#Sushi thread

1 messages · Page 8 of 1

plain hare
#

And you can see how that could save cost, due to bus belts / width, even tho this uses 7.5 belts per assembler instead of 4.5 belts per assembler that a 24 red circuit assembler build could use

hoary trail
#

yeah that makes sense

#

I messed up because I needed to load science onto a train, but I didn't have anywhere to put the train initially

#

now the train is blocking my vertical expansion for the bus

plain hare
#

Science in a train! So fancy! Lol

hoary trail
#

that was actually my first train haha

plain hare
#

Since this is the thread… sushi is great for science collection haha

hoary trail
#

I wanted to put my labs somewhere out of the way

#

my science is produced too far away to be sushi'd together on site

plain hare
#

I think, you could do a flow limiter at each science production cell

hoary trail
#

so I use a train to run around and collect it all, then dump it into my sushi kitchen

plain hare
#

Makes sense… ultimately it matters little tho, as long as the train doesn’t buffer too much science

#

A train could buffer less depending on distance I suppose

hoary trail
#

well I moved the labs pretty far away so the train was definitely less buffer

#

also my bus doesn't look too long yet imo

unknown.png
plain hare
#

Yep you could definitely get some more value out of that bus without a complete train transition I think

hoary trail
#

@plain hare I think active measurement methods could be effective for koravex

plain hare
#

Potentially yes, though there are some good designs already that use next to no circuits which is cool

#

My favorite is 5 stack size 8 outserters, and 1 regular outserter that fired when those other ones move

#

So basically 40 is always recycled by virtue of the stack size

hoary trail
#

I also like sushi

#

So I'm going to see if I can make it via sushi haha

plain hare
#

Or 10 stack size 4, I guess

#

But yeah sushi is cool haha

clear latch
#

I remember trying the Kovarex thing with a single stack size 10 U-235 inserter and a counter that swung the U-235 outserter when it counted 40 U-235 (so 4 swings). Unfortunately prod modules caused this simple math to desync and stop working after a while so for my real lategame save I ended up cheating with inventory sensors lol

plain hare
#

Much easier on design and by end game 40-80 U235 permanently stuck in the centrifuge doesn’t matter much

clear latch
#

Yeah that makes a lot more sense, I actually considered factoring in prod bonus in the circuit until I realized that damn purple bar was a little different every time lmao. So for each input/output it's not a consistent predictable and discrete bonus

plain hare
#

Yeah it’s a bitch to design it you add too many design constraints IMO

#

Clarifying whether you need low or high buffer, prod modules, helps a lot

clear latch
#

Tbh what I actually like to do lategame is grab 2k 235 and 238 and use Even Distribution to spam it around the centrifuges to jumpstart it. Then the onsite uranium patch keeps it going

plain hare
#

Yeah that surely helps a lot

hoary trail
#

@plain hare here's a pic of my blue, purple, and yellow science sushi builds all fitted into my bus. Yes I know I need to be using prod modules; prod modules are my next major project to build. But I liked this picture as it showed the three major sciences all building 36-38/min. Once I get my prod modules going I'm planning to upgrade them to assembly 3s, which should take them all to 60 spm.

unknown.png
#

Although I've actually been very happy with 38 spm, I'm still researching things faster than I can implement them

#

also I just realized that I accidentally bussed my lubricant on the wrong lane. I'll have to fix that 🤦

hoary trail
#

@plain hare how are you making items from your sushi castle available for the bots? I'm trying to decide on how I want to deliver items to some bot networks and I can't figure out how you're doing it from the sushi belt

#

At least, not in such a way that doesn't seem like it would waste a bunch of resources

plain hare
#

A constant combinator can hold a bunch of signals as a request for a filter inserter

#

Negative chest contents and positive constant combinator requests balance out to stop the filter inserter from working once request is filled

hoary trail
#

Oh, I didn't know that

#

That's really handy

plain hare
#

It’s a good one I think @normal gulch got this figured out

#

I used to have like 40 red chests in the sushi castle mk1 lmao

normal gulch
#

global wire?

hoary trail
#

Tbh I think your sushi castle is much more efficient, and possibly easier to set up the baseline for, when compared to having a construction train

#

I mean your as in both of you

normal gulch
#

mines not a castle its a wall

hoary trail
#

Well, it's similar

normal gulch
#

yeah similar shoob

unknown.png
hoary trail
#

Walls are part of castles

normal gulch
#

@plain hare kinda took the name sushi castle

plain hare
#

I think so as well 😄 hence why I made it lol. That being said late game it can synergy well with a building train. Sushi builds the rails / defenses, train comes in with mining / smelting / production equipment

hoary trail
#

That makes sense

plain hare
#

Haha I guess I did in a sense shoob

hoary trail
#

Since you'd have to be specific with where you want the machinery to be

plain hare
#

I think part of the problem with most construction train designs is, theyre either very much preset to a certain ratio of items or the combinators required to make them not a preset are fairly intense

hoary trail
#

That makes sense

plain hare
#

So I kinda think of it as an end game sort of thing, whereas the sushi construction belt can be brought out quite early in a game and modified easily as well

normal gulch
#

have u seen De hacked's?

hoary trail
#

I like the sushi construction belt a lot, I'm considering trying it on my next save

plain hare
#

I have not, but I’m assuming it’s one of the request based building trains?

hoary trail
#

It's either that or try some mods

plain hare
#

Or use the sushi construction belt in a mod pack shoob

#

I’m about to use it for my 100x seablock run a lot I reckon

normal gulch
#

i think in chests u have 1 item of each ( or a bit mroe for belts and stuff ) and if that drops the train goes there and stays for a wghile

plain hare
#

Oh, yeah, that’s classic building train style I think that’s how I used to run my auto construction

#

Until I decided I needed the rails auto constructed too haha

#

I’ll have to return to the building train soon for later game malls

hoary trail
#

Wait how do you guys remove the loop at the end of the rail line to extend it when you're adding to the end?

normal gulch
#

no loop at the end is how i do it

hoary trail
#

How does it not back up then?

plain hare
#

Right hand drive sushi belts

#

Belt tries to go right all the time but if it can’t it flows left

hoary trail
#

Oh I see

#

This is very clever

#

And also very expensive

plain hare
#

Yes fairly expensive. Early game sushi rails (landmine defenses, 2 lane rails, yellow belt) used about 3 yellow belts of iron ore to automate 50 meters of track / minute with 2.5 tile thick surrounding landmine walls

#

Late game with 4 lane track, sushi castle defenses , etc, might be looking at idk 10-20 blue belts iron ore

hoary trail
#

Maybe this sort of scale isn't for me to try yet haha

#

I'm still working with 38 spm numbers

plain hare
#

The thing about it is that, it’s an investment that pays for itself

#

If you invest in the automation system, you can claim more ore and it will ‘pay off’ running itself and more

hoary trail
#

If you had built a straight piece and later decided you wanted a T there, what happens?

#

Do you have to tear it out and replace it?

plain hare
#

The intersections are designed so that they can be placed on top one another.

#

I built the T by starting from 4 way and stripping it down to less lanes

#

So it’s seamless upgrades, I can start from 2 lane straight track and make it a 4 lane 4 way

hoary trail
#

This is so much work haha

plain hare
#

Yep it’s kind of a huge project of mine lol.

hoary trail
#

After seeing you spend so much time building it, I don't think I'd want to try to replicate it myself, at least not at that scale

#

Not yet, anyway

plain hare
#

I think, that may be wise shoob

#

I am kind of notoriously slow at executing my ideas tho

#

Too obsessed with perfection I reckon

hoary trail
#

My ideas are much smaller than yours

plain hare
#

I do like helping folks so I’d hope I could help you achieve similar stuff faster

#

Well, it’s not like I got here all at once right.

hoary trail
#

Maybe first I should make my sushi bot rush build

#

That'll be good practice

plain hare
#

Yeah, I think that could be good. I’d focus on smaller systems that can find uses many places

#

This kind of grandiose do it all at once sushi castle has so many overlapping systems I won’t lie it’s the most ambitious project I’ve ever undertaken

#

I don’t think it’s something I’d… suggest you just go and do if that makes sense

hoary trail
#

Making purple and yellow science was really easy after I'd built my sushi designs, I just plopped them on my bus, connected the outputs, and started researching stuff

plain hare
#

The power of blueprints 😄

hoary trail
#

The power of sushi

plain hare
#

But yeah I’d focus on making more ‘single build’ blueprints that can be used in a wide variety of situations

hoary trail
#

Those would have been way harder to make without sushi imo

plain hare
#

Haha that’s a funny thing to say xD I’ve been struggling with designing active measurement sushi stuff. I think I just need to try things a bit more and see how it goes, been trying to pre plan too much probably

hoary trail
#

I just set a target, like 38 yellow science, and write all the numbers down in excel and then run a bit of percentages to get the totals for the line

#

I also usually balance the materials so that there's an even split (more or less) on both sides of the belt

#

Since I don't balance the two sides

plain hare
#

Mmm excel probably would help here

hoary trail
#

After that it's just a jigsaw puzzle trying to fit in all the machines into a rectangle

#

Then I plug in the final belt length into my excel sheet, put the numbers in the combinator, and I'm good to go

plain hare
#

Also I need to finish my sushi rails probably before I get into active measurement, it’s so enticing but I really must finish my projects one at a time I’m a one track mind

hoary trail
#

That's fair

plain hare
#

Feels like pulling teeth to castle these huge intersections , but I’m like halfway done… just gotta finish

#

Yeah I think, from your design process described , I need some excel in there

hoary trail
#

It really makes things go pretty smoothly now

#

The only ones I struggled with were purple science simply due to the mass of materials involved, and yellow science due to the variety of materials involved. Yellow science I had to split into two sushi belts to get it to work

plain hare
#

Mmm yeah makes sense. I’ll probably only partially sushi things, focusing on just the most expensive products that are far from needing full half belts

hoary trail
plain hare
#

It is… ha

#

In the finished versions they should look like this right

#

The sushi belt and ammo and pipe connections ‘reach’ into their source location, this helps out construction speed overall, and prevents flamethrowers roasting your own stuff

#

However… this reach is very hard to design

clear latch
#

Those give me Battleship vibes, except maybe in space 😄

plain hare
#

I just can’t wait to use it to cut through aliens like a hot knife through butter lol

#

My goal is sushi rails completion end of January I need to turn up the heat on these if I wanna do that

clear latch
#

I should probably learn sushi myself sometimes for this weird lane switching rail that FARL struggles with...but my backlog in this game is immense and I had the misfortune of finding crafting combinator and merged chests first lol

plain hare
#

Hmm interesting. Sushi is enticing even in the face of mods like FARL?

clear latch
#

Yeah, FARL straight up can't signal the lane switching track between the two normal tracks, I think it's limited to 2 lanes at a time or something and that counts as a third

plain hare
#

Huh… interesting

#

Well maybe you can use the sushi rails book as inspiration or something when it comes completion time I’m gonna ‘publish’ it

clear latch
#

Nvm it's not a lane limitation, I think it might actually be related to the shape of the track. It can't generally place curves and apparently won't "Read" perpendicular straights as valid blueprints either, has to be simple straight or diagonal. Normally not placing curves is a good thing because it means intersections act as a natural brake so you don't cruise into a nest while afk, in this case I'd need bots aka sushi rails

unknown.png
plain hare
#

I’d suggest robo rails as potentially useful solution

#

Notice robo rails has to use a belt to transfer the first logistic bots to the neighbor network, however. This is what originally inspired the sushi castle - I tried this robo rails a few years ago - realized I needed a belt for robots, and thought… couldn’t I move most of the items with a belt? And well. The rest is history

clear latch
#

Ohhh they use inserters to bridge the networks...I think I've seen this before but that never clicked somehow. At the time I tried pairing logistics chests 1 gap across and naturally nothing happened xD

plain hare
#

I suppose I should of edited my statement to say, belt, inserter, train can serve as internetwork logistics

clear latch
#

Ngl there's definitely appeal to building entirely in map view. What I do now is pre-place an aligned intersection and then use The Fat Controller to remote control a FARL builder train to 'connect the dots' one segment at a time, it's nice for long distance but inefficient for close intersections

plain hare
#

Oh yeah. I love the idea of map view building (as you can probably tell from my designs) ha

#

I’m planning my sushi starter base around 100% map view building once you hit blue science and robots come out of the bot rush, and build up a few more smelting lines / build my sushi mall.

#

Ideally this means the engineer(s) can adopt more of an oversight - combat role. Kill aliens, connect belts between mines - stations - designs, pumpjacks, build blueprints.

#

The original idea came from watching professional RTS players - they have their game down to an art. I realized, you can do the same for factorio in terms of the economic growth stuff - if you have the blueprints

clear latch
#

Automated skillz 😄

plain hare
#

Heheheh yeah. I hope Ill be able to make a glorious time lapse of conquering a max deathworld, where the factory grows actually exponentially.

clear latch
#

I can actually see a future where I need sushi for certain mall inputs. Merging chests have a default limit of 42 tiles and bumping that creates extra sprites for every possible length and width combination for every type of chest allowed to be merged, basically a huge memory hike. So for stuff like engines or LDS that gets really long or has weird ratios adding up to less than 1 belt I might run into that

plain hare
#

Yep, sushi is fantastic for the random low throughput mall ingredients

#

I expect I’ll be using it a lot in mod packs especially

plain hare
#

May add a small green circuit thing onto the front of it

hoary trail
#

Nice! Also wow that's a lot of science

#

Also is it just me or are you only using one side of the sushi belt?

#

It looks to me like everything is only loading from the outside in

plain hare
#

The lane balancer at the front end mixes it up enough for it to work.

hoary trail
#

Oh that's a lane balancer

#

Any particular reason why you chose a lane balancer? Asking so I can consider it in my builds as well

plain hare
#

Uhhh. I dunno. I just realized I was only using one lane shoob

hoary trail
#

Oh lol

plain hare
#

I think it could have been done better

hoary trail
#

I usually try to balance the numbers so there's mostly equal numbers inside and out

plain hare
#

I guess ultimately though the configuration being the way it was for red / green science led me here

hoary trail
#

But it does make for a wider build

#

So less ideal for bus space

plain hare
#

Yeah I like the thinner build, I’ll have a Bus of materials off to the side coming from smelting.

hoary trail
#

What is your spm with this?

plain hare
#

This is 90 spm.

hoary trail
#

Wow that's a lot

#

Most of mine are only designed for 60 max

plain hare
#

I think 45-60 is a more sober goal for single player for sure

hoary trail
#

Haha

plain hare
#

Part of my planning here is I’m either gonna be playing multiplayer or using companion bots

#

So early game bots or teammates to help build

hoary trail
#

I'm not sure 90 spm would even be reasonable for purple science tbh

plain hare
#

Yeah, it won’t be

#

Well not immediately at least

hoary trail
#

That one required a red belt to even get up to 60 spm, and that was just barely

plain hare
#

The plan is gonna be 90 spm red- blue, with 45-90 military, and then 14 spm purple / yellow

hoary trail
#

End throughput was 1.75k /sec, iirc

plain hare
#

That should let me afford the useful cheaper purple / yellow techs, while using my leftover resources for the mall.

hoary trail
#

Oh that's another thing to think about that I forgot to mention

plain hare
#

Oh also, I’m not actually committed to sushi everything 😄 I just try and do it where it can make the most practical sense, if that makes sense ha

hoary trail
#

It's not just about setting the item caps in the combinator and calling it a day, you also have to look at the max flow rate and make sure it's within the belt limit

plain hare
#

Yep yep

hoary trail
#

Putin not sushi-ing everything? What is this heresy?

plain hare
#

That’s what caused the lane balancer haha

#

Well, I’m gonna use sushi for low throughout mixed item spots 😄

#

Which is a lot of places but not everywhere

hoary trail
#

Purple science is sort of low throughput

plain hare
#

I’m sure military / blue science will have some sushi.

#

Yeah, I’ll probably sushi the prod mods / electric furnaces.

#

The rails tho… idk. We will see

hoary trail
#

Yellow science actually is low throughput if you ignore the LDS

plain hare
#

Yep there it would probably be good too

#

But like, I may just do a half belt of pipes / gears since those are pretty cheap items all considering.

hoary trail
#

LDS required almost an entire half belt of copper by itself, so I just let half the belt be copper and sushied the other half

plain hare
#

Yeah the more I talk to you the more of this base will be sushi 😂

#

I just want it as efficient as I can, within… reason so I’ll probably be doing some more sushi yeaaa

#

That Being said direct insertion for engine units / electric / robot frame seems almost better than sushi

#

Dunno I’ll have to get there, I really should be making sushi castles first anyway, as that’s first on my priority list

#

For reference I’m kinda recreating this design with some tweaks for 90 spm and cheaper to build / easier to hand place

#

But you can see how the engine / electric engine stuff is pretty efficient already without sushi, robot frame / electric engine belt maybe could use some work tho

#

I’m trying to separate the sushi starter base into 3 100x100 chunks - smelting, chemical refining / power / circuits, and then an assembly science / mall block

#

In the past I tried to connect the bot rush base to a train station but I think it’s frankly unnecessary, im gonna embrace the spaghetti belt connections between blueprints, then let robots build my big train station / bus

#

Separating it out helps a lot more in the beginning of the game because of all the obstacles- cliffs, forest, water, biters

hoary trail
#

Yeah I like the smaller blueprints too for the same reason

#

That's interesting that you prefer direct insertion for the engines and robot frames, for me it seemed easier to do it with sushi because of the variety of numbers of machines involved

#

I can't quite tell what you did to get the direction insertion to work so cleanly though

plain hare
#

I paired 5 engines to 6 blue science (perfect ratio)

#

Similar to 3 wire: 2 green circuit

#

And it’s 1:1 engine: electric engine

#

And 1:2 electric : robo frame

#

So engines are quite solid to just direct insert

hoary trail
#

Hmm I see

#

I'll try that when I update my designs for botrushing

#

Right now I'm more focused on some other projects, like a 4-reactor nuclear design

#

Sushi doesn't do much for that haha

#

I'm also at the point where I'm considering making some bot-based designs, like a bot-run military mall. I'm starting to wonder if at this point my base is past the point of sushi being the answer for everything. Sadness.

plain hare
#

When you go to fuel the damn thing you may not want to buffer so many fuel cells on belts.

#

Less fuel cells on belts -> faster 40 U235 for kovarex!

plain hare
hoary trail
# plain hare Actually it does!

true, although my biggest bottleneck was the 50k stone I needed for landfills to place my nuclear power plant safely on the lake. I hadn't realized how little stone I was mining until then lol

hoary trail
# plain hare Bots do tend to compete with sushi use cases.

I feel like my argument for bots instead of sushi is that expanding the mall to produce a new item is easier with bots, especially when that item might require a random ingredient that nothing else would require (and therefore wouldn't make sense to put on a sushi bus)

#

at least for active measurement sushi, to me the primary advantage over spaghetti or regular bussing is the ease of routing the materials, and the cost and space efficiency of being able to minimize inserters and belts and buffers. But bots are more space efficient, easier to route the materials, and can have smaller buffers. They're just more expensive to build, and slow and cost a lot of power.

#

Flow-limit sushi is different, in that it has consistency over long distances, something that active measurement is not feasible for and something that I wouldn't trust to bots

#

but flow limi sushi also requires a kitchen which is typically big and expensive

#

I've considered whether it is possible to integrate flow limited sushi into a smaller build. I think it would be possible, but I don't see any advantages to it as it'll take more space, be a bit limited in how fast the inputs can be (or have a big expensive kitchen), and probably be more difficult to set up without any true performance benefits.

plain hare
#

Yep I’d definitely Say that’s a good analysis of the sushi options compared to bots

#

Once you have the logistic bots, it definitely makes sense to use them for the mall (this is what i do)

#

I just have a blue science age sushi mall for before my bot mall cause it’s a smaller resource investment

#

Late game I’m not really using sushi in production ,- not often at least - but just in construction sushi belt

normal gulch
#

for example, robot frames I made with sushi i think it costs less due to UG's but I have a new way to design thigns that doesn't use undergrounds as much so idk 🤷

plain hare
hoary trail
#

@plain hare I made a new sushi thing. I feel like I'm going to have to start referring to it as "cursed sushi" haha

unknown.png
#

This one is two separate sushi half-belts (red and green wire) that share a common center that is also used to route away the red circuits. The only part I don't like is the belts required to bring in green circuits and copper plates, but it was necessary to fit the chemical plants in there. If you were to do plastic separate then those lines could obviously be shortened.

#

It seems to produce closer to 178 rather than 180 red circuits per minute, but that's due to the transit time of getting plastic and green circuits to the end, and because there's essentially zero buffer. That could probably be improved by using red belts on the outside tracks but I don't really care about getting 2-3 less red circuits per minute.

#

Also the input inserters for the plastic and the green circuits need to be limited to a stack size of 2 otherwise plastic has a tendency to outcompete green circuits on the input side.

#

actually nevermind, it looks pretty stable at 179

unknown.png
#

The more I do this type of thing, the more I become convinced that sushi is way more viable than people say it is haha

#

I mean it's the same output as most other regular 24-assembly designs, but with far fewer belts and inserters, and I think also a smaller footprint

plain hare
#

Nice build star penguin, seems like it saves inserters compared to a standard red circuit build for this assembler count , but it’s capped at 24 red circuit machines

#

I think it would be good for the first red circuit design but as you go up to 24+ probably more efficient methods could be found

hoary trail
plain hare
#

True you could but at that point the design would be inefficient because red belts are inefficient 😛

#

And you’d have to redo the sushi a bit

hoary trail
#

Well, yes, the red belts would then make it inefficient haha

#

Although the sushi should be the same, just have to add higher limits to accommodate the longer belt

#

Same pattern should work though

#

Or I could just use two of these side by side

#

48 assemblys would last me a while

foggy junco
#

my current setup for sushi artillery

unknown.png
#

am i missing anything?

#

i'm thinking one sushi lane and the one in the middle is one lane landfill, one lane uranium ammo

foggy junco
#

my mall is becoming a mess though

#

2 belts of steel, 4 belts of greens, 1 belt of reds, 8 belts of gears and it still isn't enough for my sushi castle

plain hare
#

Where’s the castle part 😛

#

But that seems… fine enough. Maybe a few things to consider improving

#

I like having the belts connecting to the next castle behind the blueprint, so the belts ‘reach back’

#

This means when you’re constructing a new tile you don’t have all those belts just filling with items and sitting there

#

Might be worth having a constant combinator feed signals to an inserter to pickup multiple items but idk

#

I’d suggest decon plannering the output of merging splitters to avoid items that just sit permanently in the splitter too

foggy junco
#

Ill plan the other ones later today haha

foggy junco
#

that's like 100-200 items buffered at most

#

that's a few seconds worth of the sushi

plain hare
#

It’s not a super big issue but You have to at least do it a bit to not get the belt extending past the wall of the sushi castle so you may as well extend all the way back to the splitter

foggy junco
#

@plain hare how many belts of materials would you recommend as inputs for your mall while doing sushi construction?

#

right now i have 4 belts of iron, 4 belts of steel, 4 belts of greens, 1 belt of reds and 8 belts of gears

#

and it's still not enough

plain hare
#

This was my sushi castle mk1 production target - I used a 1 castle / minute goal and used the blueprint of the castle to put in exactly what was needed. This is a good approximation but will fall slightly short of 1 castle / minute because of the buffer on belts and the buffer on chests. https://factoriolab.github.io/list?z=eJwljs0KwjAQhN-mhzklaVpy7SZaEUGU-nObix4sCG0FDx767G5aQr5NZodhBhF4EygR3lEaeEqCNTXlBx.YwXnGDRzjFpXyqoJh2sMy3tUdbwjs4SvD-IRVx4g68IFahZVTZpORNhpNGfNbPsvSZb40bciJetMZttQ0beHU9YatAlOjw-j3hFobdmvDHtaXlANcxV2ueIEtpiMi0rwcBLRzW3zLP9KVO6Q_

FactorioLab

Factorio calculator, Dyson Sphere Program calculator, and Satisfactory calculator.
Determines resource and factory requirements for desired output products.

#

I think you’re close to what you need

#

That production target is rather old I’m gonna be repeating this process again soon as I make later game sushi kitchens / malls

foggy junco
#

11 blue belts of copper

#

:thonk:

#

oh wait that includes the circuits haha

#

so around 4 belts

plain hare
#

But the best way to do this is really just get the design first enter it into the lab maybe adjust some things for the buffer you’ll see on belts / chests (though this is imperfect cause if you increase item flow on belts you increase buffer on belts)

#

Realistically you’d need some fancy math or just be fine with an approximation / over estimation in order to hit your build rate desired

foggy junco
#

i mean i'm using the mall for other stuff too, so just overestimating should be fine

plain hare
#

That’s kinda like a rough idea of what building with cursed beacons costs

foggy junco
#

i kinda fucked up the planning of the mall so it's kind of a headache to add more materials to it haha

plain hare
#

24-30 yellow belts of smelting into product / hour

foggy junco
#

you still use yellow belts in lategame?

plain hare
#

That’s my mid game 😉 so no

foggy junco
#

ah haha

#

my entire mall

unknown.png
#

so far

plain hare
#

That’s like the scale up process…. Want to very cheaply grow production and all of that fits on my yellow belt sushi rails.

#

24-30 belts of ore into product / hour for like 200 items / m on the sushi belt

#

It’s like my proof of concept for replacing a building train with a sushi belt in the mid game

foggy junco
#

early game sushi castle

plain hare
#

Mmm yeah maybe it’s time you follow in fireballs footsteps here and go for a mega mall

#

Hehe did you see the costs on it tho! It’s so cheap with landmines

#

And I’ll just kill everything in the way of it with my tank 🙂

foggy junco
#

haha

#

usable in max deathworld?

plain hare
#

Yeppp

#

That’s the plan at least

foggy junco
#

nice

#

spearhead through the nests

plain hare
#

My sushi mall is blue science compatible

foggy junco
#

nice

plain hare
#

So the sushi rails can get deployed in late blue sci and I think I can automate that ore capture quite well

foggy junco
#

yeah, red circuits seem to have been the bottleneck

unknown.png
hoary trail
#

I wanna see pictures of the sushi, not just the results haha

plain hare
#

I think for my late game malls I’m gonna pull on trains of intermediates and let logistics bots go wild

clear latch
#

When I did that I decided to try making everything except modules and solar in one mall. I even made science which was a big mistake lol. I had 20+ unique trains and certain ones like gears and LDS would be guzzled in almost less than the time it took to unload the stuff at peak demand. It was really funny because I ended up setting their stops to limit 2 with no stacker and there was never more than 1 onscreen at a time. I really needed stackers or fewer trains or more specialized "malls" at the time

plain hare
#

Mm yeah I’m gonna design it for super high throughput I think. The first one I do will be cursed beacons so it doesn’t need super high throughput and should be good practice.

#

I like the idea of a max deathworld run where we go low pollution all the way to megabase shoob

clear latch
#

In my own run I'll actually be using lategame eff with double slot SE beacons, like half eff3 half speed3 and it's still 20 beacon equivalent for the mkIII beacon

#

Another throughput issue with my mall was I had this weird roundabout XL parking lot on the mainline to fit a 2-8 trash train and an artillery train. With sushi rails I could just use the return half of the belt loop for trash, probably throw some fuel and logistic bots on the outbound half too, so that's quite a few utility stops saved

plain hare
#

With sushi rails you say 👀

#

Yep, sushi rails are great for those random little things you want to supply to the base I think

clear latch
#

Well it's just rails so far, I figured I should get the block sizes and new stops planned first

unknown.png
plain hare
#

3 sushi in one place

unknown.png
hoary trail
#

I ended up making an early game military sushi build that produces small amounts of defender bots and explosive cannon shells. This time I ended up using sushi for the iron plates, copper plates, steel plates, and green circuits. Everything else ended up more or less with direct insertion. A very different sushi approach than normal, as I usually try to sushi the intermediates rather than the raw materials.

#

But it wasn't hard to do, it seemed to simplify the inserter requirements, and I think it also cut down on the buffer (not that the buffer would really have been an issue with such cheap ingredients)

#

I also linked the iron input inserters into the furnaces to the sushi belt so that the smelters only run when the sushi belt is running low. I think this keeps the smelters from filling up with 100 steel in each

glass needle
#

There's a lot of conversation here but I'm pretty interested. Anyone able to sum it up for me?

plain hare
glass needle
#

Thanks!

hoary trail
#

@glass needle another sushi convert potentially?

#

Sushi is amazingly efficient in the right applications

glass needle
#

Really?

#

I thought it was just a meme

plain hare
#

It used to be meme status but it turns out it’s just hard to use and hard to find applications for

#

It makes some amazing builds

glass needle
#

Got any blueprints for it?

plain hare
#

A lot, I mean idk depends on what you’re wanting to do with it

#

This automated artillery system is pretty nice - the ‘sushi castle’ as I dubbed it

#

Separate logistic networks let you turn rail pieces and they auto construct with bots from map view - sushi belts carries items between networks

glass needle
#

Hard to see what's going on

#

Why do you have trains if you're using the sushi for long distance transport?

plain hare
#

I’m not using sushi for long distance transport, I’m using sushi to build the rails and defenses

#

(Automated expansion to ore outposts )

#

It’s blue science compatible and delivers all items as requested or desired on the sushi belt to outposts

#

About 200 items / minute of miners, furnaces, inserters, etc on the sushi belt and you can grow production massively in about an hour - you can add like 20-30 yellow belts of ore -> processed goods by my rough calculations (if you keep it running the whole time)

#

This has a sushi belt input for rare intermediates (red belt) meanwhile everything else is output to the yellow belt sushi belt that goes on to auto construct rail segments and factory parts

#

Remember circuit wires are free in blueprints so you can place that from the get go ^

glass needle
#

Wait automated building? Is this a mod?

grave oak
#

sushi chat sushi chat

timid cosmos
#

Automated building comes with construction_robot , they are vanilla.

glass needle
#

That's not automated, you still have to put bps down

plain hare
#

A single click grid aligned print is the most automated construction you get in vanilla

glass needle
#

Ah, yeah, I guess being able to just click and be done with it is nice

#

So I guess the conveyor belts just move the construction materials then?

plain hare
glass needle
#

Oh right cuz there be trees

plain hare
#

Yep trees… rocks. Fish. Old builds in the way…

hoary trail
#

@glass needle I've been making lots of small scale sushi builds for early and mid game. For me, sushi really shines in such applications. I'm currently using sushi for red science, green science, black science, blue science, red circuits, and basic war supplies. I have sushi designs for purple and yellow science too, although those need to be updated to match the others in terms of spm and overall size.

#

I'd also suggest if you're thinking about learning sushi, it's worth it to put in the effort to understand what's really going on rather than to just use blueprints. It makes it so much easier to debug a system that you broke when you understand how it works.

#

Plus you can better customize it to your own needs

glass needle
#

Okay! It helps to have some bps to study in the first place is all

hoary trail
#

That's fair, but every one of my sushi build ends up being pretty specialized

clear latch
# glass needle Okay! It helps to have some bps to study in the first place is all

For learning flow rate based sushi, I found it helpful to start with splitters and ignore the return half of the loop. The middle working versions create the same effect (belt /2 /2 = /4), but I got the idea of using these loops to flow limit from one of Putin's screenshots. So that funnels all the outbound belts into only 1 return belt to manage. The loops aren't magic; no consumption + unlimited production = back up completely/jam
So you might want to disable the source inputs to eliminate that eventuality. That means either circuits or carefully planned production/consumption ratios (like science), and if you go for circuits you'll get something resembling active control

unknown.png
glass needle
#

Where would I best use sushi?

clear latch
#

Low throughput demand (< 1 belt by definition) and high ingredient variety I guess. So a pre logistics mall is a good case, there's also lategame ones like LDS and maybe red circuits that are really slow recipes with each thing taking up a fraction of a belt. You can break this rule too, like for science you can use 3-6 sushi belts to feed labs; also with warehouses/long chests you can get something very similar to wiring every belt segment but higher capacity and probably more throughput with inserters/loaders feeding them. And any kind of modded belt compression or high speed belt would also change all throughput considerations

plain hare
# glass needle Where would I best use sushi?

Early game malls (no bots), construction belts for feeding robots items to build with, low buffer belts (helpful to avoid massive buffer on production lines where it’s not needed)

plain hare
#

Seems like you did that in the last screenshot by looping back the output belt but I think you just let it buffer by not putting on input priority

clear latch
# plain hare Your backup problem can be entirely fixed by using input priority on the back fl...

TIL, I think that's the first time I've ever used input priority 😄. Pretty cool how it chokes the input without having to "know" your flow limit, just that there's backup ahead. It doesn't even need much buffer, I got a moving average counter from calciumwizard that stabilizes over a minute and 2-3 chest slots' of space is enough to consume the temporary surge. In practice just a realistically giant loop would do

unknown.png
plain hare
#

Yep. Now you know why the new circuit flow limiters have a chest to handle surplus surge 🙂

clear latch
#

Oh wow really - haha yeah for a while I thought active provider or active control was the only way, how am I gonna get that before yellow science xD

plain hare
#

I’m gonna add a chest (2 total now) in my next model, a buffer chest for surge and supply, and an active provider to let it dump into storage chests should the buffer chest fill past a threshold

#

That will let me deconstruct thousands of items safely.

#

(Any amount really)

#

Heheheh yeah no need for the yellow science chests, they improve the ability to deal with surges but a few thousand items is usually enough

#

This sushi thread has really just been going and going lol. Fun stuff.

hoary trail
#

Is rate limiter sushi really the best place to start though? I feel like active measurement sushi is easier to set up, although it does require more math

plain hare
#

I really don’t think I’d describe either one as the ‘best’ place to start. Rate limiter sushi has the advantage of the belt loop length being basically any length, for a larger setup overall, while the active measurement is best for a small tidy setup

#

Also I feel like the rate limiters take more math overall

#

(Until you build the circuit flow limiter at least)

foggy junco
#

well, both are arguably better than any sort of memory cell sushi

#

because that shit can break way too easily

plain hare
#

Too true, too true

hoary trail
#

Yeah I don't think I'll ever use memory cell sushi tbh. It's too fragile. And it also doesn't really have any advantages over active measurement other than that it looks slightly cleaner. But it's not cheaper or more compact. I suppose it's a bit easier to extend the loop, but that's about it.

#

I definitely prefer active measurement and rate limiters

#

I use active measurement when I'm doing a nice standalone build that I don't intend to change, especially when I'm intending to slot it into my bus. I use rate limiting for things that I intend to expand, like labs or my eventual possible sushi castle experiments.

plain hare
#

The advantage of memory cells is they allow theoretically the lowest buffer

#

I think there are some ways you could make them less fickle to work with

#

They mainly break from players interacting with the belt when they shouldn’t, so one of my ideas of protecting them was to encase them in walls.

#

You can also make a manual switch to override the memory cell.

#

You could also make some automatic things to reset the memory cell should you notice errors, with some effort

#

You could for instance turn the memory cell design into a loop put a single fish on it as a hormone item and get an active count of what’s still on the belt after a loop. You could theoretically compare what Is on the belt to what should be on the belt, and reset it from that info by Reading a belt on pulse

#

A mix of active measurement, memory cell, and blood bus I suppose 😛

hoary trail
# plain hare The advantage of memory cells is they allow theoretically the lowest buffer

It isn't necessarily a lower buffer than active measurement, though. I always set mine to have the belt 50% full, but you could easily set it to 10 or 5%, or even just one of each item if you wanted to. I just use 50% because I like watching my items on the belt, and because the lower you set your buffer the more delays you see in terms of belt transit time, although I don't think the overall effect on output is really very significant

plain hare
#

The belt never buffers to any degree really

timid cosmos
#

Memory cell designs will break if you ever just forget to wire a single inserter. Come back and theres no iron in the system, gotta take everything off the belts and reset the combinator.

#

Thats why I like reading the contents of the actual sushi belt, you can take as many items off the belt as you want, and it self repairs.

#

Also with active sushi, you can have as small of a buffer as you want. It just depends on how you design it. The design where you wire every single belt would be simple to have a small buffer. The Items/time setup would manage a small buffer with either a longer time scale or more frequent reading sections.

plain hare
#

You can have a small buffer with active sushi but it’s only as small as or more than what you’d have with a memory cell.

timid cosmos
#

in theory memory cell designs will have a very precise reading on the buffer, that's true

#

but i don't think it's worth having to read every inserter that takes something off the belt

#

active sushi designs you don't have to read any inserters that take from the belt, only those that add.

plain hare
#

I’d agree the active sushi seems to have more use cases or more easily constructed use cases, but what exactly is the ‘cost’ of wiring Inserters? All those wires are free given a blueprint.

timid cosmos
#

It's a lot less inserters to set settings on

#

thats the main advantage to me lol

plain hare
#

Potentially, I dunno. My first sushi designs were memory cell and I made a standardized cell for the assemblers taking from the sushi belt.

#

This meant I only had to change a setting or 2 to set it up.

timid cosmos
#

yeah same

plain hare
#

I think the only thing I had to put in was a constant combinator acting as a request for items for the sushi belt.

timid cosmos
#

I haven't been working on this lately but when I get back to it I think it'll be a solid design

sushirail.JPG
plain hare
#

Hmmm

#

I think it may struggle with proper flow directions

timid cosmos
#

what do you mean?

plain hare
#

I’ve always made mine with priority outputs going to the right always, to ensure one flowing belt instead of half going one way half another way

#

The way it’s set up right now it gets the opportunity to go straight first, then left, then right

timid cosmos
#

It should even out over time

#

I am going to clip the end belts on that 4-way

#

or maybe i'll leave them

#

it all balanced out in time

#

I need to design a "capper". The part I can stamp down to start the system with inputs and combinator controls

#

Having the loops will mean a slower build time, but a faster balance

#

in theory

#

Maybe I should design a mall that just builds this stuff

#

@plain hare do you put stuff like miners on the belt? stuff to build outposts?

#

Or just stuff for the rails, and then send construction train?

#

I guess you could do both, but building outposts in this way I imagine would be really slow

plain hare
#

So for early game I’ve got the on demand requests set up, then later game I’ll set it up to just call in the construction train.

#

Oh and also I can at any time change the permanent contents of the belt from the sushi mall so if I’m unhappy with latency things I can put miners and eff modules on the belt permanently for instance.

timid cosmos
#

yoo. sounds like you're about ready to set up the full automated beacon swapping

plain hare
#

Well I did get the first test of the on demand items going yesterday. It seems to work as expected.

#

Oh also got the sushi smelter built - one cursed beacons furnace column usable by iron ore, copper ore, iron plate.

#

My next plans are mostly finishing off mid game cursed beacons builds, finishing the sushi starter base to rush robots. Then some testing. Then a new bot mall for higher throughput of building materials, and red belt construction sushi rails

hoary trail
#

It will run smoother and have slightly better output though if there's at least some buffer, due to belt transit times (this can be alleviated with red or blue belts but I feel that a larger buffer size is usually more affordable than those belts for my purposes)

plain hare
#

Nothing would move at all

#

You actually do inherently need some amount of buffer for active sushi to work, in my opinion. It’s defined by measuring what’s on the belt and adjusting to that.

hoary trail
#

It's always a negative number on the combinator 🤔

plain hare
#

Oh I see what you mean.

#

Well that would be buffering 1 or 2 items on the belt.

#

Which would be, in theory, equal to or less than what a memory cell would buffer.

hoary trail
#

Well, an inserter buffers an item even in direct insertion, so I don't consider one item a buffer haha

#

It's just on a belt instead, but it's still only one or two items

plain hare
#

Inserters don’t buffer items 😛

#

Assemblers do tho I’ll be fair to you there

#

However you can cut that down, with memory cell you can measure the output of the assembler, and only request inputs when the output lacks.

hoary trail
#

Inserters do buffer an item if the target is full

plain hare
#

So that would mean zero buffer of the ingredients (only some of the output)

hoary trail
#

Which is what the belt would do as well

plain hare
#

If the belt is full. If the assembler is full they don’t grab.

hoary trail
#

Well, true

plain hare
#

And we are taking about inserters taking from belts right?

clear latch
#

You inherently need some buffer to sustain high throughput at any given point in time though, I just call it a day if I'm not buffering more than needed to achieve that

hoary trail
#

Yes, so technically there's a bit of extra buffer with the extra inserters

#

Can't beat direct insertion for low buffers

plain hare
hoary trail
#

But anyway, my point is that you can do a low buffer with either a memory cell or with active measurement.

plain hare
#

You don’t need a buffer with memory cells though!

clear latch
#

There are a couple of very hoggy recipes with such high assembly buffers that they actually hurt throughput, like nuclear reactors and rocket silos

hoary trail
plain hare
#

Well that’s fair but that’s just like… the comparison you set up right. Inherently, the advantage of memory cells is the ability to reduce buffers

#

How much difference the buffer is compared to other designs? An open question.

#

However the question of why you’d use memory cells in this day and age compared to another design style -> answer: lower buffers

hoary trail
#

Well, I feel like active measurement also has the advantage of reducing buffers, almost just as well as memory cells

plain hare
#

It does reduce buffers, but in comparison to what

hoary trail
#

I'm just saying that I don't see hardly any advantages of memory cell over active measurement

plain hare
#

Systems where I’ve seen memory cells work well: artillery shell delivery, mall item delivery to building trains.

#

What if you wanna deliver say nuclear reactors to your building train?

clear latch
#

Yeah active measurement is not enough for trains, once they start moving goodbye read contents #combinators message

hoary trail
#

One application that works well with memory cells is long distance requests

#

And yeah, trains is another application too

plain hare
#

Yep. Long distance requests of expensive items.

hoary trail
#

But long distance makes me nervous because it's far easier for such a system to break if it's run by a memory cell

plain hare
#

Oh absolutely. Memory cells are rife with issues I won’t lie

#

But similar to the bloodbus category I fought to include it so eventually we’d have conversations like this one and eventually maybe use them again 🙂

hoary trail
#

That makes sense

#

I do think some kind of checkpoint or trash disposal/reset could make memory cells viable

#

But I haven't figured out how I'd want to implement that

clear latch
#

I think the potential problem is more that the reset mechanism may itself not be 100% robust - say you use train returning as reset and it gets destroyed by biters, now the cell is stuck unless you included a manual reset or some kind of contingency timer, and a timer could break in other ways

plain hare
#

Considering most breakage is in fact caused by unforeseen events, a manual reset seems necessary.

#

However some at least partially automated resets seem viable.

#

I like the idea of putting a hormone item on the belt as a marker for checking everything on the belt and then resetting that counter

#

Like a fish that goes around and around forever, every time it comes back to the start you have a list of what remains on the belt. You could for instance clear everything off the belt each rotation, reload and reset.

#

I think making a system that actually does that in practice is easier said than done.

hoary trail
#

I like the idea though

#

Although at that point I feel like trains are probably more viable although it might be less automatic

plain hare
#

That also requires a belt loop - but one of the advantages of the memory cell sushi was maybe not needing a belt loop. So there’s that.

#

I don’t see how trains necessarily relate here?

hoary trail
#

I mean for long distance requests

#

Rather than using sushi

#

You can still use memory cell stuff with trains, though

plain hare
#

Ah right yeah. That’s true. Just replacing the medium really.

#

Ultimately tho, I’m probably not gonna be using memory cell sushi any time soon 😛

hoary trail
#

Same

plain hare
#

Although… I’m tempted to install it in my nuclear reactor design.

#

Right now I’ve got active measurement sushi but it buffers a bit more fuel cells than it needs to when it goes through underground pipes and stuff.

#

Although that’s only so relevant - goal is to get kovarex faster, but this makes maybe a 1-5 U235 difference. Not much.

#

And once kovarex does unlock you just eat through the U238 buffer anyway.

#

Tbh the only good idea I have for memory cell sushi is a way to deliver items from a belt mall to a building train to an outpost with no excess buffering

#

I just chose not to do that lol cause ew

#

That’s hard

hoary trail
#

Yeah I always try to avoid undergrounds as much as possible with active measurement, although you can avoid that issue by just lowering your limits on the combinator

#

I hadn't thought about using sushi for my nuclear designs, that's interesting

#

I wouldn't want it for koravex though as I use a design that's very easy to extend, and active measurement is not easy to extend

plain hare
hoary trail
#

I usually just try to minimize belts, or I'll wire the input inserters to the chest

#

I particularly like using bots instead of belts when moving expensive things like fuel cells

#

Although I can't do that when I first build my nuclear

plain hare
#

True but, bots do take yellow science. Yeah.

#

That’s the niche where sushi belts come in

#

First reactor - poor on U235, want nuclear energy but also wanna save U235 for kovarex

hoary trail
#

I usually just build several centrifuges so I can send one to my reactor and the rest will save u235

#

I also don't really mind having extra fuel cells as they're really not that expensive

#

Within reason, of course. A full cargo wagon of fuel cells is likely a tad overkill

hallow sonnet
#

what is this?

#

woah

#

is like a chat inside a chat?

plain hare
#

It’s a thread that’s just stuck around yeah

#

Scroll for the insider sushi secrets lol

#

Quick! Post sushi belts!

hallow sonnet
#

woah

#

sushi

#

that base looks really good

timid cosmos
hoary trail
#

That's an old picture but it shows a bunch of sushi belts at least

hallow sonnet
plucky isle
#

Wrong type of sushi

foggy junco
hallow sonnet
# timid cosmos

I really like the look it! its so good for some reason its so satisfying to watch

hoary trail
#

There's some really cool things you can do with it

#

It's not good for everything though, and if your items are already completely filling your belts sushi clearly doesn't work

foggy junco
#

main place where i use sushi is construction

plain hare
#

construction sushi belts intensify

timid cosmos
#

@plain hare do you think defense is going to be necessary?

sushirail2.JPG
plain hare
#

I would suggest maybe a landmine wall that’s optional. (Blueprints with and without)

#

The thing doesn’t pollute but geesh it would suck to let aliens take it out.

timid cosmos
#

yeah...

#

Maybe I could do minimal defenses on the inside

plain hare
#

It wouldn’t break in any significant way if they did get on top of it, but the system just seems destined to take care of its defenses as well.

#

Ooh yeah yours is small enough for that to work actually

#

A few gun turrets, small ammo / min

#

That would be cheapest

#

And smallest item / min on belt

#

Landmines end up taking a decent chunk of the sushi capacity

timid cosmos
#

This is something I was working on a couple weeks ago, i forgot about it

forgottendesign.JPG
plain hare
#

That seems unnecessary to me unless you put flamers in there

#

I feel like land mines would make a better ‘wall’

#

3 second stun 👀

#

It does look really good tho I must say.

#

I wonder if 4 flamers could fit around the roboport

#

Or 3 at least

timid cosmos
#

Just barely can't get fluid to them

#

Almost works perfectly

#

and would require medium poles in the center

#

I could do 3

timid cosmos
#

@plain hare check it out I've created a circuit priority system. If the chest needs stuff, it cuts off the flow from continuing and forces it to the chest. If the chest is satisfied, it pushes the stuff on down

Factorio_Screenshot_2022.02.05_-_00.09.49.59.png
timid cosmos
foggy junco
#

On my bp, i just put two turrets or so per segment

#

Better than no defense at all where i have to watch biters chew away at my sushi for free

plain hare
#

Okay sushi things

#

Well I have a new circuit to implement that’s gonna take rate limits past the belt capacity and transform it into a perfect 900 / 1800 / 2700 items / minute output belt

#

Also, the global wires in my rails carry the rate limits

#

So not only will it be easy to have one central rate limiting area, but even if I did double the rates it would just work anyway

#

So: (item request * belt capacity) / (sum of all item requests) = adjusted safe sushi rate limit

#

So say you ask for, one item at 1000 per minute, one at 2000 per minute, one at 3000 per minute, 4000 per minute, Normally this just fucks your sushi belt. With this circuit you’d see new rate limits of (blue belt) = 270, 540, 810, 1,080

hoary trail
#

So if you have two separate rate limiters for the same item, they'll automatically share the total rate?

#

Also ngl this seems so insanely complicated haha

#

I'm reconsidering whether I want to actually attempt any form of sushi rails or just stick with simpler things

plain hare
#

Tbh the sushi rails are not that complicated. It’s just a very large scale rate limiting setup, which iirc you know how to do?

hoary trail
#

Well that's true

plain hare
#

I have added a number of smaller systems but it’s mostly extra stuff and each system is kinda self contained

#

That being said before you make your own sushi rails you should see what @timid cosmos is up to they’ve made a new sushi rails that has 1 roboport per train grid segment

#

So it’s possible it will just be a lot better than mine as an individual system lol.

#

A lot cheaper I think.

#

But I’m sticking to my guns with the 4 roboport models. I think they may have a better late game use case when I start making proper sushi castles

hoary trail
#

That makes sense

hoary trail
#

I updated my yellow science sushi

unknown.png
#

it's not quite as efficient as some of my previous designs, but I think it's good enough for my purposes

hallow sonnet
#

why is there every belt signaled?

timid cosmos
#

It's a technique to make sure the belt doesn't overflow. If you read every single belt, you can read everything on it, and ensure it doesn't fill up

clear latch
#

@normal gulch Threw together an example of the moving average counter from calcwizard in case the new base also uses flow limit sushi. Only 4 combinators + CC, but it has a resolution problem where if you take a 1 second average it's responsive but only works with 60/s items total, and 1 min average 3.6k/m but then the belts jam unless you have a large buffer to absorb (like a train wagon) until it settles down. Might be better to just wire all belts that early tbh

hoary trail
#

@plain hare the 60 spm "starter" base is complete and in working order!

#

(I apologize for the double ping, I uploaded the wrong picture the first time)

#

It's interesting how all this sushi made my bus a whole lot shorter

#

There's a total of 8 different sushi designs (12 individual sushi belts in total; the red circuits design is used twice and it has two sushi belts in itself, and the yellow science also has two sushi belts as well)

plain hare
#

It’s so colorful with all the circuited belts 🙂

#

Put some prod modules into your processing units, purple science, yellow science, acid plants wonk

hoary trail
#

Yeah I know, I need to add prod modules to lots of things

#

I just finished some other major projects - got a 4-reactor nuclear setup running, moved my labs off-site, set up some bot networks, and cleared some surrounding land with my new laser weaponry

#

My purple science is currently bottlenecked by stone bricks, which was unexpected. I think I need to add a stone brick train station

#

I also need to make a rocket fuel production station as I have now secured my top border (I think)

clear latch
#

I made a picker that only picks up requested items that are actually on the belt. Output 1 for items missing from logistic network, output 1 for belt items, set filter for items that = 2. Was going crazy watching 30+ items being prioritized by item ID order over rails and poles lol

unknown.png
plain hare
#

@stuck scaffold you made it to the sushi thread!!

#

Don’t let your memes be dreams, join the sushi train! (Or should I say sushi belt?)

#

That’s a funny expression cause it works both ways, you can also say don’t let your dreams be memes ha

clear latch
#

There's probably a way of setting dynamic stack size without using like 6 combinators total, for now I'm just using separate inserters for big and small requests and duplicating the circuit

unknown.png
stuck scaffold
#

Don't let your memes be dreams

unknown.png
#

Honestly I'm going to switch to liquefaction and just not deal with liquids at all I think

#

Oil is too many items

plain hare
clear latch
#

I just realized those are different barrel icons, and the right one looks blue 😂

clear latch
#

Finally got a working version going but it feels really slow. I could probably cut roboports almost in half to get rid of the concavities and request more bots so there aren't 20 spread over 4 roboports lol

unknown.png
plain hare
#

Yeah my super early game one has 20 bots but it can be sort of slow as they place 250 landmines. As you progress you can boost the construction speed and spend more on each section.

clear latch
#

Figured out why it was slow...the belt loops reversed orientation in an alternating pattern. In this example they insert correctly only on the "near" side, and skip the "far" side inserters until looping all the way back down. The one I used had it the other way around so the far side inserted first lol. Still beats the earlier versions that either never looped back home (L output), never went out (R output), or always split the flow in half (no output priority). Worst case scenario I could double inserters to pick up from either side but that would probably suck for UPS. Probably best to just merge 2 roboports and have 1 inserter side only

unknown.png
clear latch
#

Learned a couple of unintuitive things while making it build faster. 1) Inserter count was the bottleneck, peak bot usage for straight rails was only 30-ish. 3 stack inserters can empty a 1 item blue belt, they won't pick up 30+ different items at anywhere near that throughput since they can only take 1 item type at a time and it's mixed randomly. Worse, they can be stuck picking up common items while the rare item whizzes by, not to be seen again for a long time. So not only 3 filters > 1 stack filter, but often 1 filter = 1 stack filter. 2) Appropriate requests weren't 1:1 to blueprint material costs; that caused the scarcest items to bottleneck construction. Equal requests all around always underrequested rails and belts. I ended up weighing everything below a "too scarce" threshold the same and everything higher as they were, then multiplying up to (modded) belt capacity. Worked well for straight rails but the city block skeleton with 30+ items might need a second lower threshold and dedicated inserters so it doesn't bottleneck the rails

unknown.png
plain hare
plain hare
#

Let’s assume a 1 minute production target per blueprint. So if you had say 300 belts in the blueprint, and you had 20 in a chest for construction, and you can expect 200 of the belts to ‘buffer’ fully with moving sushi items, then those 200 belts would end up having on them, if a blue belt, (8 * 200 * (item / minute on sushi) / 2700), so if you set it at just the needed amount, 300 + 20, 320, you’d end up buffering another 189 belts on the sushi belts themselves. This means construction would take 35 seconds longer than expected. But if you overcompensate and add that into your desired rate on the sushi belt, 300 + 20 + 189,well then you’d also buffer more. since you sent more on the belt. So now you’d have 302 belts sitting on the buffer inside that blueprint. Now, construction would take 300 + 20 + 302 belts, or 622 belts. And you’re sending them in at 509 belts / minute. So now construction would only take an extra 13.3 seconds. So you see how it’s tricky to find exactly the needed rates xD

#

Rate needed = (blueprint fixed cost) + (rate needed * 8 * # of full sushi belts / belt speed )

#

Okay I found a better equation I lied about some of that stuff up there it’s not as bad as I thought

clear latch
#

I'm imagining that this will be more complicated for my bp when factoring in the inserter count, swing speed, behavior of them cycling through any filters, construction ghost queue limits...I've seen them just pause inside the roboport for several seconds when they're in the network and the stuff is in the chests, almost like they're processing how to build next. Sometimes they even stop building a segment while another cluster of bots halfway across the map starts up, and the number of bots/materials is not huge compared to solar where they build almost nonstop, maybe it's a problem with multiple networks idk

#

What I did so far was brute force flow rate requests to be as close to 3.6k (modded blue belt speed) as possible and then put separate chests requests in another CC to 4 items of every type (modded filter inserter capacity), at 3 inserters per segment that was only 12 rails or belts picked up at a time so I bumped those again to 16/8. Flow rate requests are dumb and blind, chest requests go on a global wire to the rails being built

plain hare
#

Well I’ve officially confused myself trying to write and solve equations for this one lol.

clear latch
#

I don't remember anything from calculus tbh, I saw it as a statistics problem where you are forced to take samples because you have less inserters than the total number of unique items. So the common stuff is more likely to be grabbed and the rare stuff less likely, and just like in a gacha game missing the rare stuff can feel worse than expected because of variance xD

#

Like power poles, combinators, roboports...I actually stopped using CCs because I realized a global wire could double as a CC without having a ghost CC bottleneck a whole segment

plain hare
#

Right yeah, that’s definitely another element of the problem, I was more considering exactly how many items should we put on the belt, assuming everything on there finds its home where I should

#

Ah that’s pretty smart, just re using the global sushi belt contents to double as a buffer capacity for the blueprint itself.

#

You could probably just take whatever the global belt item rate contents are, Divide by 10, mod by 10, add those up and give that to the inserters as conditions to activate.

clear latch
#

No the global wire just tells each segment what to grab for itself haha, I don't keep live track of total items on the belt at all since it's just a dumb flow rate at the source

plain hare
#

Right, I didn’t mean a global absolute count of items but a global count of the flow of items, the count of items / minute on the belt

clear latch
#

Interesting what I actually did at first was use Ghost Counter to get a CC of all the items from a straight track blueprint, then sprinkle in a few chain signals for intersections, then arithmetic add+0 => total, 3600/total, each item*total = requests. I still do something like this except I fudge the lower numbers higher and have a circuit adjust the higher numbers lower to not go over 3.6k total

#

The other approach I tried was taking a whole city block blueprint and dividing by number of segments in the block. Then I had some problems with too much variety of exotic items that went in the city block train stops but not the rails and only 1 chest inserter that prioritized rails/belts too low. So I'm brute forcing with 3 inserters per segment for now lol

plain hare
#

Yeah there’s definitely an interesting challenge to be met on the pickup side of things.

#

My original sushi castle actually performed well here because I super brute forced it with 1 filter inserter per unique item on the belt , some even had more than 1 (belts, robots, etc)

#

I think I may look into a signal cycler for my next attempt, that would rapidly change the filter signals the inserters get.

clear latch
#

I actually only need exactly 1 AAI Transmitter building per city block so the builder train can transmit ghosts back home lol...I have at least 8 segments per block and that's not exactly a cheap structure to buffer haha

#

There are so many 1-ofs like that, and they make stack filter inserters look slow

#

I'm working out a slave inserter array that shares requests between all inserters while prioritizing the belt items near specific inserters but it looks like it will take a lot of deciders. And the more cute little steps that need building, the more unreliable it is until every last piece is built, bots really suck at that lol

unknown.png
plain hare
#

I think at some point you gotta consider every item placed into this has the same value as far as it’s slots taken on the sushi belt, so at some point it’s worth adding more inserters and just giving them each 5 filters from a cc compared to like, a ton of deciders. Hard to say.

#

I wonder if spacing out the inserters would help a bit if they’re sharing signals as well. By the time one Inserter swing finishes, the next one up the line could have modified its filters accordingly with the signal being sent that the chest has its item

#

Whereas if 2 inserters are right next to each other, important must grab items may have passed on because a filter inserter didn’t have its filter ready.

clear latch
#

Yeah that would only take 1 decider per CC group. The problem is the set filter takes its sweet time cycling through page by page of the recipes and sometimes the items further back by id are right there on the belt lol

plain hare
#

Lots of interesting challenges with the sushi rails! XD

plain hare
clear latch
#

Oh yeah spacing inserters out reminds me of building train inserters, they also had this problem of inserting too much because all of them swing at once. But a belt inherently adds a bit of a delay compared to a shared prebuffered wagon

#

Wouldn't a signal cycler also "miss" more and more as the number of requests gets higher and higher?

plain hare
#

It might miss a fair bit, I dunno. I haven’t attempted it.

#

Maybe a setup with basic filter inserters would be best, simple and each could have a dedicated 5 filters set to it.

#

You usually don’t get great value out of the higher stack size picking up off the sushi belt anyway

clear latch
#

Yeah I just like the idea of all 5 of them reacting to a belt segment that's all rails and all picking up at the same time right then

#

Watching my belt it's not very evenly mixed and there's a lot of clumps, sometimes big ones if I don't choke the kitchen output belt to stop it, and choking hurts flowrate

plain hare
#

I think at blue belt speed you have 6 ticks for an item to exist on a belt

#

Interesting, my sushi belts usually end up very mixed up.

clear latch
#

My blue belts at 3.6k items/minute so it can be very feast or famine with uneven mixing xD

#

One second it's all rails and then no rails for several seconds for example

plain hare
#

Huh, interesting.

#

I guess the timers we chose must operate differently for the circuit controls, i remember explicitly trying to get fireball and calcium wizard to make a timer that wouldn’t bunch up items

clear latch
#

Like technically the flow rate for rails might be high, but that's an average and locally the variance can screw it up

plain hare
#

Yeah, there’s a decent variance between straight segments, and intersections, as far as rails needed and chain signals needed.

#

(If that’s what you meant)

clear latch
#

No I meant clumping still - averages out, but locally it clumps

plain hare
#

Oh, that as well then ya.

#

Hmm maybe we are going about this the wrong way.

#

What if we used UNFILTERED inserters to grab everything when all the requests aren’t met, then use filter inserters on the other end of chests to take out the excess?

clear latch
#

I tried a version of this yesterday actually with belt => loader => storage chest => loader => belt but no inserters...would have to be a merged chest or 2x2-4x4 warehouse this time to empty it to avoid bottlenecking by inserter throughput

#

Yeah that was when I gave up on 1:1 requests, the common stuff was jamming the chests and the belt just shut down ahead of it...and the rare stuff still wasn't showing up lmao

plain hare
#

Alternatively, we could try using the ‘blacklist’ on filter inserters loading the chests, might let us tell them what not to grab. (Would that mean they’d be better at grabbing rare things?)

#

So many choices! XD

#

What do you mean by 1:1 requests btw?

clear latch
#

Request 1 item in the chest for every item strictly needed to make its segment bp

plain hare
#

Oh oh. Yeah that’s quite the buffer potentially as well.

clear latch
#

Well later I used that to upscale calculate flowrate and used 1-4 inserter swings buffer for the chests. But it's what led to all these issues with common/rare items and trying to find a good compromise

plain hare
#

One of the considerations I used for choosing buffer was to see how many items were needed to create the intro to the next sushi section

#

So like how many belts, inserters, splitters, power poles etc to get the next ‘sushi mouth’ operational

#

It’s hard to find an ideal way to do this stuff.

clear latch
#

Like just this stretch basically? The part where the next segment can start building on its own?

unknown.png
plain hare
#

Yeah, for me with 2 roboport deep segments, there was a key connecting section from 1 roboport to the next, and all those belts needed to get placed before the new roboport areas bots could start working.

#

Of course, part of that consideration for me was that the new section I refer to was getting built into undefended territory - so it needs to go up ASAP so the castle can build.

#

I guess some of the other considerations that go into the inserter sushi belt pickup lottery - what items are essential to continue sushi construction? Which are not? So for me, I don’t mind as much if the rails / signals lag behind a bit. I’m gonna assume one of the primary bottlenecks is the player actually remembering to place new prints, so as long as the time until you can place a new print is as low as possible, the internal construction speed as far as track, signals, construction equipment, that kinda matters less in a way.

clear latch
#

That's fair, I haven't tested this in a real save yet and there was definitely a point when I requested the same ratio of all items that the "brain" was zipping way ahead and left rail ghosts in the dust haha

plain hare
#

Ah. Yeah, I guess if the difference is drastic enough it could definitely add up to be an issue.

#

All these considerations hurt my head haha

clear latch
#

I kinda put this on myself by deciding to design everything backwards from this save. A lot of this inserter stuff is borrowed concepts from any filter builder trains. I still need to make an explorer rail leading up to these blue belt blocks, not sure if I can afford blues right away, might be possible with modded wood recipe and deep core mining

plain hare
#

Yeah, in my sushi castle mk2 I realized the early game potential and chose to design around minimum belt underground distance with yellow belts, which should be nice as I can smoothly just upgrade planner belts to get the next tiers going

hoary trail
#

And then there's me, who has never even built a blue belt before haha

plain hare
#

(That may be for the best) they’re very expensive. The kind of thing you wanna have a spare iron patch to start making them.

hoary trail
#

That makes sense

#

Now that I have a spidertron for my war endeavors I can definitely obtain several iron patches that I had been waiting on, if I so desired

#

Even without any rockets, spidertron is so op lol

plain hare
#

Dooo eet 😄

#

Oh yeah especially when you’ve been going low evolution / low pollution!

#

What’s your evo at?

hoary trail
#

0.64

plain hare
#

Lmao those poor poor aliens

hoary trail
#

Lol

plain hare
#

I was playing around with the spider and found I really enjoyed having manual control of the rockets with plds

hoary trail
#

I'm just trying to decide how to structure my expansions now, before I actually start building better production

plain hare
#

That lets you barrage the nests without self damage

hoary trail
#

I also need to train in some petroleum oil to my main bus, as it's low on petroleum and I don't know why, and I'm tired of messing with the oil line haha

#

I was wondering about that self damage aspect

#

My solution has been to just not use any rockets at all

#

My plds melt a nest in a split second anyway

plain hare
#

Well, the yellow rockets won’t do any self damage. But they seem boring I like the booms

#

Mm yeah the plds will start doing worse though as more and more bigs show up, and then behemoths to follow up.

#

Of course maybe the nests are small enough it won’t matter

hoary trail
#

Yeah, for now the plds are way overkill

#

I've basically maxed them out with my prerocket science and I think I'm running on 13 plds total

clear latch
#

When I just have 1-2 spiders I like to trace a vanishing spiral into the nests on the minimap with the remote so they avoid the worst of it and don't get chain slowed too much. Explosive rocket or modded equivalent full auto feels like too much clear speed to pass up and a few shields is usually enough for friendly splash

hoary trail
#

Hmm, I didn't think about shields, that might work well

#

I don't have a lot of space on my spider though because I have 3 exoskeletons and a fusion reactor

plain hare
#

Oh yeah, the shields definitely work well for stopping the splash.

#

I’m sort of aiming my starter base plans at a spidertron stage to just crush all the low evolution aliens and take all that juicy juicy ore before behemoths come out and I have to slow down and switch back to artillery

clear latch
#

My endgame actually is spiders but it depends on certain mods to be viable late. Initially I'll put a tank on the explorer rail outbound belt and make some cheap spiderlings follow it; eventually I'll abuse the mod settings to let them fire nukes without too much collateral damage 🍿

hoary trail
#

Spiderlings?

clear latch
#

Spidertron Patrols adds those; they're cheaper blue science spidertrons. I'll also be using them with Spidertron Weapon Switcher to get them a variety of weapons. For example MG doesn't do friendly fire, flamer is kinda cheap and good up close, cannon frankly better than rocket, etc., all good options before nukes

hoary trail
#

Those sound like they would be op

#

I already feel like spidertron is op for my evolution

clear latch
#

They are very OP with the right AI, but physically they only get 1 weapon slot and have a lot less hp, grid space, and inventory space

hoary trail
#

I see

clear latch
#

It's kinda funny what sushi building lets you get away with. Celtics and half-celtics (3-way cutouts) basically require a curved midsection to overlap properly and the extra side belts take up annoying amounts of space. 3-way can be a bit more compact but it won't fit a 4-way. Since there's roboport coverage on both sides I could default 1-way for long distances and deconstruct/build the better ones when I feel like filling out the grid. The only problem is trains freaking out lol

unknown.png
clear latch
#

Local CCs are starting to make a lot more sense now, there's no reason to buffer chain signals in the 1-way straights, in fact there's no reason for every block to be compatible with 4 parking lots, I could have a version with 3 straights and a 4-way that unloads to a neighboring block too

plain hare
#

Sushi rails… absolutely glorious

#

And yeah I think local constant combinators are my favorite

plain hare
clear latch
#

I got a plan involving the crudest use of recursive bp 😄

unknown.png
plain hare
#

Lol just to zap it and replace?

clear latch
#

Yep

#

Could probably also send the other track bp over global wire but that's harder to work out for now

plain hare
#

Truly insanity haha

clear latch
#

So ideally I could dial-a-CC, enable a certain CC to switch to that track

plain hare
#

I love it

clear latch
#

But so many combinator spam everywhere that way xD

plain hare
#

I do try and reduce my combinator spam where possible

#

(Makes sushi smelter with 1 combinator per furnace) lol also me

clear latch
#

The biggest improvement for that was wiring up the belt to redirect flow to 1 set of inserters instead of 2 for requests and bots to catch either belt direction redundantly...that went from 16 deciders to 7 xD

clear latch
#

Slightly better recursive bp abuse: getting rid of the inserter blob after the bots finish building the inside of a city block. If I change my mind I can place another blueprinted blob later since it's in construction range of a rail segment

hoary trail
#

@plain hare I tried to make sushi work with beaconed builds, but so far the sushi always has significantly lower throughput than belt weaving. I think it's because beaconed builds tend to benefit from trying to compress the belts more. Here's some of my recent beacon projects all working nicely on my current map

unknown.png
#

You can see that when I got to the one on the left I had learned some more compact unloading techniques haha

#

It's less expandable, though. But I don't think I'll need faster artillery shells for a while, and if I do, I think I can belt weave to double my production at least, which would be 100/m. I imagine that would be good enough as I use the spidertron for expansion. The artillery are just for static defense

clear latch
#

Lazy sushi explorer rail defense, should be vanilla viable too with regular spidertrons. It's definitely cheaper than a yellow belt + tank, and the train can even be automatically replaced if destroyed since it's blueprintable. Haven't figured out how to make it leave based on train count/ID if it actually stops at a disabled train stop...maybe it thinks it's stopped properly as long as any stop ahead with the same name is open 🤔

hoary trail
plain hare
#

This is the sushi thread dammit I expected sushi!

clear latch
#

The undergrounds are great throughput, but being undergrounds, wouldn't work for moving a vehicle. And a separate belt, even a yellow one, would feel like a waste when there's already rails there. This literally just staples on some stops and preserves the regular signal blocks

plain hare
#

‘Sushi was supposed to defeat the spidertron! Not join it!’ Duel of the fates starts playing

clear latch
#

Actually this whole track is designed to work with them...that enemy detector top gates a special CC to request ammo and logistic bots just for the spiders, and along with the resource scanner pauses auto construction. No mines or other static defense at all, I'll just rig an alert when enemies are detected or 50M+ ore is found

hoary trail
clear latch
#

The idea is to invest as little construction time as possible and auto build with recursive blueprints in a straight line until I get to the good ore, then start up the "civilian" city blocks with a spider sentinel perimeter and eventually decon the old base, rolling decon the umbilical rail, delete empty chunks for those marginal RAM savings. So the same sushi belt that builds this will one day be a giant trash belt. Mods definitely feel key to automating spiders, I'd recommend spidertron patrols if sentinel seems too OP

plain hare
#

It’s like my plan with more mods involved xD

timid cosmos
#

why does this channel randomly appear and disappear?

desert tide
#

It's a thread, it disappears when it's inactive and reappears when someone talks

hoary trail
#

Sushi isn't cool enough to have its own channel

foggy junco
#

i believe threads get archived whenever no one types in them for 48 hours, or something

#

then you have to go to the channels threads, on the "archived" tab and select in manually

distant gorge
#

I'm starting to think sushi might be the easiest way to deal with some stuff in Angel's at the moment

#

like... I feel like having a counter for all this stuff and have the outserters contingent on that would work better than relying on backpressure everywhere

unknown.png
#

I'll try the normal way first though I think

plain hare
#

Those are pretty small item counts you could probably do a super simple active measurement sushi loop measuring the entire loop with circuit wire and inserting when each item < some count

#

Takes a little optimizing the count for each item but it’s easy on the mind

distant gorge
#

the tricky part is figuring out which output is actually the trigger for each recipe

#

some of them are clearly going to always work without controls, the neptunium conversion and plutonium recovery namely

distant gorge
plain hare
#

Ah yeah, makes sense. Maybe try and include a little manual reset switch or something

#

I was just always breaking my memory cells left and right

distant gorge
#

I ran the equivalent thing on my train system before train limits, should be fine

#

in hindsight it's almost exactly the same system actually insertereyespy

plain hare
#

I think my problem was I was using it for a mall and then I’d expand it….

distant gorge
#

trains on a rail = items on a sushi belt

plain hare
#

Broke it every time lol

#

Oh yeah that’s wild, I never did get that creative before train limits

#

Just had huge stackers and disabled pickups..

distant gorge
#

I had a working version of this nuclear thing before the mods updated, and now there's another isotope I have to deal with and some different ratios

#

It's probably going to be.... I'd guess around 1% of the entire factory I'm planning

plain hare
#

Yeah you’ve got 4 different fuel cells there

#

Does that mean 4 reactors?

distant gorge
#

way more

#

well, not different reactor tiers, no

#

but there's a fast burner reactor specifically for burning off cells when you don't care about the power

plain hare
#

Oh that’s nice, I was wondering how youd balance this

distant gorge
#

What I actually care about is the muons you get from reprocessing thorium cells

plain hare
#

Like a uranium flare stack ha

distant gorge
#

kinda haha

#

The plan is to void nothing but nitrogen gas at this point

plain hare
#

Oh wow that’s fun, all the byproduct management is the best part

distant gorge
#

yup!

#

especially when it's self-contained puzzles like this one

#

mostly self-contained anyway

unknown.png
plain hare
#

10.1 GW

distant gorge
#

that's an upper estimate based on no beacon sharing

#

I have much more built than this and am only pulling 4.3... it's not that much lol

#

OH lol

#

the cheat recipe for getting the fuel cell burning to show up in YAFC is consuming energy instead of producing it

#

that's how much energy the thorium cells are making

plain hare
#

Ohh makes sense

distant gorge
#

and a bit from the mixed oxides

plain hare
#

I was gonna say it was only a few machines…

#

I was beginning to wonder what kind of cursed modules your mod has ha

distant gorge
#

nope, no Bob's :p

#

actually, I could get 16.5 GW out of the thorium cells, before reactor bonuses, if I really wanted to

#

probably will need to do that by the end to power things, since I'll be running reactors anyway

foggy junco
#

how does something like that happen

clear latch
#

I would guess no filters in the blueprint when placed or something wrong with inserter conditions (including logistic connection). Assuming the setup is individual inserters wired to individual chests. I've mostly run into this when wiring to roboport or belt through bad circuits that leaked signals from stuff not on a CC

foggy junco
#

the wire only disables the inserter

#

there's a filter on it by default, the storage chest is filtered as well

normal gulch
#

bots took coal there?

#

or there was coal on the gruond and a chest built onjtop

hallow sonnet
#

Sushi legacy!

foggy junco
normal gulch
#

Perhaps idk

timid cosmos
#

that only happens with belts. with a chest, a bot will remove the coal to place the chest

#

i think

normal gulch
#

this happened when I put stone on the ground

unknown.png
foggy junco
#

Good to know

timid cosmos
#

if you do the "drop" action into a chest of course it's gonna put stuff in there, but if you build a chest on top of that stone, it'll go into your inventory

normal gulch
#

no, i put a chest on the ground and it put stuff in the chest

#

uh now thats not doing it 🤔

#

Ok, only if a bot places a chest wehre stone was broken from a rock

#

or any item on the ground

hallow sonnet
#

on phone so just theroy:

sushi starterbase

limitet to belt throuput right? so what do i put on the sushi belt and what not? i think everything but smeltet ores and greens right? but then throuput will be the issue right? as i cant have like 7,5 reds going into blue science and still provide materials on the same belt for all the other crap?

#

maybe you have an idea of what iam thinking about @plain hare

distant gorge
#

For fun/academics, or actually trying to be functional?

hallow sonnet
#

functionality + my first try with it

plain hare
#

I’d suggest as a first gambit with it try and make a sushi lab setup: it reduces buffer by having 1 belt with every science on it, and it’s not too ambitious as a first project

#

I would not really suggest 1 belt with everything

hallow sonnet
#

sushi is pretty esay it wont be hard to do labs

plain hare
#

Yep, it’s a good first step into the wonderful world of sushi belts 😄

#

If you want specific design advice I can throw you some tips as well - there are many ways to get a sushi belt functioning

hallow sonnet
#

my plan is buffering somewhere everything and have then inserters connectet with the labs sushi belt around it only when something is missing they get activated to refill buffer and output on the sushi belt

#

my only concern is throuput cuz iam thinking too big

#

i want 60spm and 0.5 modules per second

normal gulch
#

use factoriolab to check if u have enough throughput

hallow sonnet
#

its not about doing it on one belt its just the organization where to put what so they have short travel time

plain hare
#

Other things to consider: sushi has certain early game advantages - I’d argue theoretically easy to hand build (if done right) and the second big one is low buffer - Reduce costs and get what you need faster

hallow sonnet
#

do you also do smelting on sushi or what?

plain hare
#

I did do one, it’s not really worth it in my opinion, but what you can do ( in the spirit of lowering buffer) is add a circuit wire to measure output belt and control input inserters to prevent furnaces from filling to 100 items

hallow sonnet
#

ah i see

plain hare
#

That’s huge on steel furnaces in particular

#

I think the added logistics cost of a sushi smelting array can only possibly pay off once beacons are involved

hallow sonnet
#

not really

steel i would make near the production where its needed

plain hare
#

(Even steel furnaces near a production line can benefit from this)

#

1 steel furnace buffering 100 steel is 500 ore that could be used elsewhere. Which is kinda big when you’re bottlenecked on ore to maybe 3000 ore / minute or so (early game starter patch)

#

Like we all cut out belt stubs and like make perfect ratio assemblies which honestly saves less plates

#

Or we use yellow inserters instead of fast inserters

hallow sonnet
#

but then again i probably gonna make 4 belts of all materials as input and then processing it into 4-5 blocks of upgrades on a sushi

#

maybe my spm target is too high

plain hare
#

I think 45-90 is pretty good tbh

#

Though consider going lower on purple / yellow science

#

Key phrase: 2 stages of Factorio - before and after bots

#

Bots are blue science so I’d say plan around that, in particular, you can scale up in a larger way after you obtain construction bots

#

I like automating maybe 2 assemblers on purple / yellow science in the bot rush base - you can rush out cheaper techs that help a lot and keep most of the resource flow for re investment in more production lines

hallow sonnet
#

this seems pretty much doable to fit in a single block

#

for 60 spm without space science

#

that would be just the smelting

well i tought thats gonna be smaller

#

desinging this gonna be a pain and will take more than this evening oof

hoary trail
#

Otherwise, I did red circuits, all science, and even a few military stuff with sushi

#

I think I didn't do sushi for laser turrets also, because it was so simple that sushi seemed to give no real benefit

hallow sonnet
#

good to know

foggy junco
#

the stack size for nuclear fuel is 1 to begin with, so i don't think that's a good item to demonstrate with

foggy junco
#

Huh

plain hare
#

@limpid hornet welcome to the rabbit hole xD

limpid hornet
#

lol thanks

desert tide
#

What did you expect to happen? That inserters, which have one stack for their hand inventory, to be able to carry more than one stack?

foggy junco
#

I mean, isnt that obvious

pulsar lance
#

Someone remembers my "half"-suhsi rocket fuel factory? It backs up sometimes, so it kinda does not work perfectly

foggy junco
#

Can you show a design?

#

Do you output the rocket fuel onto the input belt?

plain hare
pulsar lance
foggy junco
#

Doesnt work in threads

pulsar lance
foggy junco
#

You can hook up some circuits to your belt and only allow input if there's less than a certain limit on the belt maybe?

#

Input from solid fuel

#

I mean

#

Or you exactly calculate the input, as in you only allow 10 solid fuel in once a rocket fuel is outputted

pulsar lance
#

Maybe i try these, but i am not sure, whether any of these will help... I want to input almost a full belt and those options sound, as if it would happen quit often, that the belt is not full enough

pulsar lance
plain hare
#

I think if you want the non sushi belt loop to work you should measure a buffer of rocket fuel at the other end (train stop?) and then either run it full tilt or not at all -> SR latch on when rocket fuel <10% off when > 90%

pulsar lance
foggy junco
#

Still not too hard to calculate it with prod

#

What prod are you using?

pulsar lance
#

40% productivity, means i need 7,1428571428571428571428571428571 solid fuel, per rocket fuel

foggy junco
#

For 7 rocket fuel, you'd need 50 solid fuel

#

Should be doable

#

You multiply both input signals by 7 and it shouldnt be too hard to calculate

pulsar lance
#

SR-Latch worked, not sure how much impact the buffer alone had.

#

Now I am heading towars two inputs, 1 output on the same belt 🙂 ....Edit: Finished

queen lodge
#

Hey what's going on

foggy junco
#

Sushi

hoary trail
#

I feel like this thread is rapidly becoming more of an advanced circuits channel rather than sushi

foggy junco
#

I mean its still sushi

#

I dont see a problem

hoary trail
#

That's true

#

But for someone who is new to sushi, this channel might be pretty intimidating

foggy junco
clear latch
#

Maybe some pins linking to the faq and supporting discussions would help. There seems to be some incidental discussion of how to make the most basic stuff work, but it's scattered and they'd have to scroll pretty far up.

plain hare
#

I’ve been meaning to make a Better guide with simple setups…

hoary trail
#

There's a misconception it seems that sushi is inherently too complicated to be practical

#

It would be unfortunate if this channel fed into that

jagged light
#

lol

normal gulch
#

It’s super complicated the way I’m planning on letting every item go on lol

distant gorge
#

An academic exercise more than being inherently useful

timid cosmos
#

I never got far eneough in BA to see that stuff but it looks painful

distant gorge
#

That just Angel’s

hoary trail
#

@plain hare so I'm working on building sushi rails now. It seems to me that the splitter priorities only work if there's no loops in the rails, otherwise the sushi belt gets split into two distinct loops that get cut off from each other. I could solve this by making a separate pieces specifically meant for joining two pieces of track without allowing the sushi belt to join. Do you not have this problem?

timid cosmos
#

the loop will be turned off when the chest is satisfied, otherwise all the materials get priority send to the chest

#

not sure if that helps with your question

hoary trail
#

I'm not sure, either, haha. Although it intrigues me that you only have one inserter and one chest supplying everything

#

I wasn't aware that I can do it with only one chest

#

I've been using nine chests, one for each material lol

timid cosmos
#

Using circuits you can set the filters on a filter inserter. So a constant_combinator sends a positive signal "+5 transport_belt". This sets the filter to belt since the inserter is getting a positive belt item signal. You then read the contents of the chest, and *-1, to make it negative. Now when there are 5 belts in the chest, that +5 from the constant combinator is fighting with the -5 belts it's reading from the chest, it cancels out and becomes 0, and the inserter un-sets the filter.

hoary trail
#

Wow, that's very nifty. I usually send negative signals, but I need to switch to be able to do that trick

timid cosmos
#

Sending negative signals can be useful too for other things

#

i use it a lot in my mall

clear latch
#

Yeah I use negative signals in my local CCs with a decider instead of arithmetic. Either way can cancel the items with what's available

timid cosmos
#

and then the way I have it set up, I can read if ANYTHING is >0, turn the belt on going to the inserter, the splitter givers priority and directs 100% of the sushi flow to the inserter

hoary trail
#

Do you find that construction ends up being bottlenecked by only having one inserter for all the items?

timid cosmos
#

so far construction is bottlenecked by my feeding system I'm considering using one of putin or fireballs giant belt limiting arrays

hoary trail
#

I built a small belt-limiting design a long while back, I'm using that to feed mine

#

This is how mine currently looks now. It's a bit different because I have a unique limitation of having to use a 38x38 grid, and I'm only allowed to put junctions every other section

unknown.png
#

I'll fix the giant chest array later

timid cosmos
#

I will probably end up coming up with some circuit design, i need to get motivated to work on it again

clear latch
#

It can be pretty slow yeah, I use two stack filter inserters in my rails and it takes a while. Definitely not gonna compete with a 48 inserter builder train that's already there but that's the price you pay for convenience + 100% uptime + smaller unloading footprint I guess

timid cosmos
#

builder train has trouble building the actual rails, which is the whole point of this exercise

hoary trail
#

Because I can only put intersections at every other section, I can bundle the straight piece of track with the intersection itself to make a 76x38 blueprint

timid cosmos
#

my design will only send items on the sushi for building the rails themselves, i'm considering not even doing defense

hoary trail
#

It feels a bit unorthodox, but it works so far and it should be able to integrate seamlessly with my current rails system

clear latch
#

Ahh I'm contemplating actually building everything minus defense with mine

timid cosmos
#

I thought about using it for outposting

#

but thats so much crap to put on the belt

#

refineries, chem plants, pipes...list goes on

hoary trail
#

I feel like it's more efficient to use a builder train or a spidertron (or several) to build the outposts

#

I like the spidertrons because it means I can walk over to the ore and start building the outpost while the bots work on the rails for me at the same time

#

A builder train would need to wait for the sushi rails to be finished

clear latch
#

My thought was that in terms of belt throughput the raw number of materials inside a city block isn't that much more than the number of rails, belts, and combinators that go in a block itself. In fact it might be much less, especially with modded buildings that I'll be using to save UPS

timid cosmos
#

i have a nice construction train blueprint , the sushi will build the rails for this

contrain2.PNG
#

i'm proud of my construction train works like a dream

hoary trail
#

But I haven't figured out how to do it

timid cosmos
#

the only way I think you could do it reliably is with >>ghost scanner<<

#

well

#

mod that reads ghost blueprints in the logi network

#

i've thought about ways you could do it beefore, it would be complicated but it would be cool

hoary trail
#

Yeah, I don't want to do it with mods

clear latch
#

Ghost scanners can really hurt UPS though, I prefer ghost counter to make CCs in advance

timid cosmos
#

yeah I tried the ghost scanner and it cause about a 5-tick lag every 360 or so ticks, it felt like.

hoary trail
#

I mean you could do it with a chest that requests items only when it begins to be depleted, so for every item that a bot takes out another item goes onto the belt. Sort of like a memory cell. But it would be far too slow because the item would have to travel along the belt all the way to get to the outpost

timid cosmos
#

you just typed out my idea, except i would send in bulk. find out how much the blueprint needs and send it all at once in exact amounts

#

too much trouble though, imo

hoary trail
#

And really, the only reason that the sushi belt is even remotely efficient at getting things done is because there's a constant flow of every item on it. So having a conditional outpost supplier would be too inefficient to be worth it

#

When you have items on the belt all the time, there is minimal lag in getting things built because there's always a supply of what you need

#

Oh wait I need radars haha, lemme add that

timid cosmos
#

yeah true, i learned these lessons in my Sushi Village

#

that was fun

clear latch
#

CC offset already has the effect of replacing items taken out. I tried setting flow rate based on an average of CC requests in every rail but it kept shutting down to re-average every time a new segment was built, basically the kind of lag you wanted to avoid

hoary trail
#

This is my current sushi kitchen. It could scale up to red belts, and it can even do blue belts, but it can't do really high flowrates. That isn't a problem with this application as no individual item should ever be more than a third of a blue belt. I think I started running into throughput issues after that point iirc

unknown.png
timid cosmos
#

here's what i've got so far

sushirails3.PNG
#

here's the kitchen still not functioning as intended

sushikitchen.PNG
normal gulch
#

i have bigger limiters that can work for this

hoary trail
#

I like my tiny limiters

normal gulch
#

what

#

hows that possible i need that blueprin

timid cosmos
#

my rail and item catch design works nicely i just need to work on the official input engithink

#

I want it to be an endcap print that I can slap down and use to start the rail system from

#

i just need to play with the circuitry some more

hoary trail
normal gulch
#

can u share it?

#

unless u playing rn

hoary trail
#

I can paste the string here but the bot won't work here

normal gulch
#

yeah i just want to play with it

hoary trail
#

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