Optimization idea: abstraction

[blazespinnaker]

Yeah, I'm beginning to wonder if you've thought through how this would actually be utilized in a megabase where UPS is an issue.

If you have, can you share the larger architecture / design that you have in mind?

As I mentioned above, my concept is the city-block megabase design. Each block is a virtual factory, with trains stacked in common areas per 16 city blocks.

I'll leave it here but I really can't think of a design using your approach that would either be UPS efficient or frankly workable, as factories would constantly be falling out of sync. Which, as I mentioned, is an artificial mechanic which is a result of the underlying CPU issues with no analogy that exists in factorio today.

I just answered to your former post, and that simulating a sub-factory like so is not guaranteed to be a correct simulation.

You are refuting by tautology here. You're saying x is y because x is y. I hope you understand the deficiencies of such logic.

I am saying that your approach is an inaccurate simulation because it does not account for intermediates. Yes, you can artificially 'wipe' factories constantly to keep them in sync but that is not how factories work in Factorio. That is some other game that you are simulating that I am not aware of.

[ptx0]

...
I am saying that your approach is an inaccurate simulation because it does not account for intermediates. Yes, you can artificially 'wipe' factories constantly to keep them in sync but that is not how factories work in Factorio. That is some other game that you are simulating that I am not aware of.

wowza it's like I'm reading my exact response that I had to you earlier..

[ssilk]

wowza it's like I'm reading my exact response that I had to you earlier..

I’ll let it stand it here, because this discussion is currently useless.

[Dimava]

I made a mod for this
https://mods.factorio.com/mod/dupebox
please test and review

[steinio]

Wouldn't it be easier to prototype 4 times faster smelters?

[NotRexButCaesar]

Wouldn't it be easier to prototype 4 times faster smelters?

The video presented there is just a demonstration. Separately from that: the point is not ease, but UPS

[ptx0]

Wouldn't it be easier to prototype 4 times faster smelters?

The video presented there is just a demonstration. Separately from that: the point is not ease, but UPS

it'd be better for UPS, too.

[Qon]

The point of the abstraction idea (to me at least) is to not make the game easier. Faster assemblers just makes the game of logistics and game of space management and factory design x times easier when you make assemblers x times faster.

[ptx0]

The point of the abstraction idea (to me at least) is to not make the game easier. Faster assemblers just makes the game of logistics and game of space management and factory design x times easier when you make assemblers x times faster.

it doesn't though.

belts and inserters have maximum throughput limits. it imposes new design considerations unless you want to just leave throughput on the table, wasting it.

[blazespinnaker]

...
I am saying that your approach is an inaccurate simulation because it does not account for intermediates. Yes, you can artificially 'wipe' factories constantly to keep them in sync but that is not how factories work in Factorio. That is some other game that you are simulating that I am not aware of.

wowza it's like I'm reading my exact response that I had to you earlier..

Sorry, I must have missed this response. Can you point to it?

[blazespinnaker]

The point of the abstraction idea (to me at least) is to not make the game easier. Faster assemblers just makes the game of logistics and game of space management and factory design x times easier when you make assemblers x times faster.

Agreed. Maintaining some of the current challenges through abstraction helps maintain a lot of the fun of the game, IMHO.

That said, I think once you get to the scale where abstraction is UPS beneficial, I don't think these challenges really exist. You likely have many robots and practically infinite supplies from your mall. At that point, it's less about production, and more about logistical management of resources (miners + trains)

Using modified modules or uber assemblers is probably a reasonable stand in for most folks and is a common reply I've found when I've brought up this topic. It certainly will work just as fine for many purposes.

Abstraction is likely only interesting to a niche few, from what I've seen. For example, if you have a particular interest in building optimal unitary factories for various things.

There is an argument for using abstraction to remove some of the copy/paste element to the game, but that's a different discussion and really a different game.

Wouldn't it be easier to prototype 4 times faster smelters?

Now that you mention it, I have to say, you're exactly right. I don't see the practical difference.

However, this can be alleviated somewhat if OP removes the constraint of having the duped factories adjacent. Still, you run into this sync thing which really, uber furnaces and assemblers seems like a better approach with pretty much all the same aspects. The benefit of something like schalling's mod over this would be a lot less finnicky silliness that is going to come in play as this is used at scale.

https://mods.factorio.com/mod/SchallMachineScaling

I'd also be curious to see if the Mod author has tried using this at scale as well (say, around 100k spm). I tried, but the mod kept on crashing so I gave up.

Steino, let me know if you get a notification for this quote. Trying this instead of double posting.

[ptx0]

However, this can be alleviated somewhat if OP removes the constraint of having the duped factories adjacent. Still, you run into this sync thing which really, uber furnaces and assemblers seems like a better approach with pretty much all the same aspects. The benefit of something like schalling's mod over this would be a lot less finnicky silliness that is going to come in play as this is used at scale.

https://mods.factorio.com/mod/SchallMachineScaling

I'd also be curious to see if the Mod author has tried using this at scale as well (say, around 100k spm). I tried, but the mod kept on crashing so I gave up.

Steino, let me know if you get a notification for this quote. Trying this instead of double posting.

ok now I'm not sure if you're trolling me because I absolutely mentioned Schall Machine Scaling.

I'm using it with Bob's for 1 million SPM..

[blazespinnaker]

Not trolling at all. I'm not entirely sure we're saying the same though, though perhaps we are.

I think those mods make more sense than having to sync up factories using this abstraction approach, as practically speaking, it's pretty much the same thing and you don't have to deal with the synchronization silliness.

However, I believe there are other ways to abstract and not have to sync, which I think is an improvement over the uber assemblers .. at least for my use cases. The reasoning is that I see a benefit to abstraction when trying to devise atomic factories using base rules of the game. The issue with uber assemblers is that they take you away from the shared experience that is a part of vanilla.

Synchronization, though, is just a pain I see no profit from, and yes schall is the way to go if that's your thing.

[ptx0]

The issue with uber assemblers is that they take you away from the shared experience that is a part of vanilla.

i thought your intention was to focus on larger train networks? the larger assemblers that let you load up a train with fewer entities are perfect for that. they shift the UPS bottleneck to your electric network updates and train loaders.

[blazespinnaker]

Yeah, it's something I'm considering more for sure. We'll see how it plays out and what I have time for.

[ssilk]

Wouldn't it be easier to prototype 4 times faster smelters?

You’re right. That was the point, why I think this suggestion is at its dead end. The cloning idea is in my eyes the way out, because that behaves really differently. The point there is, that it needs to be feed equally to work efficiently, which is for bigger setups hard to achieve.

[blazespinnaker]

For the record, you have it backwards. But yeah, this is going round and round, and there isn't much point in discussing it as I really don't see the idea going anywhere anyways. I will say, to your credit, the kernel of the idea is pretty good (deterministic measurement of currently executing factories). It's a shame you stopped there.

[ssilk]

I changed my opinion. That happens. I don’t see that as bad.

But now when you directly point to it, thank you blazespinnaker, I thought about why.

My basic opinion about making Factorio faster is this: the abstraction/simulation must behave exactly as the original. That is only possible under certain circumstances. Because that little details is, what Factorio differs from most other games, because it brings in some kind of “reality”, that I even don’t see in all other new Factorio-like-games as Dyson sphere, Satisfactory.... it would brake my understanding of what Factorio is, if an abstraction is not simulated exactly enough.

That is not seen so by most here. They want just faster assemblies without that hazzle... I exaggerate!! But I saw the signs. So I changed my opinion, because keeping the spirit is for me much more important, than having faster factories.

And it’s unfair to actively vote against it. So I will try to be silent now.

[ptx0]

My basic opinion about making Factorio faster is this: the abstraction/simulation must behave exactly as the original. That is only possible under certain circumstances. Because that little details is, what Factorio differs from most other games, because it brings in some kind of “reality”, that I even don’t see in all other new Factorio-like-games as Dyson sphere, Satisfactory.... it would brake my understanding of what Factorio is, if an abstraction is not simulated exactly enough.

at risk of going off-topic, Satisfactory has some mechanics from reality that Factorio lacks.

if you run over your power limits, the fuse trips and the whole thing goes down.. and if you're overproducing or underproducing at a building, there's an add'l penalty of the startup cost for the machine - it takes a few moments to spin up, so keeping them at 100% efficiency is best. this mirrors real life.

[ickputzdirwech]

[ick] Merged two threads about abstract "black boxes" to increase performance.