Quality and Speed: How much waste does haste make?

[ichaleynbin]

So there is net positive combination of quality quality and speed module, where lower chances are more than countered with increase of speed.

This is the thing. Legendary Speed 1 modules can be quite effective because they don't provide as much of a bonus or penalty. For fewer module machines, like crushers or chemplants, you may want to even add an empty/eff beacon just to dial it back even further. For cryos, because it's 8 quality modules, you get some grace on both numbers. The speed bonus will mean more, going from -40% speed to +272% speed from the speed 3 module is a 620% speed gain, because of how bad the speed penalty was to begin with. The quality penalty will also mean less when compared to a machine with fewer quality modules, taking 6% away from ~50% is about a 12% penalty, whereas taking 6% away from 25% is a 24% total penalty, dropping it to 19.

10% of 1,000 items/sec is significantly larger than 25% of 100 items/sec. "Oh but 900 waste" Yes but you see, those are second order chances. You recycle them and get another chance to roll higher quality materials. Particularly with something like electric engines, generating more low quality engines allows you to roll more dice at the second order, and then the ones that fail again get to be third order.

The numbers are usually not so extreme in game, that's just to show clear examples. The plastic one is just the most cut and dry because it's purple plastic; If the quality goes up, it's going up to legendary. .371* 178 is 66 legendary plastic a second. .434*96.7 is just shy of 42. .496*15.6 is 7.74.

Second order chances on the remaining 112 purple plastic is *0.25 *0.248(no need for speed on this step but you could) is 6.94, third order is 1.31, so that's an additional 8.25 or so, just over 72 legendary plastic per second. Just the second and third order chances on the speed beacon version generates more plastic than the one without any speed.

Obviously if you can't keep a machine supplied, don't speed it. Particularly higher quality products, those can have pretty serious impacts. Stuff with prod techs are also obvious candidates. This is one of the better ways I've found to generate legendary iron on nauvis so far

bcs.png (1.07 MiB) Viewed 798 times

It takes like 3 miners of each so like... yeah that's relatively compact for what it yields, compared to most of the other stuff I checked that's going straight to legendary. The beacon on that build straight up halves the size. And it's 100% efficient for materials b/c it's recycling blue chips so like... it doesn't even matter if the chances are lower. Just roll the dice again.

[ichaleynbin]

Which of these three setups produces the most legendary plastic per second?

None of the above.

First off, you didn't answer the question. Second off, OP is an order of magnitude wrong lol.

[Junorus]

I could not watch this topic... so I did some calculations.
For single machine it is quite easy to calculate how much transmission of speed module will be making more quality items per time unit. It does not matter if you are using t1 or t3 speed modules, the legendries have the same speed/quality penalty ratio. Just that t3= 2.5 t1.
Different machines can have different number of module slots 2,3,4,5,8. I cannot find anything with 6 or 7, but I will put numbers for 6 just in case I missed it. You can get nearly any transmission by combining different quality beacons. I will put numbers for Q5 t3 speed modules transmission.
To make thing easier to understand let's assume that the machine is doing 1item/s without any modules.
Slots(s) number of quality items/s with Q5 t3 quality modules(Qs) max number of quality items/s MaxQs transmission needed for max t relative increase is speed

12-10-2024, 18-29-53.png (8.89 KiB) Viewed 741 times

(9.68 transmission is exactly 15 legendary beacons, 1 module each)
So if you do not care about waste if materials, you can go quite far with beacons, but there is hard limit how fast you can go. But if you want to optimise for module number, setup details will start to matter.
So how to image above max legendary plastic per second will need more than 3 beacons, but 3 will be reasonably close to maximum. Exact will be for 15 single module beacons...

[RockPaperKatana]

[Moderated by Koub : Off Topic]

[ichaleynbin]

(9.68 transmission is exactly 15 legendary beacons, 1 module each)
So if you do not care about waste if materials, you can go quite far with beacons, but there is hard limit how fast you can go. But if you want to optimise for module number, setup details will start to matter.
So how to image above max legendary plastic per second will need more than 3 beacons, but 3 will be reasonably close to maximum. Exact will be for 15 single module beacons...

Interesting numbers, that behaves about how I thought it would. There's a pretty big spectrum of combinations you can have to dial in T precisely, but having a T to shoot for is a great step in the right direction. If I'm interpreting this correctly, T is just the craft speed multiplier you want to shoot for, to maximize quality crafts/sec?

There's a lot of ways you can reach those T values, One beacon has a speed 1 and a speed 2, another has two speed 1's, different beacon qualities as you mentioned, etc. Modules should stay legendary, they get a bonus but no penalty for their quality. 15 beacons is probably not the most optimal way to reach that T value, but maybe it is, idk. There's going to be a certain amount of 'close enough' here I think?

This doesn't take into account second order chances but I think that's fine, I'm fairly sure as a simplification method, it's fine to just take Quality% * Items/sec and call it a day. You do get additional second order chances for more speed, and this may affect the optimal points.

[Koub]

Moderator-me here, please keep the discussion on the thread's subject. I moderated the last post for that, anything unrelated to the thread's subject will be moderated past this point.

[coffee-factorio]

I could not watch this topic... so I did some calculations.
For single machine it is quite easy to calculate how much transmission of speed module will be making more quality items per time unit. It does not matter if you are using t1 or t3 speed modules, the legendries have the same speed/quality penalty ratio. Just that t3= 2.5 t1.
Different machines can have different number of module slots 2,3,4,5,8. I cannot find anything with 6 or 7, but I will put numbers for 6 just in case I missed it. You can get nearly any transmission by combining different quality beacons. I will put numbers for Q5 t3 speed modules transmission.
To make thing easier to understand let's assume that the machine is doing 1item/s without any modules.
Slots(s) number of quality items/s with Q5 t3 quality modules(Qs) max number of quality items/s MaxQs transmission needed for max t relative increase is speed

12-10-2024, 18-29-53.png

(9.68 transmission is exactly 15 legendary beacons, 1 module each)
So if you do not care about waste if materials, you can go quite far with beacons, but there is hard limit how fast you can go. But if you want to optimise for module number, setup details will start to matter.
So how to image above max legendary plastic per second will need more than 3 beacons, but 3 will be reasonably close to maximum. Exact will be for 15 single module beacons...

THANK YOU! This is valuable!

So here's the deal. You're right. But you have to realize you're right in general.

So if I move your logic earlier into the production chain, it benefits me.

It's absolutely correct to say that for the machine energy given and the units involved, a brute force solution might produce a better rate. If you accept that your vertical solution is the best, you're going to have to scale off 5 beacons and collective modules cost to get the "best".

However, at the speeds we are going (red/blue belts for items/minute) and the investment costs, it's worth looking at alternative solutions to maximize that brute force. For coal ->plastic as a route you get into an interesting situation. Because the productivity bonus on the building can bee made so high that you don't need to invest beyond 8 productivity modules to have an assembly line that is valid and more performant regardless of your other choices. And when you add beacons to that line, it can scale vertically more because the only thing it is sensitive to is the rate of legendary coal being delivered to it.

Even toy layouts have to be constructed carefully because if you void wrong, you'll significantly impact performance numbers. It's a good exercise.

The logic here though is that drills scale with mining research, so in the land of imagination a 300 ips science line is an achievable thing that can burn through productivity researches to make it so a drill can hit a belt of interest. Practically, the speed you need to benefit from that logic w/out applying speed modules at the point of drills is greater than productivity 20. And if I just don't care and want to shove in greens, I can still do it from a non-legendary building that's just going to get better once I'm in a position to care.

For other routes (say Foundary vs. Electric Furnace on legendary iron plates) you also have to factor in recycling to see a bonus. Coal->Plastic is uniquely at the disadvantage to placing speed modules on the production facility because it can be made so productive. I have to test out the recycling element on a Foundary vs. Furnace, because from the perspective of going speed vs. productivity an electric furnace is so much less productive that on the basis of resource usage it is beat before I try recycling.

Until I test out what recyclers can do to that line it's open to me. A recycler system in Foundary vs. Furnace is going to have 66K chances to produce a better ore than can be multiplied by 1.5. A Foundary is starting out at 63K choices but it's material won't revert to ore, so at that point I might be able to say I get more but I have to do a 5 recycler layout on each.

[ichaleynbin]

Coal->Plastic is uniquely at the disadvantage to placing speed modules on the production facility because it can be made so productive.

Yeah it's actually quite interesting, and there's a few dimensions to play around with. I'm still trying to figure out how to optimize properly, if I'm being honest, that speed beacons have a potential benefit in quality crafting is just a smaller truth that's part of a much bigger picture. I've constructed a thing in Foreman2 which should allow me to find some solutions, I just have to give it the options it has to choose from. This is legitimately what factoriolab is for, but I can't get Flab to work properly for quality yet.

COalwashing.png (1.15 MiB) Viewed 622 times

For some reason, given what I specified as my desired output products, it wants to wash <=rare coal, go purple plastic in the 8*qual3 cryoplant, and then wash the purple plastic. I'm certain this has to do with the bonus productivity on plastic, but I'm still trying to figure out what the proper approaches are. The thing with plastic is that if you turn it into plastic from common coal, yes you do get a ~50% roll, but you also get 8 plastic. You get the same odds of jumping when you convert from coal to plastic, so having fewer machines is an investment advantage, but also it's better to wash the coal because it's an 8x multiplier, every coal you wash is the same as washing 8 plastic. The order really matters here, some places the item/sec increase might actually dominate for UPS concerns, even if material concerns don't exist b/c of 300% prod.

[TechyBen]

So there is net positive combination of quality quality and speed module, where lower chances are more than countered with increase of speed.

This is the thing. Legendary Speed 1 modules can be quite effective because they don't provide as much of a bonus or penalty. For fewer module machines, like crushers or chemplants, you may want to even add an empty/eff beacon just to dial it back even further. For cryos, because it's 8 quality modules, you get some grace on both numbers. The speed bonus will mean more, going from -40% speed to +272% speed from the speed 3 module is a 620% speed gain, because of how bad the speed penalty was to begin with. The quality penalty will also mean less when compared to a machine with fewer quality modules, taking 6% away from ~50% is about a 12% penalty, whereas taking 6% away from 25% is a 24% total penalty, dropping it to 19.

10% of 1,000 items/sec is significantly larger than 25% of 100 items/sec. "Oh but 900 waste" Yes but you see, those are second order chances. You recycle them and get another chance to roll higher quality materials. Particularly with something like electric engines, generating more low quality engines allows you to roll more dice at the second order, and then the ones that fail again get to be third order.

The numbers are usually not so extreme in game, that's just to show clear examples. The plastic one is just the most cut and dry because it's purple plastic; If the quality goes up, it's going up to legendary. .371* 178 is 66 legendary plastic a second. .434*96.7 is just shy of 42. .496*15.6 is 7.74.

Second order chances on the remaining 112 purple plastic is *0.25 *0.248(no need for speed on this step but you could) is 6.94, third order is 1.31, so that's an additional 8.25 or so, just over 72 legendary plastic per second. Just the second and third order chances on the speed beacon version generates more plastic than the one without any speed.

Obviously if you can't keep a machine supplied, don't speed it. Particularly higher quality products, those can have pretty serious impacts. Stuff with prod techs are also obvious candidates. This is one of the better ways I've found to generate legendary iron on nauvis so far

bcs.png
It takes like 3 miners of each so like... yeah that's relatively compact for what it yields, compared to most of the other stuff I checked that's going straight to legendary. The beacon on that build straight up halves the size. And it's 100% efficient for materials b/c it's recycling blue chips so like... it doesn't even matter if the chances are lower. Just roll the dice again.

When the Factorio math helps with your Balatro meta... second order/third order re rolls... amazing!

[scottmsul]

For what it's worth I've been working on a linear solver script that can optimize number of modules used, buildings used, inputs used, etc. It can also check speed beacons. Linking it here in case anyone finds it valuable: https://github.com/scottmsul/FactorioQualityOptimizer

It can be difficult to predict ahead of time exactly how many speed beacons to use on each step. If you just guess a number it'll probably be wrong and "look like" speed is bad. The solver generally likes to sprinkle in some speed beacons here and there but not much, and usually only on lower quality steps.

[coffee-factorio]

For some reason, given what I specified as my desired output products, it wants to wash <=rare coal, go purple plastic in the 8*qual3 cryoplant, and then wash the purple plastic. I'm certain this has to do with the bonus productivity on plastic, but I'm still trying to figure out what the proper approaches are. The thing with plastic is that if you turn it into plastic from common coal, yes you do get a ~50% roll, but you also get 8 plastic. You get the same odds of jumping when you convert from coal to plastic, so having fewer machines is an investment advantage, but also it's better to wash the coal because it's an 8x multiplier, every coal you wash is the same as washing 8 plastic. The order really matters here, some places the item/sec increase might actually dominate for UPS concerns, even if material concerns don't exist b/c of 300% prod.

Say x is the chance of getting a green. From x/10 you get rare plastic * 1.5 out of the cryo-8q on rare. That means you're rerolling based off that 1.5x/40 pool of candidates down that cycle instead of 1x/40 coal. But when you cycle, you have to remember that the upcycle is always relative. So you're 1.5x share of purple coal is going to become 1.5x/10 plastic share of legendary.

With plastic it is not so good because it's a chemical product. So your cycled product is always going to be the chemical product.

What if it's a component that benefits from inherent productivity? With green chips every time you wash you multiply by 1.5, giving you mechanical advantage. With something that also gets a research bonus like LDS the potential is insane (and I've already heard that with a high enough inherent bonus this turns into a breeder cycle for rocket components at legendary quality).

Want to see something cursed? Maybe not in this thread, it's tangentially related but I can't say it's practical.

[coffee-factorio]

Alright, pics to come for this and I'll edit them in when I'm less exhausted from work.

So I ran benchmarks of the things being discussed. I did plastic and iron. The cursed, tangentially related thing is what happens when I try and do this stuff using biochambers, because going to plastic through artificial coal is an exercise in magic.

I'm half convinced the sulfur part of that particular exercise will all spoil and you half to ship in petrol for it and... yes. It's mad.



So for plastic. We start with a Q8+Legendary Speed Beacon, 2 marks + Legendary Machine. We have legendary recyclers but for practical purposes I'll show a test build that doesn't use them. The idea of this toy is to show what you can start investing with. The square patches of coal where based on finding 4 million patches that where bigger on Nauvis, default settings. They're 32x32.

1 Hour mark that makes 6.6-7.5 ipm. I need somewhere to start and this shows what haste makes. Granted I could copy and paste machines with q8 till I overmatched, but scaling horizontally like that is an always option and it's more valuable IMO to show where you can start and think about how you can vertically scale designs with toys. All coal consumed is 1.3k/min based off histogram - close to the machine speed * 60. This includes recycling. So a red belt for 6.6ipm. Recycling is the hardest part. But when I tried without, the input was about 1/3 of what it is so do it.

Of course you need drills but that's arguably easier done in this pattern: you can straight speed boost a drill and I'm giving them productivity 20 because you should have >10 ips science of all flavors when playing with things that cost ten grand unit. So basically, a normal beacon and normal drills set to scream and you're good.

I'm using legendary recyclers, arguably you could just use more non-legendary ones.



Now if I switch over to productivity my input seems like 24ips coal. So 1440 approximately. My output graph claims 58K which I don't believe it because I can't see the full belt of green that various things say I should have. Switched over to a red belt to show it isn't getting filled so \o/. I'm going with what the speed on the drills read. Drills are legendary but legendary is equivalent to not from a speed perspective: resource drain is lowered so the patch lasts insanely long. Everything else is normal quality here, except the modules.

It does 38ipm.

I cut all the buildings back to normal quality and it still does 38 ipm for the nearly the same amount of resources. That's the why to what I'm saying, it just shows an increase from 6-7 to 38.

This is bench marked in game over ten hours, so it's what you get if you get the machines working. I benchmarked similar last night but tossed everything out, I got ~3 ipm out of it. So you need to do both to get this result.



Right, today's project given that example and the idea from the linear solver. So the start then I have filtering out and benefiting the t4 input material. This pushes it up to 60ipm/1 ips. In short, that 1.5x productivity doubles output. It's more or less the same as the above and it should be able to scale down. So that's just what productivity was doing to the thing. This is off the same 1440 and I'm not injecting speed into this process.

What if I did? Or "Why was the constant factor of loss higher than 2 before benefaction, looking at only productivity?"

Well, depending on where you speed it up your quality is dropping from .0006 to 0.0004. But that's happening on a geometric scale as parts are brought your building - it affects the initial roll and all rerolls, which occur at a 1/4 rate until rounding ensures they disappear. I'm not good enough to say what the sum of the geometric series is but I know that's the stick I'm getting hit with. I'm not saying you can't use brute force to get 'round it, but from a point of view of "where do I start/how much do I lose" then brute force is kind of the final option when you say 3.6K legendary plastic in an hour isn't good enough. But the bonus round will show that the situation is lot more nuanced and arguably plastic is a special choice.

And again, all the stuff in this build was converted to non-legendary buildings to show it's the modules themselves that you want to start investing in from a practical point of view. I won't say you can't just buy a million tickets, but if you do then as a strategy it seems like you want to avoid using quality modules in the production step. The benefaction step on epic to legendary is slick man, I dig it; and that's where you want those quality modules.

And todays bonus round is plate from a foundry vs plate from an electric furnace.



Here things are more difficult. A foundry just slaps and it has an inherent productivity bonus instead of one added in.
My calculations show use at 10 ips ore for ~1.9-2 ipm



But if I do benefaction and productivity on an electric furnace I get ~11.8 ipm for 24 ips input (same drills, can't quit 'em). Assuming that the epic trick works the same here, half of that is from that step. And then a little over half of that is from material advantage. So you need both those steps to compete with iron from a foundry, but once you do you go from about 1/5 output per ips to like 1/2 output per ips input. The foundries productivity lets it keep up.

I might like to try out the OP's suggestion of electric engine units next. But it's rough because I need an example of pure speed and an example of doing it right with something I'm just learning too.

[ichaleynbin]

Coffee, it feels like we're talking past each other. You keep talking in items per item, legendaries per common. I don't find this number particularly important, and when it's important it's for a different reason that a different optimization step will catch. I keep talking in items/sec output. This isn't a situation where you can say "Look at this, I found a design where speed didn't improve it, that means it doesn't work!" no amount of counterexamples can prove it never works, I only need the one example where it does work and plastic is something. And yes, productivity is a better item/item comparison. I just do not care about item/item. I'm not entirely sure what point you were trying to make with your last post but it seems like you're still taking it as me saying "All speed all the time" when from the beginning I've been talking about it from a chain maximization standpoint lol.

Of course speed generates more 'waste' as well. I never said otherwise. But where you take your prod step matters, and plastic in particular has free prod, which changes things as well. I think that's why it wants to put the qual step at purple, it's specific to coal->plastic with 300% prod. Also, breaking these into multi-step is often critical, generating more green to generate more blue, taking jumps one quality at a time by design. The multijumps are always 10% and so can be largely ignored if you come at it from i/s, better i/s is better multijump i/s.

"Not recycling" is also an option, there are a few free steps in some of the sciences where you can't do prod. Notably, the ingredients for green science and purple science all may not accept prod. "Opportunistic quality" might be reasonable for the sciences, I put together a ratio'd build (that includes speed beacons on appropriate steps) last night. Forcing quality minimums with recyclers doesn't seem as good, but the qual here is just extra science for "free."

Opportunistic RGMBPY.png (1.46 MiB) Viewed 422 times

The reason I don't care about items/item is because by doing recycling, we're signing up to yeet 95%+ of something in the recyclers for even a single step. Once you're introducing the recycler step, the overwhelming majority of the items are getting thrown away. If you care about i/i, don't yeet them, try something like that opportunistic version which multiplies rather than divides. If your target is ONLY legendaries, and non-legendary are an unwanted byproduct, recycling a lossy cycle deprioritizes i/i for me. When doing something like blue chip or LDS upcycling with max productivity, however, you're getting 1:1, so if you really care about item conservation, you should use blue chip upcycling to produce your electric engines. It's 1 common input == 1 legendary output with techs. Use LDS to print the legendary steel for you.

BC to EE.png (723.63 KiB) Viewed 422 times

Using blue chip upcycling allowed me to produce 300 electric engines/minute, 750% of OPs version. The version with a few properly sprinkled speed beacons:

BC to EE Speed.png (747.26 KiB) Viewed 422 times

Beacons are eff3, speed1, both legendary, two beacons a piece. It does nothing but reduce machine count. "If the only tool you have is a hammer, every problem starts to look like a nail." Open your mind, there are so many fun solutions hiding if we only know how to start looking for them. 10 EM plants, 5 foundries, less than 10 recyclers, four asm3's, 750% improvement on EE output.

[ichaleynbin]

Just for good measure my materials efficiency is also an order of magnitude better, ~4800 molten iron compared to OPs >40,000.

Materials.png (221.25 KiB) Viewed 418 times

[coffee-factorio]

Ah, I had some breakthroughs. Although one is a correction in errors made on Wednesday but they are educational.

So the 5.8 machines speed not filling the red belt is because the machine automatically stacks. And arms of all kinds will automatically destack. This is different from Nullius/Deadlock stacking where it is packaged and super obvious. So the actual rate of the lines pictured is 69.6 ips then. What I get for going to the map editor but putting down a drill and blue arm cleared it up right quick.

This means I have to redo the lines for speed on plastic, particularly to include 1 speed legendary build but with the full mines.

The reason for doing this is that I can go [items/second] / [items/second] and get an idea of how powerful machines are, relative to one another. Of course a good optimization will just run the competition over; particularly since these rates are so far under item/second. At that point if I decide to scale I can.

I really really need to redo the lines for furnaces. I've got a feeling that the gap there is alot worse. I've got a lot I could say about a Ichylabin but I'll keep it two things. The first is that the productivity modules in those machines from the post proceeding are going to do a lot to a line provided you can scale it.

The second is that I've got the beginnings of modeling tools, rather than just Monte-Carlo simulation using the engine and maybe sprinkling in the kindness of others when it seems appropriate.

So your initial pass for quality between a layer is going to yield ax/(10^t-1) parts, where a is either your rate in items per second or some large pool of items (4 million coal, 3 million iron, 4800 inserters). So epic to leg will do this and normal-uncommon. At this stage, a*x/1 and x = 0.1 for say 4 q1 chips. But if you improve to q3t5 that's 0.24. When you divide out your items you'll see that for a given pool you get 0.24/0.1 = 2.4 times the amount of good product for a fixed rate. Productivity is more of a concern here because it will flat multiply that rate. And that's going to be consistent down the chain, your total legendaries if you just generate then produce and toss everything out will at least be multiplied by your productivity. Unlike normal production you're committing all your items to get 1/10 to 1/1000 items in that case, so needing 1.5x less energy over thousands is really huge. Since machine utilization is low, you can dump greens on the last step and cut the size of the power back-end.

Recycling is going to add the sum of a geometric sequence to them. Each recycling step makes an amount of parts a/4. When you add in quality to the cycler, it'll color (ax/40^q-1) parts based on tiers. Benefaction schemes will pass the good stuff out. For chemical->recycler as in plastic the cycle is 'cycler->cycler'. But for components it's cycler->machine->cycler. If the machine trivially has quality modules that don't do much. But productivity of any kind will multiply the population going into the next cycle and drive the rate up products produced up. Numerically, you're getting the sum of the series out which is given as: quantity/(1-passing) out of a benefaction cycle.

See https://en.wikipedia.org/wiki/Geometric ... _its_proof for where I'm getting quantity/1-passing, it's given as a/(1-r) in that source. Alternatively AMS Math has free calculus textbooks.

Passing is going to be at least quality. But if you can sneak productivity in there it drives 1-passing lower. (1-passing) < 1, say if that where 0.5 you'd see that as quantity/0.5 = 2quantity. By adding a 10% penalty for worse quality chips, say. That becomes 1.66*quality. Speed is going to gouge the rate like that. As you're adding steps of recycling together, a small difference over all those steps will effect your benefit by alot. In practice just adding a benefaction step on legendary plastic brought it up to 1 ips from ~2/3 ips.

Where speed matters more; and this is why I have to redo iron, is when productivity is going to just make so much material that it's impractical not to go fast. So in forging, the liquid step + productivity just adds such a huge constant multiple that 1.66 to 2 just doesn't matter if you can't milk it for all its worth. I was already sort of seeing this because I could only fit down 1 forge and had to infer what it was doing, but since I'm using 3 times as much iron as I original thought that 1/2 bonus is actually 1/6.

You can't benefit in iron off most conventional items at first glance until you realize iron plates + 2x productivity gears = quality cycling in a foundry for belts.
Instead of cycling on purple though you probably want to put that cycle in on green or blue because then you'd actually have enough parts to make belts

Here's where item to item clarifies things when you have accurate rates: ~1.66 < 2. But if the iron drills can't be more productive. The forges liquid step can carry a 1.5 to a 2.5 value for the same investment of material. So even if you have to put in a speed module in you end up seeing that as (~1.66*1.5) to (~1.66*2.5): 2. Since forges are 5x5 you're in a situation where you can more reasonably look at things and say "Well that won't fit on a Space Platform". We'll see.

And with the rocket unit cycles, you get inherent bonuses that can outweigh having a 3-4 deep supply chain with research. Your 20 greens and 2 reds each can benefit from a 1.5 minimum bonus off the EM plant. And then blue benefits off an EM plant so that's 1.5^2 on everything. And then you multiply that by what research gives you. And then you recycle, is sort of the game plan. And you target that on uncommon and normal because you need 20 green and 2 reds to play. In principal it'll look something like ichaleynbins graph. It's a fascinating thing to think about because there's several supply chains hitting that point for material, and researches play a great role in demonstrating what it can do.

I do know I'm babbling on about the why when the what has been shown between either a linear solver or a Monte Carlo. I'm aware that this doesn't apply to all sciences and well, engines is the focus.

So that isn't one but a few builds. I have the weekend now baby.

[ichaleynbin]

Where speed matters more; and this is why I have to redo iron, is when productivity is going to just make so much material that it's impractical not to go fast. So in forging, the liquid step + productivity just adds such a huge constant multiple that 1.66 to 2 just doesn't matter if you can't milk it for all its worth.

Yeah, the thing that originally ruffled my feathers was someone posting "optimal builds" on the wiki. This math is all doable, though I'd prefer to use a tool like Foreman2 to do the math for me. I'm about 99% sure I'm not asking the right questions yet. For instance, I did do a layout on naive irons, and indeed, if you have a choice between prod and quality, it would seem prod is largely more effective, unless footprint goes up in priority.

Irons.png (443.14 KiB) Viewed 262 times

But this is the wrong question. The answers are mathematically valid, but they're all incorrect solutions. This is not how to generate legendary iron. There are multiple variables to even evaluate optimality on, and those approaches don't win for any of them, I think. Blue chips are a lot better because they can get to +300% AND roll qual, if nothing else, it's just Wube showing us that linear scaling has a...**cough cough** tendency to beat exponential scaling.

[Romayne]

Before I even get through all the discussion I'd like to say thank you so much for helping to further this discussion, I didn't expect to get this much going and I'm very happy it's gone the way it has.

The one and only reason I've been absent from this is because.. well.. I didn't get any notifications I actually managed to disable them somehow even when I thought I didn't. I might add on to this in a bit, maybe not, who knows.

[coffee-factorio]

But this is the wrong question. The answers are mathematically valid, but they're all incorrect solutions. This is not how to generate legendary iron. There are multiple variables to even evaluate optimality on, and those approaches don't win for any of them, I think. Blue chips are a lot better because they can get to +300% AND roll qual, if nothing else, it's just Wube showing us that linear scaling has a...**cough cough** tendency to beat exponential scaling.

It is simply unfortunate this is where you choose to end your search for understanding.

This is my previous work.

Right, today's project given that example and the idea from the linear solver. So the start then I have filtering out and benefiting the t4 input material. This pushes it up to 60ipm/1 ips.

The blueprint for this test case is here.

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

Copy blueprint

Instead of showing other things lets use this as a high water mark. That your idea works is a tautology- if the build does not work the linear solver won't consider it. But the question is how much we waste ends up being crucial for felt speed. You might come to see that the reason why I look at your work in the light I do is because when I look at your graphs I notice that you put a full beacons square on a machine making pipes - which take a half second to make. And if I fix the rate, I can judge what would be appropriate to affect horizontal scaling from 5 ips. It's actually a single normal assembler 3 if I don't change anything else. And felt speed will be the same, although it might not look so impressive to have one normal machine.

It adds some weight to your argument about linear scaling till you work at megabase scales where there is a meta-budget of UPS use, and cutting out assemblers from a table is going to directly increase what everything else can do.

At the end of the day we all run out of seconds; less morbidly our coal patch runs out. And then the comparison with which we are judged is items : items. If fairly scaled though items/second : items/second can be judged since that is what the layman might be interested in achieving. This is why fixing input size and understanding exactly what was going on with those drills turned out to be worthy of being called a breakthrough. As I am to demonstrate what haste wastes; but doing so unfairly is inappropriate.

If the high water mark can be beat, then I was slow; but more importantly haste makes no waste or might save us material.

Modeling suggests pessimistically that decreasing a in a/(1-r) is not such a good idea.

But the map editor is here. And I can use it run a Monte Carlo simulation: I throw an hour of seconds at items/second so the poor rates of return don't cloud my vision. And regardless of the accuracy of my model or understanding, it will tell me what the truth of the matter is if I take the time to read it correctly.



This shows what going to legendary speed with the correct scaling factor does. The belts are 3:4 balancer so all machines get even feeds - the drills do 73 ips instead of 69 but when I fix a in the equation a/(1-r) I see that as a productivity multiple. (73/69) is so small that unless I see you doing an item/second in your advantage. The actual speed is 22 items per minute or ~1/3 or an item per second. Productivity previously boosted to 37 ipm.

In the model this is shown as (2.5)a/(1-r). The building doing 60 ips does so by adding on a geometric series as your linear solver suggested.

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

Copy blueprint

The blueprint for this test case is here. A user concerned with my choice in using a balancer could instead run a single belt to each building and see no difference in rate. A blueprint that can demo this is given later.

We can perhaps optimize in terms of increased quality. OP's original question suggests that we could do what ichaleynbin suggests with linear scaling and just drop beacons and scale horizontally. By using your speed 1 legendary 1 idea to benefit quality production, I will show what that scaling is when you fix item per second input like that. Linear scaling here isn't such a great idea. This is suggested by the above post about amount of coal being produced being to your advantage. But a 1/3 speed drop is what it is. Quality difference is felt as a drop from (44% to 37%) when you go to tier 3. So that's going to be seen as (1.07)a/(1-r)



Ah we went from 3 buildings and a beacon to five and another. Output rate is about 1-2 ips more. Still at 1/3rd disadvantage when we use comparable resources. But it says something about speed that will inform my future test cases (both on a practical "how I'm going to do it" and a "they directly inspired me to make some direct cases". Those two more buildings add 16 more t3 modules at legendary rates to, basically reduce the cost of the beacons to a t2 module. That's time spent building and placing all that stuff for a micro-optimization. Starts from speed are going to be inherently unfair like that, so the lower investment cost is more of my concern when doing Monte-Carlos because I have to fiddle all this stuff into toys.

The blueprint for this test case is here.

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

Copy blueprint

The demo blueprint alluded to earlier is here. By swapping out t3 speed modules in the head of the production line you can see directly how speed effects output and without being concerned with how a balancer might change that.

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

Copy blueprint


OP you didn't perhaps have the most efficient build but these test cases show that by going rather slow and optimizing correctly we go 3 times as fast, by producing 3 times as much, when working plastic. Your logic continues to be correct and exploring it gives me no small amount of pleasure. But we're in an unfortunate position where productivity volume, independent of the speed, and geometric fractions of parts saved are going to be what ultimately drives output speed up. I believe it's been said that the effect on research is significantly different, and when I desk checked that with a calculator I found no fault with that logic. But for your concern for manufacturing it's a constant factor decrease in both parts received and felt speed on trivial cases.

The situation is rich from a mathematics perspective. It's unfortunate from a technical one. It's a scenario where a person is invited to demonstrate optimal by the technical constraints even though speed plays a role. My earlier work in iron showed that one forge is going to bottleneck another and I believe ichylanban correctly shows that the (2.5)a/(1-r) situation will dominate the production at that point. I put speed modules in the forge making iron - blasted work getting in the way of play - reduces that to (1.5)a/(1-r) which suggested better. The misunderstanding on drill clouded that. But basic iron production is going to be very rich and I hope it is not out of place to show that here.

This was the work of my evening. The work of the day is nothing short of comedy but I feel some humor is appropriate. Plastic has more stories to tell.

I feel like I have to explain my idea about blursed plastic a bit more (factoriolab has a version of it and it chooses the biolab route to sulfur). So you get 2.5 bonuses each pass through a biochamber from fruit -> prepped -> packed (bioflux) -> cooked (nutrients) -> overcooked (carbon). And with some patience you let everything spoil so you don't take a 1/4 hit. But from a numbers standpoint you need 6*5 = 30 spoilage to pull off plastic in the biochamber itself. God may leave us at this point because 2.5^4 is 39.

The cryo-plants probably deliberately have so many modules slots in the first place so they can compete with this unholy abomination. Since sulfur follows a similar route up... and then you factor in that you can make quality nutrients for your fuel line, and if you overmatch fuel needs that you get some quality spoilage as well. It suggests a micro optimization of using productivity on steps of the bioplastic route instead of just loading productivity which, from the a mathematical standpoint as well as a practical one, could be described as cthulu's tentacles. I'll leave elder gods for others to play with

[ichaleynbin]

There are at least 2 different regimes to care about legendaries for; Mall stuff, and science production. For neither do I care about i/i. First off, this is a low level of mining productivity:

Legendary.png (267.53 KiB) Viewed 203 times

I'm getting ~110 ores for every ore that leaves the ground. Lava is virtually free, the exchange rate through calcite is even better and calcite also gets the mining bonuses. If the physical materials are a concern, mall on vulcanus. Caring about legendaries per input materials is a very 1.1 way to think, we have access to more materials than we could possibly use. Doing it for materials conservation purposes, not in 2.0. Will the coal patch run out? Yes, one day far in the future, and when it does, you can get another. There are deposits for days. I reject the premise of this entire argument and have from the start.

Specifically for malls; UPS should not be a concern at this point, and IMO the highest priority is speed, if optimization is to be considered. Waiting on your mall to produce more legendary furnaces or powerpoles will slow you down, but honestly, optimizing the mall is almost always going to slow down overall progress. I tried a lot of things for mall items, and they all seemed passable, some better than others. But at this point in the early game, while legendary miners will be hard to come by, legendary quality is also not the target. The thing that's being saved here is investment cost, which is significant at this stage. Extra legendary machines have their own cost too, so if materialism is to be a concern, I don't think ignoring infrastructure cost is fair. One legendary t1 speed module touching multiple machines is a lot cheaper than a few extra legendary machines with legendary t3 modules. Amortization time is the normal way of evaluating this, how long does it take for the design to pay back. If it's a week, then it's not worthwhile for early mall, if it's an hour, definitely worthwhile.

The problem is that for any given investment cost, speed can put less additional investment in to match rates. Non-speed adds 3 machines, okay speed adds 1 and doesn't need to add an additional beacon. Coffee, I think you may be comparing apples to oranges here, for that reason. Quality items/second/total machines is a metric which finds a balance point rather than extrema; If you choose certain metrics(that I do not think are good metrics to use), speed will always lose, but if you choose other metrics like quality items/second/total machines, you can find interesting middle points. Probably it'd be fairer to compare quality/second/investment, if we're going to continue to be unnecessarily materialistic? Maybe I missed something, but I haven't been struggling and I haven't been optimizing yet either.

Miners.png (3.52 MiB) Viewed 203 times

For science production, I think the optimization to target is science per swing? A swing of uncommon gears is worth twice the effective red science as a swing of common gears, so I've been toying around with "science exchange rate" as a way to evaluate some builds for quality vs. UPS. This also allows some things like shoveling to provide unexpected advantages.

Shoveling.png (3.7 MiB) Viewed 203 times

Nearly 78% of the recycled materials will be same quality, while 22.3% go up in quality, so I'm going from 1 insertion on everything, to 2 insertions on 22.3% and 0 on the rest. The amount of materials going up-chain will be the same as such builds are built to output target for me, so adding or removing beacons actually doesn't change the amount of insertions done, it only changes the number of machines. The bonus qual science per swing doesn't change, but per machine does.

Number of input items does show up to be a concern but elsewhere, it ends up being a concern if you examine from the i/s side also, and care about UPS. It's not so much that I'm burning the items, as it is I have to move that many items/sec. I haven't found a favorable way (espm/swing) to include recycling for science production aside from the 300% prod techs, recycling balloons the i/s requirements regardless of the introduction of speed, and UPS concerns start to rear their head. I will definitely include speed on the list of things to test regarding UPS and qual science, whenever I get around to it, but so far I haven't found anything I even want to test yet.

[ichaleynbin]

FWIW I expect non-beaconing to have some decent shots at optimal points in high UPS quality science builds, if those ever exist, but UPS concerns may show recycling to not be worth it on the sciences. I'm a ways out from that still, but I have been poking at it, and legendary iron is a problem.

Recycling lossy steps (those without 300% prod techs) is going to be some form of exponential, as you've been observing, coffee. Lossy steps loses for UPS reasons, but adding speed makes it worse if you can't shovel the extras. I keep this stash around just so I can show people the gravity of my mistakes before I learned to not be precious lol, legendaries are cheap.

memes.png (2.25 MiB) Viewed 197 times