Factorio Forums

This might be a bit philosophical, but I've started to wonder about measurements of so-called "mega-factories". Lately many of the new "mega-factories" I've seen consist of repeated copies of self-sustained micro-factories (hereafter "blocks") - including copying of perfectly shaped infinite ore patches right into the factory! This copying conveniently avoids all large-scale logistic problems, especially the one of getting ore into the actual base.

But if a bunch of copy/pasted independant blocks constitutes a megafactory, then why not copy a 100SPM block a thousand times and claim 100kSPM? I don't really understand the logic.

"It must run at 60 UPS" doesn't seem to be important either, because sometimes people claim X SPM with 30UPS, sometimes with 70UPS. And honestly if it was UPS bound, that would mean that the exact same base would have different SPM-capacity on different computers.

So...eh? Anyone care to explain?

I don't want to downplay anyones effort. I still enjoy reading UPS-optimization threads. I just don't understand where the "SPM" numbers come from anymore.

The speed-running community seems to have fixed definitions for "Something%" runs which make different runs comparible. Perhaps the megabase community lacks such definitions and everyone just claims "SPM" as they see fit? If that is the case would such a definition even be desirable?

I think if you are looking for UPS optimisation in a similar form to speedrunning, UPS Wars are close to that. Though that is somewhat far from megabasing.

The definitions for speedruns that you describe exist on the speedrun.com leaderboards and the attached community which organizes and votes on rules. There are also speedruns outside of those leaderboards that follow other rule sets, that are not comparable. I don't find this very unlike the UPS wars <-> general megabasing relation - defined rules allow for comparison, but there will always be creations outside of those rules.

The disadvantage of any defined rule set is that not everyone will like it. For something with an already high entry barrier like megabasing (not UPS wars), I could imagine that the entry barrier to any possible leaderboard would be so high that virtually no one will create things for it - there are already very few up to date megabases >= 10kspm as is. But I'll leave more detailed assessments of that to more experienced megabasers than me, I only have a meekly 1kspm to my name :)

I too feel that 100~500 blocks of X spm bases aren’t really the same as one large base. Maybe a format like 1000x100spm for your 100k example would be a nice way to convey what you mean.

For the editor and infinite ore patches, I think things should be separated into

1. Pure Vanilla
2. Modded Freeplay
3. Editor Freeplay
4. Custom Freeplay (ore patches edited on lab tiles)

I don’t think there is really too much competition, so these aren’t really “rules”, more like descriptions.

I find UPS management to be the least interesting aspect of building megabases, but it does have its place. Building a self-contained pod that can be "stamped out" is interesting. Stamping them out until the computer can't take it anymore, cap'n! is not. I like to look at the interesting ways that people have developed to handle complexity at massive scale, and stamped out copies that don't interact with each other (except through the artificial construct of the game engine underneath) is not complexity.

I also find the practice of measuring SPM in a vanilla base on only 6 science packs to be...

Things about megabases I find interesting:
1) Can it hot start (shut down the labs, let everything back up, restart the labs, does it all start running properly again without player intervention?)
2) Can it cold start (start from no items at all, not even fuel in trains (trains can have a designated start location), can it start and run properly without player intervention?)
3) If it can cold start, how long until first lab turns on? How long before it reaches full sustained production?

Cold starting a base that powers entirely from solar panels/accumulators is a very different beast from cold starting one that runs on nuclear (I don't think there is a way to cold start a vanilla base without some solar, but I haven't explored that very far).

Here is the semi-official measurement of SPM that you can use if you want to be "competitive".

• Produce all 7 science packs while researching Follower robot count.
• Measure over a long enough time period to demonstrate stable production.
• It is alright to use speed modules in labs, because we are measuring beaker consumption and not research.

But you are correct that SPM is only part of evaluating a megabase. The interesting megabase question has always been "Can you build a big base?"

SPM == Science Per Minute
or
RPM == Rockets Per Minute

That are two typical numbers, to compare megabase productivity.
SPM is measured on consumption side, while you research, most often, mining productivity. With interval of at least 10min, better 1 hour. Or you can measure this on less interval but you'll get peak SPM.

Some people prefer to use RPM, because it's more visible, and numbers are about 1000 less.

Another good number to use, if you want to compare two big bases, is a power consumption. But SPM/Power might differ a lot, depending on modules used.

Bilka wrote: ↑Tue Mar 30, 2021 4:50 pm I think if you are looking for UPS optimisation in a similar form to speedrunning, UPS Wars are close to that. Though that is somewhat far from megabasing.

UPS was just one example of things I find interesting about megabasing. I also find the logistic complexity at large scale very appealing. Makes me wonder if there would be interest in an UPS War about building one of those self-contained micro-factory blocks.

Bilka wrote: ↑Tue Mar 30, 2021 4:50 pm The definitions for speedruns that you describe exist on the speedrun.com leaderboards and the attached community which organizes and votes on rules. There are also speedruns outside of those leaderboards that follow other rule sets, that are not comparable.

I see. I'm not really into speedrunning so I was unaware that there are different rule sets...

Bilka wrote: ↑Tue Mar 30, 2021 4:50 pm I don't find this very unlike the UPS wars <-> general megabasing relation - defined rules allow for comparison, but there will always be creations outside of those rules.

...but for megabases there doesn't seem to be even a single ruleset? I've only seen that some builders self-impose rules on their own creations. Maybe if I assume that ("no bots") is a complete ruleset, even if it contains only one rule, I can compare at least some factories.

Bilka wrote: ↑Tue Mar 30, 2021 4:50 pm The disadvantage of any defined rule set is that not everyone will like it. For something with an already high entry barrier like megabasing (not UPS wars), I could imagine that the entry barrier to any possible leaderboard would be so high that virtually no one will create things for it

Wasn't really thinking of a leaderbord, but maybe "comparability" implies that? Though I'm not sure if such a thing doesn't already exist at least in peoples minds. After all, SPM is already used for comparison.

Bilka wrote: ↑Tue Mar 30, 2021 4:50 pm I only have a meekly 1kspm to my name :)

Grats though! I haven't even made it that far yet for various reasons (time, motivation, hardware, starting new maps too often...).

_____

Trific wrote: ↑Wed Mar 31, 2021 12:30 am I find UPS management to be the least interesting aspect of building megabases, but it does have its place. Building a self-contained pod that can be "stamped out" is interesting. Stamping them out until the computer can't take it anymore, cap'n! is not. I like to look at the interesting ways that people have developed to handle complexity at massive scale, and stamped out copies that don't interact with each other (except through the artificial construct of the game engine underneath) is not complexity.

Couldn't have said it any better. I used to have this sense of awe and wonder when looking at other peoples builds. The 700+ wagon "mega train" factory? Awesome. But yet another "I built 20 blocks of 500SPM each" factory? Insant boredom even if the individual blocks would've been interesting. The removal of all resource related logistics by using spawned ore patches as if they were infinity chests completely ruins the awe-factor for me.

_____

DaveMcW wrote: ↑Wed Mar 31, 2021 3:06 am Here is the semi-official measurement of SPM that you can use if you want to be "competitive".

But how is "time period" defined there? If I measure in real-world time then SPM becomes CPU dependant, and if I measure in in-game time then it becomes RAM-size dependant ("how many times can you copy/paste before the map can't be loaded anymore"). Neither of these seem meaningful to me.

Maybe the thread topic question just isn't very good. Maybe I should ask "If copy/pasting independant blocks is allowed, then why does anyone still care about total SPM anymore?". Basically this is me trying to understand the appeal of "multiple factories on the same map"-type "macro factories".

_____

Would be interesting to hear from one of the block-builders why they decided to copy/paste their micro-factory exactly 20 times and not 15 or 30 times.

Maybe SPM is just whatever the builder bothers to build, and the theoretical limits of a certain approach aren't really important.

I think many people building megabases are either just picking some arbitrary SPM as their "goal" to aim towards and seeing if they can make a factory that reaches that, or are just seeing how high they can get to before they either lose interest in that savegame, or lose the ability to run that savegame at a non-glacial UPS rate on their particular computer. If someone's inclined to make modules they can copy and paste in order to achieve this then yes, to some degree it is just about how many you can be bothered to build. Unless you *do* use infinity chests or infinite patches or something for the resource input, though, there is still an actual planning/logistics aspect at the global level: even if every module is identical and selfcontained, actually getting raw resources to all of them at the speed they require *does* get harder the more of them you build.

It's primarily only relevant whether different base's SPMs are really properly comparable if there is a specific set of rules/goals set out, and generally these rules/goals would make copypasting a single factory module not really relevant. For example, if your goal/competition was:

"Assuming you have an arbitrarily large amount of raw resources coming in from 'outside', what's the highest ratio of SPM to build area you can achieve"

then copypasting the factory module doesn't change the results and the challenge is purely about optimising production vs space, and for this kind of challenge just having resources coming out of an infinity chest or whatever is perfectly reasonable - it's abstracting away the "miners to inputs" logistical challenge in order to allow people to focus specifically on the production challenge. Things like this do get posted to the forum/subreddit occasionally, but they're generally going to be pretty difficult challenges since optimisation at this level is very hard, so I've not seen many that had a lot of different people participate, and I'm not aware of any "ongoing" ones that are structured similarly to speedrun leaderboards.

So.. maybe that partially answers your question, but I don't personally copypaste factory blocks in any of my games so can't speak directly to that
I've never really pursued SPM as a goal; once I've achieved a nice endgame factory that can do infinite research at even a fairly low SPM I tend to move on to another save where I try to meet different self-imposed restrictions or just use different major content mods, and anything I've built that might qualify as a "megabase" was to deal with cases where producing that much was actually necessary to reach the endgame due to the restrictions/mods/settings involved.

torne wrote: ↑Wed Mar 31, 2021 9:05 pm Unless you *do* use infinity chests or infinite patches or something for the resource input, though, there is still an actual planning/logistics aspect at the global level: even if every module is identical and selfcontained, actually getting raw resources to all of them at the speed they require *does* get harder the more of them you build.

Just to be clear what I called "ore patches like infinity chests" is things like in this thread. On the pictures you can clearly see that ore patches have been "copy/pasted" to fit the factory, and I assume that they have a density so high that they are effectively infinite. (Funnily enough you can see that the factory also has actual infinity chests next to each miner. Maybe those were used before spawning in the "real" ore?)

No large-scale logistics for ore transport are required with that approach. I haven't seen any "copy block" based map that bothered to actually feed each block from "natural" resources. If you know any please feel free to post links.

torne wrote: ↑Wed Mar 31, 2021 9:05 pm "Assuming you have an arbitrarily large amount of raw resources coming in from 'outside', what's the highest ratio of SPM to build area you can achieve"

Space restrictions can be interesting on a small scale sometimes. But mostly I prefer to build properly spaced.

eradicator wrote: ↑Thu Apr 01, 2021 12:43 am

i don't see why that's a problem at all. i play in /editor in my bob's map (see signature) and it's plenty of fun. it's all about the player who's posting their base showing you what they thought was fun and exciting. and you're just like "i don't get it". that's fine, but it's because you don't want to get it. not because it's hard to understand.

Dixi wrote: ↑Wed Mar 31, 2021 7:35 am
SPM is measured on consumption side, while you research, most often, mining productivity. With interval of at least 10min, better 1 hour.

Mining productivity only requires 6 different science packs. Measuring SPM that way deserves an asterisk at least. I don't care if people do mining productivity (or bot speed) while building up the base, since those are useful and follower robot count is useless, but measurements for real SPM should be done with all 7.

Trific wrote: ↑Thu Apr 01, 2021 12:56 am

Dixi wrote: ↑Wed Mar 31, 2021 7:35 am
SPM is measured on consumption side, while you research, most often, mining productivity. With interval of at least 10min, better 1 hour.

Mining productivity only requires 6 different science packs. Measuring SPM that way deserves an asterisk at least. I don't care if people do mining productivity (or bot speed) while building up the base, since those are useful and follower robot count is useless, but measurements for real SPM should be done with all 7.

that's... not how SPM is measured. sorry, but military science being included is the exception rather than the norm. you are welcome to do that for yourself but expecting others to is just weird.

ptx0 wrote: ↑Thu Apr 01, 2021 12:58 am
that's... not how SPM is measured. sorry, but military science being included is the exception rather than the norm.

Why is it the norm? Doesn't make any sense. Turning biters off, manipulating resource distribution I can accept, but just leaving out one science pack leaves a lot of the work on the table. Using a 7 pack capable factory to run up your mining productivity numbers is fine (done that myself), but not building a 7 pack factory to start with? Don't really care about "the norm", it will always be an asterisk in my opinion.

eradicator wrote: ↑Thu Apr 01, 2021 12:43 am Just to be clear what I called "ore patches like infinity chests" is things like in this thread. On the pictures you can clearly see that ore patches have been "copy/pasted" to fit the factory, and I assume that they have a density so high that they are effectively infinite. (Funnily enough you can see that the factory also has actual infinity chests next to each miner. Maybe those were used before spawning in the "real" ore?)

No large-scale logistics for ore transport are required with that approach. I haven't seen any "copy block" based map that bothered to actually feed each block from "natural" resources. If you know any please feel

Space restrictions can be interesting on a small scale sometimes. But mostly I prefer to build properly spaced.

You just don't get it. This is perfectly realistic/authentic to actual settings.

On vanilla without editor, if you want to build something like that, you'll tweak map generation settings to fit it into your requirement. High ore frequency, high density, large size, no pollution, no biter, etc.

Factorio's belt is optimized so that a compressed, uninteruppted belt is count as 1 segment. so no matter how long you route the belt from patches, as long as it is compressed, the UPS is the same.

ah i guess i totally missed the context and questions here earlier, i'm sorry.

you have some good questions. and maybe i can try and explain how i see it.

eradicator wrote: ↑Tue Mar 30, 2021 3:59 pm This might be a bit philosophical, but I've started to wonder about measurements of so-called "mega-factories". Lately many of the new "mega-factories" I've seen consist of repeated copies of self-sustained micro-factories (hereafter "blocks") - including copying of perfectly shaped infinite ore patches right into the factory! This copying conveniently avoids all large-scale logistic problems, especially the one of getting ore into the actual base.

But if a bunch of copy/pasted independant blocks constitutes a megafactory, then why not copy a 100SPM block a thousand times and claim 100kSPM? I don't really understand the logic.

i guess it's situational. everyone has a different goal - megabase and SPM are just measurements, like any kind of metric, it's just a small part of a whole picture.

any base can be boring. Xterminator5 did a base review of my first base i'd ever played - it's not a high SPM factory, it never even launched a rocket. but the story behind it and some of the stupid contraptions we made on the map to compensate for our lack of understanding of game mechanics were what made it interesting. we just stuffed everything onto a single belt because it seemed to work just fine.. until it didn't..

i don't quite understand the arbitrary figures people aim for with SPM either. i think it's also that the 100k SPM would probably be unplayable and pointless as a result.

i'm working on a 1 million SPM base. I chose that number because it was absurd and I didn't think I could do it at 60 UPS, but I found some engine things to exploit and it might still be possible.. to me this is not the exciting part, but the challenge of all the direct-insertion builds, even using /editor and spawning in infinite patches. it's taken me more than 1050 hours of real playtime. explaining it to others isn't easy. not many people understand why it's neat. they think it's just easy as heck to do. one asked me why not just use infinity chests? because that'd be like the eagle bringing the ring to mordor. it misses the whole point of the story.

there's also the aspect that copy-pasting something repeatedly isn't very exciting. i just commonly see it as though they were wondering whether a certain build style would scale up if they did it a large number of times. it's a quick test to pull off. i say it's not exciting because you're hardly playing the game. you're just copy pasting. the play time was spent designing the initial module. maybe there's just some satisfaction gained in seeing it all work.. but i'm with you - interconnecting rail networks are fun to watch. junctions are fun to design. city blocks have the fun to watch aspect but no way to optimise it.

eradicator wrote: ↑Tue Mar 30, 2021 3:59 pm "It must run at 60 UPS" doesn't seem to be important either, because sometimes people claim X SPM with 30UPS, sometimes with 70UPS. And honestly if it was UPS bound, that would mean that the exact same base would have different SPM-capacity on different computers.

i guess that comes back to the context and the story behind the base. what were they going for? sometimes it's interesting to see someone's base that runs at 30 UPS. i'm a site reliability engineer, and performance is always interesting to me.

the UPS declaration is just a measurement of how much CPU time the base took on their system. maybe their goal was to maintain 60 UPS or greater - maybe it was just to see how slow it got once they achieve their goal.

eradicator wrote: ↑Tue Mar 30, 2021 3:59 pm So...eh? Anyone care to explain?

I don't want to downplay anyones effort. I still enjoy reading UPS-optimization threads. I just don't understand where the "SPM" numbers come from anymore.

SPM is measured in-game. the production graphs measure by tick clock, not wall clock. so the SPM in the game is as if you were the engineer experiencing their "realtime".

it'll be wrong if you ran at !=60 UPS and watch the items moving on the belts keeping track manually, for instance. but the "Playtime" indicator is also wrong once it's running below or above 60 UPS.. it's assuming 60 UPS.

eradicator wrote: ↑Tue Mar 30, 2021 3:59 pm The speed-running community seems to have fixed definitions for "Something%" runs which make different runs comparible. Perhaps the megabase community lacks such definitions and everyone just claims "SPM" as they see fit? If that is the case would such a definition even be desirable?

SPM is just what the map can produce. depends on mods to know how much overall 'resource consumption' is going on / how many machines are active.

as far as i can tell there's a "competitive SPM" with two categories; military science or not. and it usually has to be vanilla or no one cares. the competition is to see how big you can go and maintain 60 UPS on *YOUR* computer. then everyone else sees how fast it runs on their computer.


the number of train stations is also a pretty good one.

siiid wrote: ↑Thu Apr 01, 2021 1:29 am Factorio's belt is optimized so that a compressed, uninteruppted belt is count as 1 segment. so no matter how long you route the belt from patches, as long as it is compressed, the UPS is the same.

Segments are limited to 200 tiles long, as shown in the debug info. The FFF where they were revealed had an even shorter length, just 100.

Silari wrote: ↑Thu Apr 01, 2021 5:28 am

siiid wrote: ↑Thu Apr 01, 2021 1:29 am Factorio's belt is optimized so that a compressed, uninteruppted belt is count as 1 segment. so no matter how long you route the belt from patches, as long as it is compressed, the UPS is the same.

Segments are limited to 200 tiles long, as shown in the debug info. The FFF where they were revealed had an even shorter length, just 100.

Good to know. 200 tile long segment is still mostly long enough on high ore density settings.

siiid wrote: ↑Thu Apr 01, 2021 4:23 pm Good to know. 200 tile long segment is still mostly long enough on high ore density settings.

you're going off-topic here. but you're also wrong.. trains win over long distances.

ptx0 wrote: ↑Thu Apr 01, 2021 5:08 pm

siiid wrote: ↑Thu Apr 01, 2021 4:23 pm Good to know. 200 tile long segment is still mostly long enough on high ore density settings.

you're going off-topic here. but you're also wrong.. trains win over long distances.

The point is, on high ore density setting, you don't need long distances. There are plenty of ore patches within 200 tiles reach.

siiid wrote: ↑Thu Apr 01, 2021 1:29 am You just don't get it.

Tautological. If I "got" what other people find interesting about copy-block factories I wouldn't have made a thread to ask. So please calm down.

siiid wrote: ↑Thu Apr 01, 2021 1:29 am On vanilla without editor, if you want to build something like that, you'll tweak map generation settings to fit it into your requirement. High ore frequency, high density, large size, no pollution, no biter, etc.

I'm perfectly fine with any map settings that can be chosen in the gui. Any base that exists long enough would hopefuly eventually reach a state of "effectively infinite" ore patches. Either by expanding/exploring outwards to larger ore fields, or by reseraching sufficient levels of mining productivity.

siiid wrote: ↑Thu Apr 01, 2021 1:29 am This is perfectly realistic/authentic to actual settings.

Naturally occuring fields do not conveniently spawn in drill-sized patches that perfectly fit 50+ aligned copies of the same micro-factory. Not having to deal with natural ore distribution removes - to me - too much of the puzzle. I can understand that some people might find that part of the puzzle to be not worth the time spent on solving it. But pretending that spawned patches are "as authentic as actual settings" is too far a stretch for my taste.