I'll use factoriolab to demonstrate. It is a linear solver and can clear the matter up, regardless of what I wrote.
I'll do it as respectfully as I can if given patience Koub. I can show.
Quality and Speed: How much waste does haste make?
-
- Inserter
- Posts: 47
- Joined: Thu Oct 17, 2024 10:56 pm
- Contact:
-
- Inserter
- Posts: 47
- Joined: Thu Oct 17, 2024 10:56 pm
- Contact:
Re: Quality and Speed: How much waste does haste make?
Factoriolab is a linear solver.
https://factoriolab.github.io/spa/list? ... *40.8&v=11
edit: If you read the picture now and zoom in, the hyperlinks match.
This is electric engine units. It might be an input conditioned lie. And looking at the recipe is going to reveal that it uses something you can't get. But also is going to be honest. Plus Foreman2 could be honest if I lie.
My relevant number is iron fluid 1306.5 off of input condition 40.8 items per minute. That's Romayne's output on her original example. It took 43K or so. Edits: this is seen at the bottom of the image in low density structures production.
https://factoriolab.github.io/spa/list? ... )***0&v=11
Adding speed 1's won't change this.
This is because a linear solver is a linear solver and will always report what it sees based off its inputs. And it's inventing legendary Yumako fruit from a legendary Yumako plant.
https://factoriolab.github.io/spa/list? ... -2**0&v=11
When you put in productivity modules on the last step.
If you add the saved material to the inputs for the same amount of material then it speeds up the production. But a linear solver set to 40.8 will always produce 40.8.
https://factoriolab.github.io/spa/list? ... -2**0&v=11
I get ~900 fluids doing this. If I drive the condition up the next move is to call me a liar. edit: Actually, given some thought. You just say that's a volume increase. My temper is a bit hot now.
But in reality you always get the modules worth of bonus input felt as speed.
Edit: I've updated the image to show that 100% yumako seed also produces 100% legendary yumako. The hyperlink changed because the API on factorio labe will report if you so much as open a tab.
If you could change the input to be not legendary yumako you'd also be able to show an exponential decay at each point you applied speed modules 1. And it'd be seen as an increase in inputs in any linear solver.
As he's said, I can't prove it. An API can't be patched fast enough.
I am not claiming that ichaleynbin knows this or is intentional but he asked if I was a machine or not.
When I call the shot I need to provide my position. And you can hit that because you're a moderator so the artillery is you. But at least leave my logic up so someone else doesn't get hit by this guy.
https://factoriolab.github.io/spa/list? ... *40.8&v=11
edit: If you read the picture now and zoom in, the hyperlinks match.
This is electric engine units. It might be an input conditioned lie. And looking at the recipe is going to reveal that it uses something you can't get. But also is going to be honest. Plus Foreman2 could be honest if I lie.
My relevant number is iron fluid 1306.5 off of input condition 40.8 items per minute. That's Romayne's output on her original example. It took 43K or so. Edits: this is seen at the bottom of the image in low density structures production.
https://factoriolab.github.io/spa/list? ... )***0&v=11
Adding speed 1's won't change this.
This is because a linear solver is a linear solver and will always report what it sees based off its inputs. And it's inventing legendary Yumako fruit from a legendary Yumako plant.
https://factoriolab.github.io/spa/list? ... -2**0&v=11
When you put in productivity modules on the last step.
If you add the saved material to the inputs for the same amount of material then it speeds up the production. But a linear solver set to 40.8 will always produce 40.8.
https://factoriolab.github.io/spa/list? ... -2**0&v=11
I get ~900 fluids doing this. If I drive the condition up the next move is to call me a liar. edit: Actually, given some thought. You just say that's a volume increase. My temper is a bit hot now.
But in reality you always get the modules worth of bonus input felt as speed.
Edit: I've updated the image to show that 100% yumako seed also produces 100% legendary yumako. The hyperlink changed because the API on factorio labe will report if you so much as open a tab.
If you could change the input to be not legendary yumako you'd also be able to show an exponential decay at each point you applied speed modules 1. And it'd be seen as an increase in inputs in any linear solver.
As he's said, I can't prove it. An API can't be patched fast enough.
I am not claiming that ichaleynbin knows this or is intentional but he asked if I was a machine or not.
But 0.00005 legendaries when applied to a quality machine is what you're going to get if you apply speed 1's as a micro optimization to production on a recipe. This can be shown on honest inputs in a linear solver, by back of napkin math by comparing 0.0001 to 0.00005, or by Monte Carlo simulation using the map editor sped up to 64x speed.ichaleynbin wrote: ↑Mon Dec 16, 2024 8:51 am I'mma be honest coffee, your posts are getting more and more deranged, and less understandable. Is this an AI sockpuppet of Romayne's, or just ESL? I don't understand half of this last post. It's all english words but they're not being combined quite right.
When I call the shot I need to provide my position. And you can hit that because you're a moderator so the artillery is you. But at least leave my logic up so someone else doesn't get hit by this guy.
Last edited by coffee-factorio on Wed Dec 18, 2024 12:52 am, edited 1 time in total.
-
- Inserter
- Posts: 47
- Joined: Thu Oct 17, 2024 10:56 pm
- Contact:
Re: Quality and Speed: How much waste does haste make?
I am going to edit this to include images so that I can write a bug report.
-
- Inserter
- Posts: 47
- Joined: Thu Oct 17, 2024 10:56 pm
- Contact:
Re: Quality and Speed: How much waste does haste make?
If you could put legendary Yumako seed into the ground and grow legendary trees you could grow infinite legendary yumako fruit and infinite legendary anything. It doesn't work that way the last time I tried. You have to get legendary yumako fruit by passing normal one's through a recycler.
I've made my edits to the post at this time.
I've made my edits to the post at this time.
-
- Inserter
- Posts: 47
- Joined: Thu Oct 17, 2024 10:56 pm
- Contact:
Re: Quality and Speed: How much waste does haste make?
I produced a layout capable of showing what I mean about quality within a rational tolerance.
One layout is 4 legendary cryolabs can be loaded with 4 quality 3 modules of normal quality. By hitting research all, I apply a 10% upgrade to all volume. And a 50% upgrade to all volume. But all buildings are equal. The game gives a speed of plastic output of 8.8 ips. 8.8*4=35.2 ips.
With green modules 2 of any quality they do 300KW each or 1.2MW in total power plus the beacon.
By short loading a crylolab with 4 modules I simulate recyclers, assemblers 3, biolabs, ect. Basically the majority of machines that can take produce recipes regardless of their complexity. It also simulates asteroids and ores. But a faster machine will run a mine out of ore quicker and you would have to pause the test perfectly to show that. The game reads 5 % quality, although common reasoning says the number should be 6%. 1% - 4 modules.
I then load a legendary cryolab with 4 quality 3 modules of normal quality. I replace the green modules and get a machine 36.5 machine speed at a cost of 5.25 MW. Before running simulation this alone shows a very dangerous pattern. Speed will and power consumption are scaled on a machine to machine basis but all of them are going to be scaled similarly to 1.2MW to 5.25. But someone running nuclear power might feel this as an unnoticeable increase in uranium use the first time they do it.
The green modules 2 may be of any quality. Not loading them will increase power by 5 times. Allowing someone not familiar with this kind of scaling see a ~6MW power to a ~5MW power. This is the limit to which that argument can be distorted.
Without showing rate of return over time we can't see what it costs in material.
I gave the machines 10 hours of runtime so that if a machine had 1 bad hour it would have 10 chances to have a good one. With a random number generator I might be off, but random number generators in computers give a distribution of numbers that is realistic if well made and I can hope to be within tolerance. I get by math 13 legendary plastic over 10 hours and 130 in total.
At the end of ten hours the complex without speed did 124 plastic. Playtime tick argument is 2160000.
At the end of ten hours the complex with speed used 36/35 more material and did 72 plastic.
72/124 is 0.58064. 60/100 = .60. Machine speeds are 16/16.5 as a ratio. And when I multiply this by 60% I get 58.18%.
0.58064*100 ~ 58% after rounding and I believe this can be reproduced using the blueprints I'm posting.
That's what this so called micro optimization is doing to legendary production of big drills (that's where I caught this) over 10 hours on the input stream of normal components, to a lesser extent because legendary quality modules are several times as powerful. Productivity is not an issue and you are welcome to increase productivity of plastic by research to 10x by auto-researching that in the map editor, to run this in an hour simulated.
But if someone has extra resources and wants more big drills they can move the electric drill machine on, and it will produce a ratio that is better but it cause the same jump in power. There's not an easy answer because you have to add a second machine, but you do not have to add a second legendary machine if it simply covers the scale of material. But then you have a bad ratio times a bad ratio and that is exponential decay.
Speed:
Pictures of the runs with that tick count on 2 surfaces. This is how I arrived at my solution.
One layout is 4 legendary cryolabs can be loaded with 4 quality 3 modules of normal quality. By hitting research all, I apply a 10% upgrade to all volume. And a 50% upgrade to all volume. But all buildings are equal. The game gives a speed of plastic output of 8.8 ips. 8.8*4=35.2 ips.
With green modules 2 of any quality they do 300KW each or 1.2MW in total power plus the beacon.
By short loading a crylolab with 4 modules I simulate recyclers, assemblers 3, biolabs, ect. Basically the majority of machines that can take produce recipes regardless of their complexity. It also simulates asteroids and ores. But a faster machine will run a mine out of ore quicker and you would have to pause the test perfectly to show that. The game reads 5 % quality, although common reasoning says the number should be 6%. 1% - 4 modules.
I then load a legendary cryolab with 4 quality 3 modules of normal quality. I replace the green modules and get a machine 36.5 machine speed at a cost of 5.25 MW. Before running simulation this alone shows a very dangerous pattern. Speed will and power consumption are scaled on a machine to machine basis but all of them are going to be scaled similarly to 1.2MW to 5.25. But someone running nuclear power might feel this as an unnoticeable increase in uranium use the first time they do it.
The green modules 2 may be of any quality. Not loading them will increase power by 5 times. Allowing someone not familiar with this kind of scaling see a ~6MW power to a ~5MW power. This is the limit to which that argument can be distorted.
Without showing rate of return over time we can't see what it costs in material.
I gave the machines 10 hours of runtime so that if a machine had 1 bad hour it would have 10 chances to have a good one. With a random number generator I might be off, but random number generators in computers give a distribution of numbers that is realistic if well made and I can hope to be within tolerance. I get by math 13 legendary plastic over 10 hours and 130 in total.
At the end of ten hours the complex without speed did 124 plastic. Playtime tick argument is 2160000.
At the end of ten hours the complex with speed used 36/35 more material and did 72 plastic.
72/124 is 0.58064. 60/100 = .60. Machine speeds are 16/16.5 as a ratio. And when I multiply this by 60% I get 58.18%.
0.58064*100 ~ 58% after rounding and I believe this can be reproduced using the blueprints I'm posting.
That's what this so called micro optimization is doing to legendary production of big drills (that's where I caught this) over 10 hours on the input stream of normal components, to a lesser extent because legendary quality modules are several times as powerful. Productivity is not an issue and you are welcome to increase productivity of plastic by research to 10x by auto-researching that in the map editor, to run this in an hour simulated.
But if someone has extra resources and wants more big drills they can move the electric drill machine on, and it will produce a ratio that is better but it cause the same jump in power. There's not an easy answer because you have to add a second machine, but you do not have to add a second legendary machine if it simply covers the scale of material. But then you have a bad ratio times a bad ratio and that is exponential decay.
Speed:
Pictures of the runs with that tick count on 2 surfaces. This is how I arrived at my solution.
-
- Inserter
- Posts: 47
- Joined: Thu Oct 17, 2024 10:56 pm
- Contact:
Re: Quality and Speed: How much waste does haste make?
It's always trivially true that you can pass 1 Yumako through a recycler to get 1/4 yumako of the same quality. Here's how that plays into the tale:
Here's why. So the old hyperlinks I put up for FactorioLab? Those told me to mouse over to the machine tab. So I did.
Factoriolab is recycling 4 legendary Yumako fruit to produce 1 legendary Yumako fruit in it's input graph.
I thought about what that really meant about building linear solvers.
If you make a table of legendary items, a linear solver that is "correctly" programmed will be give you "correct output".
But it has to make a table for quality 1 1 module. Quality 2, 1 modules. And by common figuring that's a table at least 12 deep. And then you multiply 12*5 to get a table 60 deep for legendary grades.
But if I write the modules as (a,b,c). The ways you can just rewrite (a,b,c) are factorial( length of (a,b,c) ) = 6. And that shows non-polynomial scaling on memory. Factorial is: 1*2*3*...*n in mathematics.
And we aren't rewriting 3 modules here. Factorials are studied because they let you count combinations of of objects and this suggests a binomial coefficient. Choosing 4 modules from a set of 21 modules because I need to let someone else do that math. It's been years since I've needed it this much and I'm always going to be going a bit too fast and a with my head in the clouds. I don't joke about ADHD. Regardless of if I got the constants right or not you just can't hardcode that much data.
It's not that it's impossible to give an honest answer in this case. It's just that it's computationally impractical.
So if I can't call the shot on this, I think if another more healthy person checks foreman2 the outputs there might be suggesting some wild things. And a sophomore college student in a technical discipline might be able to back me up on this. 4 slots and I choose up to 21 modules.
So no modules in a recycler is the most trivial entry in a table for a linear solver to get a legendary yumako plant from a legendary yumako plant.
Here's why. So the old hyperlinks I put up for FactorioLab? Those told me to mouse over to the machine tab. So I did.
Factoriolab is recycling 4 legendary Yumako fruit to produce 1 legendary Yumako fruit in it's input graph.
I thought about what that really meant about building linear solvers.
If you make a table of legendary items, a linear solver that is "correctly" programmed will be give you "correct output".
But it has to make a table for quality 1 1 module. Quality 2, 1 modules. And by common figuring that's a table at least 12 deep. And then you multiply 12*5 to get a table 60 deep for legendary grades.
But if I write the modules as (a,b,c). The ways you can just rewrite (a,b,c) are factorial( length of (a,b,c) ) = 6. And that shows non-polynomial scaling on memory. Factorial is: 1*2*3*...*n in mathematics.
And we aren't rewriting 3 modules here. Factorials are studied because they let you count combinations of of objects and this suggests a binomial coefficient. Choosing 4 modules from a set of 21 modules because I need to let someone else do that math. It's been years since I've needed it this much and I'm always going to be going a bit too fast and a with my head in the clouds. I don't joke about ADHD. Regardless of if I got the constants right or not you just can't hardcode that much data.
It's not that it's impossible to give an honest answer in this case. It's just that it's computationally impractical.
So if I can't call the shot on this, I think if another more healthy person checks foreman2 the outputs there might be suggesting some wild things. And a sophomore college student in a technical discipline might be able to back me up on this. 4 slots and I choose up to 21 modules.
So no modules in a recycler is the most trivial entry in a table for a linear solver to get a legendary yumako plant from a legendary yumako plant.