Game optimization and playability

[ssilk]

... the ultimate goal is of course to automate the building of the factory. Automating the automation. The player gets useless and can only look, how the factory builds itself faster and faster. When it comes to the cpu limit it builds a faster computer on which it can run on. And the exponential growth goes over into hyperinflation. The dark universe is graved under this... super crazy big factory. A big explosion of light...

... And some background speaker whispers “And there was light...”

[blazespinnaker]

I've mentioned this a few time, but it really does bear repeating. Reducing UPS consumption from factories allows to build a larger and more complex train networks. For folks who want to play around with trains more extensively, this is quite useful.

Also, optimizing UPS will be useful for folks with slower computers who want to achieve something on par to the community at large.

[jodokus31]

I've mentioned this a few time, but it really does bear repeating. Reducing UPS consumption from factories allows to build a larger and more complex train networks. For folks who want to play around with trains more extensively, this is quite useful.

Also, optimizing UPS will be useful for folks with slower computers who want to achieve something on par to the community at large.

It, of course, does help, no question. Lets assume, we can double the performance (which already happened), which would be pretty huge. It won't be long, when we stand here again and again. There's no end to it.
It's of course desirable to always optimize.

[blazespinnaker]

I've mentioned this a few time, but it really does bear repeating. Reducing UPS consumption from factories allows to build a larger and more complex train networks. For folks who want to play around with trains more extensively, this is quite useful.

Also, optimizing UPS will be useful for folks with slower computers who want to achieve something on par to the community at large.

It, of course, does help, no question. Lets assume, we can double the performance (which already happened), which would be pretty huge. It won't be long, when we stand here again and again. There's no end to it.
It's of course desirable to always optimize.

Maybe. I suspect there are advanced train puzzles which you don't have to solve with current UPS limits, but they will be largely solved by extending out UPS. Unless wube adds more at scale train features, this should be the end of it. Certainly the current round of train features almost beg for larger factories. In some ways, wube is causing the problem by making trains at scale easier to use

Allowing folks with slower computers to have a roughly equal experience and achieve similar SPM as the rest of the community at large does not need repeating. Consumers of the mod will be niche at most.

Another thought, is UPS hit from adding mods and MP. This will help those scenarios achieve similar SPM experiences as the community at large. This leads to greater shared experiences and reduces bifurcation in the community.

Edit to add: if you play marathon mode, infinite tech is 4x cost.

[Zavian]

Maybe. I suspect there are advanced train puzzles which you don't have to solve with current UPS limits, but they will be largely solved by extending out UPS. Unless wube adds more at scale train features, this should be the end of it. Certainly the current round of train features almost beg for larger factories. In some ways, wube is causing the problem by making trains at scale easier to use

What advanced train puzzles are you referring to? Things like train throughput issues and congestion? Whether they will be a problem depends on how you design your base. To my mind designing to avoid congestion/throughput problems is much easier (but potentially less interesting/frustrating) than adding more tracks in an attempt to solve them.

But if you want to try solving congestion/throughput issue, then go ahead. Whatever floats your boat. Try building a 1000+ SPM base with a central smelter producing iron/copper/steel/bricks. If you can get that working then try scaling up to 2k SPM. I will note that even my 10 year old PC can handle 1500 SPM ( it can actually do it at 90 UPS, so it should be able to handle 2000 SPM) on a map with biters and pollution enabled, and with a base design where I made some design decisions based on aesthetic, rather than ups considerations. So most people should be able to run a 1000 SPM base without performance issues, assuming a decent base design and that they disable or clear most of the biters out of their pollution cloud).

Allowing folks with slower computers to have a roughly equal experience and achieve similar SPM as the rest of the community at large does not need repeating.

Anything anybody does to improve performance will simply enable everyone to build larger factories. So whilst people with slower computers might be able to build bigger factories, players with faster computers will also be able to build bigger factories. So folks with slower computers will never be able to reach the same SPM as folks with faster computers. (Unless something starts artificially capping active entities or SPM or something similar).

The real issue is whether folks with slower computers can access interesting gameplay. Obviously different players will have different opinions on what they consider interesting gameplay.

Another thought, is UPS hit from adding mods and MP. This will help those scenarios achieve similar SPM experiences as the community at large. This leads to greater shared experiences and reduces bifurcation in the community.

Many (most?) mods have a negligible impact on ups.

[blazespinnaker]

What advanced train puzzles are you referring to?

Allowing stops to limit trains simplified things a bit. It was a nice change, but removed some of the challenge to managing throughput of trains at scale. It'd be interesting to see how well it works when quite a lot more train and stops are added to the system. You don't really know until you try.

And then there is the recent UI change to improve managing mass trains.

I'm sure dev was well meaning with the changes but it does make one think about bigger factories and larger train networks.

Anything anybody does to improve performance will simply enable everyone to build larger factories.

As I mentioned above, consumers of the mod will likely be rather limited. Given the reception so far, I am skeptical that everyone will be building larger factories. However, if you have a slow computer you would at least have an option that allows you to build a base that matches what the community at large is producing.

Many (most?) mods have a negligible impact on ups.

I'll have to take your word for it. I've tried only a few mods and have yet to profile them to see their precise time usage at scale. I don't play MP that much to know what impact it has on UPS.

[ptx0]

gee, you took over another thread with your ideas and keep repeating yourself. it'd be bad enough if you just kept repeating yourself, but you're repeating the same things as if others don't understand, or didn't see it before? they just disagree with you. or think you don't know what you're talking about. you said you've only played a couple mods, haven't profiled them? don't know how MP affects performance? not sure what you've got to offer to the discussions on performance except for a whole lot of non-starters.

for one, the insistence that a factory must grow forever. it's simply not true - and you want to clone factories, but not trains? you haven't profiled the game, so I guess you don't realise that rail network update time is eventually going to be your next highest CPU consumer. and then what? abstract the trains away?

you may as well just open up a spreadsheet and type some formulas for the recipes and just keep scaling up the number of rows and columns that are occupied by machines. that will scale forever and it's the perfect game for you.

[Koub]

[Koub] I've seen enough fighting recently, I won't tolerate yet another thread derailing.

[blazespinnaker]

Yes, I am getting a bit repetitive. However, you'll be glad to know I've exhausted my supply of things to say about this topic.

So I'll leave off here except to say to new folks, if you come across this thread at a later point and are still interested, feel free to PM me. I'd be happy to share what I've done.

[4xel]

Building a factory takes time. Building a bigger factory takes more time. Humans have a limited lifetime. Therefore, if the machine and the game are sufficiently powerfull that during a human lifetime, one cannot build a factory big enough to hit the hardware limits, there is enough optimization.

This is pure bonkers. Early on, it takes 6 minutes to double the size of a factory. Later on, accounting for the bots needed to place things for you, it probably takes less than 30 minutes. If you use prod modules3 every were, it probably takes 5 hours, lets call it 10. With recursive blueprints, resource is really the only limit, but even assuming you place entities by hand, if you're serious and prepared, you can reliably double the size of your mega-base every 2 weeks.

By comparison, the density of transistor in dense integrated circuit doubles every 2 years, slowing and closing in to theoretical limits. There is no way for hardware to ever be ahead of players on that particular issue.

[bobucles]

Factory production in Factorio is incredibly fast paced. Most objects produce in seconds, and fractions of a second with beacons. Inventory space is large and trains have monolithic item transporting rates.

Frankly, if you want more factory or train networks, it makes sense to start nerfing things. Cut a cargo storage down to 10 slots and guess what? You suddenly need 4 times the trains. Increase production time by 2x or more? Now you need twice the factory to get the same result. Plus there are amazing scaling mods like clustertorio, which let the factory scale beyond the limits of 1 PC.

You can build big factories. In fact they can get SO big that humans start losing track of where their stuff is. I don't think it needs to get bigger than that.

[ptx0]

i also disbelieve that anyone has surpassed this build: https://www.youtube.com/watch?v=dY2nxVNBHQs

which should be noted - needed none of this improvements as suggested here in this thread.

[starlinvf]

I've mentioned this a few time, but it really does bear repeating. Reducing UPS consumption from factories allows to build a larger and more complex train networks. For folks who want to play around with trains more extensively, this is quite useful.

Also, optimizing UPS will be useful for folks with slower computers who want to achieve something on par to the community at large.

It, of course, does help, no question. Lets assume, we can double the performance (which already happened), which would be pretty huge. It won't be long, when we stand here again and again. There's no end to it.
It's of course desirable to always optimize.

I'm trying to find the theorem that bares this concept, but don't know how to describe it in a way that google gets me the results I need. Anyone have a name on this?

[Zavian]

I'm trying to find the theorem that bares this concept, but don't know how to describe it in a way that google gets me the results I need. Anyone have a name on this?

Try searching for stuff like exponential growth.

[gujarat6]

Here I am, 1 year later, with the same old problem. We got back into the game with some friends during summer break, as always, but our fast-paced playthrough destroyed new save games one by one. We reach the ceiling faster and faster every time we start a new game, computers aren't being developed quickly enough to satisfy our sense of "FACTORY MUST GROW", which makes me incredibly sad. Join me as we grieve over the inevitable fate of these games.

[cpy]

I built huge factories in game and I never ran into performance problems in factorio. Never.

[blazespinnaker]

Really, the problem isn't so much optimization, but rather de-optimization. Factorio has to be careful about building features which encourage too much at scale building, because it just leads to dissatisfaction seen in this thread.

map mode was awesome, but now of course, we want to use it.

[ssilk]

@gujarat6 without further knowing, what you did, on which system you’re playing, which mods you use, and many things more, this is just a rant.

So, please, before you might come back next year and continue , provide us with the information we need to tell you, why you have problems. Because — when you look at this thread — it seems that your problem is not very common.

Questions:
What kind of system you’re using? Operating system, CPU, RAM, RAM-type, graphics card. (Did I forgot something important?)
Do you use mods? Which ones?
Does it happen in multiplayer or single?
How do you build your typical factory? Provide us screenshots/save.
How fast is it going (ups/FPS)?
When in your build process do you start to see it lagging?
More things we might need to know about?

And I think you’re false here. This is more a support question. but let’s see.

[gujarat6]

@gujarat6 without further knowing, what you did, on which system you’re playing, which mods you use, and many things more, this is just a rant.

So, please, before you might come back next year and continue , provide us with the information we need to tell you, why you have problems. Because — when you look at this thread — it seems that your problem is not very common.

Questions:
What kind of system you’re using? Operating system, CPU, RAM, RAM-type, graphics card. (Did I forgot something important?)
Do you use mods? Which ones?
Does it happen in multiplayer or single?
How do you build your typical factory? Provide us screenshots/save.
How fast is it going (ups/FPS)?
When in your build process do you start to see it lagging?
More things we might need to know about?

And I think you’re false here. This is more a support question. but let’s see.

Yeah I mean I'm playing on 3900X with 32 GB 3600mhz ram and GTX 1070 Ti. Win10 obviously. We are playing on vanilla sometimes, sometimes with the factorio extended mods family, it always ends the same way lol. Single-player games last longer, but it's the same problem, just a bit further down the line. We make the factory bigger and bigger, delete the whole thing at some point, place a block of few thousand furnaces, and build around that. Biters disabled (waste of UPS). I'm not even sure if there is a point in continuing this, because it will end the same way as always at some point. You cannot get the bases big, you can design them big with perfect resources and cheats, you can't play big and do what you want, place how many entities as you want cuz you'll lag the game to death. Always there is this boundary which you cannot cross, you just can't, game engine and CPU power don't allow it simply, which is really sad, because the game is crazy good. Have a good evening Sir.

[ptx0]

Win10 obviously.

a lot of us play on Linux. you'd be surprised.