handle that increased input as well since it can deal with "too high requests" by itself

[foggy junco]

wait, wasn't the reason you designed your circuit to not have problems if the requested throughput is higher than the possible belt throughput because the requests were set automatically in some way?

how did that work

[plain hare]

It uh, re ratiod it

If the sum of item/min was over belt speed. You can do new item / min = (old item / min) * (2700 item / min ) / (old total item / min)

So if you had 10 items each at 540/ min summing to 5400/min, it would re ratio them all to be 270/min

Relatively simple in the scheme of things

[foggy junco]

nono, not that circuit

i mean the circuit that'd automatically request stuff

was that the sushi castle?

or the sushi mall

[plain hare]

Oh yea the mall

Fireball programmed in what recipes required and it asked for some items related to missing mall item stock

Also relatively simple in the scheme of things

[foggy junco]

ah so not the sushi castle itseslf

that'd actually be a smaller sushi belt in terms of length

so perhaps it would profit from an addition like that

unless you perfectly have every single recipe stored in some combinator

so the ratios are (near) perfect while crafting

[normal gulch]

I think I did have every recipe in the combinators and it was way too bulky and too many combinator and places for human error to be practical

[eternal olive]

we have purple!

[normal gulch]

how much science per minute??

[eternal olive]

I'm building machines up to 15 SPM but it'll perform way under unless I get the ratios perfect etc. I also need to build up oil processing a bit, its under built right now, particularly plastic, so purple is way underperforming

i basically got purple going then took the screenshot 🙂

[eternal olive]

lol putin why the . if something messed up let me know, im just having some fun

[plucky isle]

Spying on your sushi

[eternal olive]

hehe

delicious sushi

[plain hare]

Well I’m always up for a little theoretical sushi discussion

The style you’re using there where every belt is measured (I think?) generally does better the shorter the belt is

Are you finding any issues with item ‘clumping?’

Like the items are on the belt, but not evenly distributed, so the assemblers don’t actually work as often as they should given what you have on the belt

Since there’s just a short window of insertion when the clump goes by

[eternal olive]

i havent put any time into observeing performance, just building. clumping seems possible and aestetically displeasing but I dont know if its come up yet

[eternal olive]

yellow science GET

base

[mystic heath]

very nice

[wary horizon]

lmao and now they got whole belt readers

[eternal olive]

sushi chads .... our time has come

[eternal olive]

i knew my time trying out sushi would pay off

[plain hare]

I like making malls with one of these whole- belt- read sushi belts for the random ingredients, and then send in a few normal belts with the bulk

When I’m playing normal games and such

Or can do the same for science builds and stuff

[eternal olive]

thnk about how easy the mall can be when you can have stacked green belts

one stacked green belt would give you the equavalent sushi throughput as 16 yellow belts

[plain hare]

Oh I hadn’t thought about that yet

That will be fun

[eternal olive]

yeah green belts and stacking is what made me want to try out sushi

i think its gonna be cracked

[half phoenix]

will read all the belts in the same 'Transport line' as the belt being read. It survives going through underground belts, but is broken by splitters and side-loading onto another belt.

sushi chefs gonna be mad when they find out about all the other ways transport lines get broken...

[steep cloak]

they have custom logic that does not follow traditional transport line breaks

it's specifically splitters and sideloading that break read hold segment, normal breaks like the 200 cap or past inserters don't affect it

[half phoenix]

they think of everything. how silly of me to lack faith

[eternal olive]

just waiting on silo now

getting about 10 spm

76% ....

[eternal olive]

I win!

Final base

Things I learned:

• belt reading sushi like this this is pretty fun and easy .... before the sushi complex gets too big
• the biggest problem I had sushi is how hard it is to diagnose a problem. ie, if there was an issue I'd have to look all over the base to find the missing wire or misconfigured assembler. huge pain in the butt
• one issue I had with my sushi setup is was hard to expand.this will become much much easier with the sushi features released this FFF. but next time, I would absolutely leave 2 gaps between my assemblers and the sushi belts instead of one. making the sushi easy to extend should be a high priority for a sushi design
• i cant imagine using a memory cell sushi with how many mistakes and adjustments I made. the only time I can imagine using memory cell sushi is if I predesigned the sushi in editor mode and then plopped it down and never touched it
• barrels are so annoying
• one issue I was running into all game was needing to adjust the item count properly. i never really figured how to to do this and get it down to a science
  overall, fun run

[clear wedge]

Blueprint String

Sushi in proportion

[safe agate]

Maybe someone will find it useful

[normal gulch]

@safe agate is it just quickly enabling / disabling the belts in a ratio to get the items out that you put in hte combinator? does it use modulo?

[safe agate]

yes, it tries to fill a segment with resources in proportion, it is not used modulo... it works well, but it is not suitable for those who like precise calculations

[normal gulch]

+faq codesushi

[woeful pulsar]

Codesushi

#bot-stuff message

[normal gulch]

@safe agate there’s this cool blueprint which is perfect and uses modulo so that the belt is nice and clean as well

[safe agate]

"This content is no longer available." 😦

[plucky isle]

https://cdn.discordapp.com/attachments/334735604342325249/1076302300974415953/blueprint.txt

Copy-pasting the link should let discord generate a new access token

[safe agate]

This sushi is good and I liked it, but it has a different goal, let's say precision... My sushi is trying to fill a belt (any type), but whether something is taken from the line, all or nothing, it doesn’t matter. Endless cycle.

And the lack of resources will not disrupt the scheme either.

[steep cloak]

the circuit in the example does that

it it just doesn't do it in the example

here's what a recycle loop could look like, though the circuit used here is simpler because it's only trying to fill the belt instead of a portion of the belt

you can use the circuit in the linked example in place of the one here and it will do exactly what you're wanting it to do

[safe agate]

Thank you, I got the point.

[glass needle]

I really want to see someone build a belt reading sushi with two belts crossing using inserters so all the items are still being read

[normal gulch]

sounds cool in theory, but in actual practice ive found the small inaccuracies of items going on underground belts does not matter much

[glass needle]

Yeah it's definitely an "in theory" thing

[normal gulch]

itd be helpful when you need 100% perfect sushi I bet

[steep cloak]

I think we have established previously that inserter based belt crossings get way too big and require way too much setup to be useful lol

[normal gulch]

ive only seen a yellow inserter design

[glass needle]

we did, I still want to see it XD

Technically though red inserters and such based designs could be a lot simpler

[steep cloak]

long inseters make it a little better lmao

[naive rivet]

It should be possible to have a circuit that can acccurately output the items on an underground belt as a signal

Dunno how big that would get tho

[foggy junco]

You could technically build memory cells for every segment that isnt measured

But then we're back to memory cell sushi, which was exactly what we were trying to avoid with active measurement sushi in the first place

[steep cloak]

small scale memory cell setups should work fine, the larger they get the easier it is to make a mistake and ruin the whole thing

[steep cloak]

here, have an exact underground reader

[clear wedge]

Blueprint String

Blueprint 1

[steep cloak]

this can also be a global circuit, you do not need the combinators on every underground

left pole is for reading purposes, also I forgot to include the substation and energy source in the blueprint (sorry)

[hardy fable]

I have my rate-limiting splitter sushi setup done for SA planning 12 science packs and green belt, just managed to get down to 7 tiles after the priority merger

it uses yellow and red instead of red and green, to keep the 1:2 ratio

enough for 1200spm

[foggy junco]

Belt based throughput limiting?

[hardy fable]

yus

[foggy junco]

nice

[normal gulch]

1800 red science is done. I have already grabbed 3 ore patches before logistics is researched, this map is going to be a grind

now the question is, do I splurge the 600K science for fast inserters or go into green science with out

[hardy fable]

holy shit, isn't that massively overkill this early in the game? 😵‍💫

[mystic heath]

600k science for fast inserters
super expensive recipes? O.o

[hardy fable]

that's... 20k times more expensive than it should

[mystic heath]

hrmm... fast inserters are almost 3x faster than yellow inserters so it would let you use a few fewer machines to get the same amount of production. Probably worth going for fast inserters now to get them out of the way and get their production started for later expansions, and so you aren't caught out later with an inserter bottleneck once you start upgrading your assemblers.

[hardy fable]

depends how fast your machines are to begin with

I use for my unbeaconned

and boilers as well

[mystic heath]

I do the same, and use for anything that takes a long time to craft, like sciences, where the output rate isn't helped by upgrading the inserter speed. as long as the furnace can empty faster than it fills it's all good. I was thinking more for the 0.5s craft time items if you want minimum number of machines working on it. buuuut, you can pretty easily just add a few more machines so overall you're getting enough items in and items out to fit requirements. Once the upgrade needs to happen the old can get recycled into the

[naive rivet]

Seeing the amount of resources those researches require I'd only do the minimum required for bots lol

[normal gulch]

I just realized I put this in the wrong channel.. @slim wave here the progress I got done last night

[mystic heath]

:D

[sage whale]

I have found my people :)

[inland terrace]

I guess this would be a fun place to tell people im a sushi chef irl.

[eternal olive]

teach us your secrets

[red obsidian]

base screenshot pls

[inland terrace]

Base screenshot for you

Feel free to DM me for more base pictures if you really want them

[sage whale]

that actually looks so fuckin good

i need it

the sushi around where i live is spensy

[pulsar lance]

Got some nice sushi, though it’s not working 100% as intended

https://media.discordapp.net/attachments/546453369271615503/1240068789413089351/grafik.png?ex=6645378e&is=6643e60e&hm=a9f20464bf68f115b9ac3fb7075c8f8678a44493e49d3c94ae1ba1d6574a0caa&

The problem seems to be, that I need more than one belt of throughout through

[normal gulch]

It looks like most items are on 1 side of the belt ?

[pulsar lance]

where exactly do you mean?
quite everywhere I see

I havent really dealt with the bottom loop at all
I need to do demand based sushi, or multilane sushi If I want the top one to succeed at all....
2 of the 8 items aren't used at all, and some of the others are used more and other less depending on the current needs

The main goal of the sushi contraption is to save space/belts because most of the stuff isnt needed anyway: The problem is though, that no one knows which stuff is currently needed though....

I think just using more belts is easiest here, but cooler would be demand based sushi:
Either I buffer infront of each factory and only send materials to those buffers
Or I keep track on which machine is currently activated and always put the ingredients for that factory on the belt

I stell kinda need to think a lot of it through at this point, especially how to make sure machine a isnt running, when machine b needs the ressources more importantly

[slim wave]

Sushi thread?? How did I get on here and why didn’t I know about it

Oh fireball invited me

[hardy fable]

welcome to the big boys club

[hallow sonnet]

whats the capacity of 1 belt?

[slim wave]

Depends on what shape the belt is

[hallow sonnet]

curved

[hearty relic]

Man am I behind

[slim wave]

I think a curved belt holds 7 items

Oh that’s days old

[near patrol]

wait we have to be invited? I kinda just started seeing it ages ago

[hearty relic]

nah, i joined this thread without a particular invite. Just gotta find the thread

[steep cloak]

when you ping someone in a thread it joins it automatically for them

[balmy willow]

you can either be invited by someone pinging you or you can look in the thread list and join yourself

[near patrol]

ohhh I see, nice

[distant gorge]

sushi of a sort

[hearty relic]

hows copper getting on there?

[distant gorge]

what copper?

[hearty relic]

its hard to see the red ores on my screen, but they look like copper

[distant gorge]

It's thorium

[slim wave]

What mod is this

[distant gorge]

Angel's

[shut current]

@formal cipher ping me here once you're back.

[slim wave]

I just had an idea to do a sushi ultra cube

[normal gulch]

I want to see how that’d work

[shut current]

Oh thats what you meant. As in the practical issue of science just not being consumed because not all sciences are present, thus filling up the belt and not leaving space when the sciences come back.

Tbf, the design needs one last combinator check to make sure the final splitter per color has enough input, and if not, not to output to the condenser splitters (the pyramid-shape arrangement thinging)

[hardy fable]

that can be done with a buffer between the recycler and the priority merger

big enough to store the looping belt full of its science pack alone

[hearty relic]

i feel like im overcomplicating this

regarding the sticks to rails

[distant gorge]

probably

[slim wave]

Yea

Without any circuits it’s not going to work

[fierce merlin]

does anyone know any good guides for how to make a sushi science belt using circuits

I want to have a belt that goes around my labs with equal proportions of all sciences but I dont know where to begin when it comes to wiring up the inputs

[hearty relic]

one simple strategy is to only release science when there's an equal amount of science to release, but this isnt flawless. In particular, if the research you are doing doesnt use a particular science, this can become desynced.

Building from this, it would make sense to have buffers in front of all the labs then check the contents. Populate the belt based on most needed items.

[hearty relic]

man i forgot about this;

[plain hare]

try this Sushi 101: https://cdn.discordapp.com/attachments/419527276875481111/1004819025441865818/unknown.png If you want to replicate take red or green circuit wire, choose 1, wire every belt to its neighbors. For the belt settings use measure belt contents, hold. Now you have a giant rotating measured chest. Take your wire and connect it to belts / inserter loading the belt. Enable them when (specific color of) science < X (X being desired count on the belt) .
One belt / inserter per science loading the loop

[sharp valley]

I just use this type of thing for sushi:

https://cdn.discordapp.com/attachments/139677590393716737/856194363259617350/unknown.png?ex=666fade3&is=666e5c63&hm=3c4a19849d9e575e1013a14406352cbca9ccd2fbf1c1e6011900bb43556cfb82&

[hardy fable]

I swear I am genuinely crazy

[normal gulch]

You definitely are crazy, circuits are so much easier lol

[hardy fable]

I like the beauty of it

[eternal olive]

we need to rapidly design sushi that can handle up to 15 sciences, at least one of which can spoil, as well as quality

[sand estuary]

Attempt at belt sushi

I didnt realise the use for the 2nd filter splitter

so it clogs

Which is a shame

Because its perfect for tiling

[hardy fable]

why 15? 🤔

[sand estuary]

The new sciences I think