Optimising UPS; The True Enemy of Megabases | Factorio Tutorial/Guide/How-to

[@Nilaus]

All Master Class Blueprints are available on FactorioBin Overview and direct links to all Blueprints: nilaus.atlassian.net/l/cp/HBEUm524 (Pastebin links no longer work)

[@bobbybobman3073]

If you select the the check box labeled show-entity-time-usage, (it should be the one directly under, show-time-usage), it will display a more detailed break down of entity time usage, with classes like, inserters, mining, and assemblers (and many more). Also feel free to delete this comment and post a separate pinned comment about this. Just figured that you'd want to know this, and hopefully this info can help others.

[@juanfranciscoronanduano2441]

Sometimes the production bottleneck of your factory is your PC and you just need to expand it. The PC you run the game is just another component of the factory

[@delcim]

this comment is underrated

[@nitrotypegod9598]

The factory must grow

[@Akira23453]

Masterclass on optimizing bottlenecks in pc performance?

[@kevin800404]

Mega component

[@KeinNiemand]

The problem with that is is that the game mostly needs more single threded performane to run faster something that doesnt increase more then like 5-10% cpu generation so even always buying the best cpu wont do much

[@bekar_au]

I think one important thing that wasn't mentioned is power. For UPS stability, power needs must always be met. The moment power dips below, the UPS will tank due to all of the performance calculations that needs to happen on any electrically powered entity. Now by the time most people get to a base this size, power levels should be fine, but for pure solar builds that sometimes dip (even a little bit) when accumulators run out, bad things happen.

[@TijmenZwaan]

The information you're providing about the transport lines at 13:40 is incorrect. The compression of belts is completely irrelevant for performance. What matters is the number of sections that the belt gets divided into. Splitters split up belts into multiple sections, as do inserters that interact with the belts. You can see these sections by enabling the "show transport lines" debug setting. You will see lines on the belts, with arrowheads indicating the start/end of a section. Each of these sections will only need to be updated once per tick. The way this works is that the game only needs to update the distance between the start of the section and the first item on the belt. All gaps after the first item remain exactly the same so they don't need to be updated. Which is why gaps absolutely DO NOT MATTER. (This is explained in more detail in FFF-176) What does matter is whether a belt is *moving* or not. A belt that is backed up will not be processed anymore, so letting your belts back up is still a good idea, but not because it removes gaps. You can also just leave it empty, or just have a few items at the end of the line with no new items coming in. All of these are handled the same. It seems like a lot of people have misinterpreted FFF-176, because it mentions the gaps between items, so people seem to assume that the gaps are what matters. But the point of the optimization is that the gaps never change except for the first, so they don't need to be updated, except for the first. So gaps don't matter. Only sections do. What this means is that balancers should be avoided, because those actually cause multiple belt sections to be created. As well as minimizing the inserter interactions with the belts. (not only because inserters are expensive themselves, but also because they break up belts into sections). One thing to note here is that you can have up to 3 inserters in a row (directly adjacent) grabbing from the same belt while still only breaking up the transport lines once. So it's preferable to place inserters directly next to eachother where possible. It's also preferable to make sure that inserters are only grabbing from once side of the belt at a time (for example by using the lanes for different resources) as the inserter will only break up the transport line that they're grabbing from, leaving the other line of the other lane untouched. If you can, you should probably add a disclaimer in the comments/description, as this is basically misinformation that seems to still be very prevalent in the community. (I was directed to this video because of a discussion of this topic in the official discord) If you're unsure whether I'm correct, you can check out a bunch of benchmarks that have been performed by the fellows at the technicalfactorio subreddit and discord. I'm also happy to answer any questions in more detail on the official discord (if I know the answers, of course). Greetings, Swan - Factorio discord/subreddit moderator

[@dragonarity]

Woah thanks for the information

[@bradleyhamilton4409]

You've made this game much more fun for me, thanks for the good content

[@Borthralla]

Thanks for this Nilaus, I often here people talking about UPS but now I feel I understand it a lot better. Now I know why you hate steam storage!

[@E-102_Gamma]

Something that I feel is worth clarifying: a UPS-optimal belt doesn't need to have items in both lanes. All that matters is that there are no gaps between the items on the belt, so a belt with a single full lane of items (a half-belt) is fine.

[@dfarmer_]

That's VERY helpful!

[@Xisdown]

Now I understand why you don't like efficiency modules. You prefer speed modules to have less entities. I liked this video once again. But still: I need an advanced train master class

[@E-102_Gamma]

The higher-tiered efficiency modules are also just kind of laughably underpowered. Eff mod 1s are great for reducing the pollution of mining drills and such, but the upgraded versions are quite niche. Two tier 2 eff mods can bring a machine's energy consumption down to 20% (the limit) by themselves, which can be helpful when working with machines that have only two module slots, but that's not terribly important in practice. And then Efficiency Module 3s... are a joke. They don't even begin to live up to their absurd price tag, and don't really have any good use aside from crafting Spidertrons. They are out-performed by the other types of modules, and they aren't even competitive with the lower-tiered modules that they're made from. They're just awful.

[@Xisdown]

@@E-102_Gamma yes. The upgrade for the higher tiers really is low. But I still use them sometimes. In combination with speed modules you "need" tier 3. 1x speed module 3, and 3x efficiency 3 perfectly results in 20% energy consumption. Of course the production of the efficiency modules comes last, and only works, when other modules are already produced

[@ColonelSandersLite]

Here's a tip - From what I'm seeing, you're balancing the amount ore belts just before they go into the smelters and you are then balancing the freshly smelted plates and, where you aren't using direct mining, you are also balancing the ore before it goes onto the train. You don't need to do all of that. You only need to balance that once, not 3 times! Since you have mostly direct mining resulting in uneven ore loads, I would suggest balancing before smelting. In doing this, the plates will already be balanced. If your base isn't using direct mining, you need only balance the load of ore when you load it onto the train.

[@mertserimer6928]

Well no, because if you are not using bots and using belts for feeding assembly machines, then one side will be used more. Thus chests for trains' unloading stations will be used unevenly which will make belts not saturated from stations. Thus you need to make sure that you are getting resources from chests evenly. But aside even with that you dont need 3, you only need two.

[@Bastelkorb3]

Regarding the biter performance cost, the biggest increase was in "entitys" from 10 to 13. After you deleted all biters and pollution, entity update goes down to 7. Just to make clear its not the path finding which kills your performance but calculations regarding pollution (it would be interesting if the pollution is the biggest problem tbh) and creation of biters and nests.

[@dj_spare]

I am running 50 UPS at the moment, base is getting out of control. Will kill pollution and biters tonight in my save to see how many entity updates it saves

[@dj_spare]

Disabling didn't do much for me. However I have another problem with Transport Lines taking 4 alone on average... Entity Updates stays around 10 for me

[@keonix506]

Nuclear is actually not that bad in terms of performance if you use reactors to transfer heat instead of heat pipes and feed water directly from pumps to boilers (one pump for 9 boilers without connecting to others, so there is no sloshing in pipes)

[@ChrislyBear]

What a great video! This explains so much semi-knowledge that is floating around in the Factorio community!

[@murrayboyton8779]

Would appreciate a Master class on Death Worlds, I know you have a series on one, but I would like more detail on starting strategies and priorities. Death Worlds scare me as a noob.

[@melind82]

I'd actually suggest just diving in. They aren't as hard as you may think if you are familiar with the game (have launched a rocket). My first death world games were some of my favorites.

[@reina4969]

In addition to what melind82 is saying, Nilaus never really felt the pressure of a death world because he used bots (mod) to accelerate base development before his pollution grew enough to spark biter attacks. We all know why he did it, but it wasn't a death world.

[@Hawk7886]

Yama Kara has a really awesome death world series where his biter settings are absolutely nuts.

[@reina4969]

@@Hawk7886 Have you come across any quality Rampart mod death world series?

[@pauljs75]

Moar dakka, as the old adage goes. And biter behavior tends to be predictable. Also cluster turrets when you can early on, as biters tend to focus on the first one that draws aggro while the adjacent turrets work on the biters. Thus small clusters last a bit longer than single turrets spread out to have the same perimeter coverage. Also all weapon types are complementary, just when you'd think the first tier of turrets aren't counting for much - uranium ammo becomes available and makes them quite valuable again. They're balanced to overlap with differing ranges, DPS, and resource demands, even if it seems easier to play favorites. If you have good power redundancy with a surplus and have the defense perimeter all botted and belted, then at that point you're pretty much set.

[@ScorpiusZA.]

17:31 Now I understand why you keep spamming solar array's. I would have thought that going whole hog on nuclear would have been a better idea.

[@ztac_dex]

It's a good way to use a lot of empty space when you expand

[@CreativeFunction]

I would be interested to see someone actually fully build nuclear vs solar. The fluid and heat time to process in this video was tiny. Trains, bitters, and many other things took up much more. I wonder if there's an argument for one massive perfect ratio nuclear plant instead of any panels.

[@juanolotgn]

Seems like the bulk of the load of biters is in the "entity update" - it went from 14 to 6

[@Hexadris]

Hey guys, today i upgraded my CPU from a ryzen 9 3900x to a ryzen 9 5900x and really got noticable results in factorio. i tested with my factorio extended megabase which isnt fully completed, it currently provides about 8k science/min but space science isnt build yet an purple is cant keep up at the time. anyways under full production my game updates where at about 2.3 - 2.6, now with the new cpu it sits at 1.6 - 2.0 with da tendency to stick at 1.8. So this leaves me space to build a base 8 times bigger while my old cpu only permitted around a times 6, thats about an increase of 33%. color me impressed.

[@TheKetsa]

Upgrade from a Ryzen 9 3900X ... Oo Talk about first world problems.

[@Hexadris]

@@TheKetsa well i play mostly only cpu intensive games and i also need the cpu for my work, so it actually was a worthwhile upgrade. considering that the old one still had a considerable resell value.

[@Newspaperman57]

Nice. I have a 3900x too, and have been thinking about how a 5900x will perform in factorio with the higher IPC and boost-frequencies. This is almost spot on with my "calculations" of theoretical improvement, but nice to hear real world use too. Did you change out the RAM or anything else at the same time, or was it only the CPU you changed?

[@Hexadris]

@@Newspaperman57 i didnt change the ram, i am having 3600mhz at cl18, i did try upgrading it to 4000mhz to improve fabric clock, but thats not an automatic uplift and requiers manual tinkering, i didnt consider it worth the hassel and returned the ram.

[@djsnowpdx]

If you are indeed building a PC in late 2022 for Factorio (lucky you!) then the Ryzen 7 5800X3D is the only CPU worth buying, as it can process three updates in the time it takes an Alder Lake i7 to process two. Make sure you understand the difference between the 5800X and 5800X3D - the 3D chip is 50%-100% more expensive, but in Factorio specifically, it’s 70%+ faster. In early 2023, the 5800X3D is set to be replaced by Zen 4 3D V-Cache CPUs, and at that point, the advice will change, as those processors should be roughly 25% faster again. All that said, The Ryzen 5 5600X and up, Zen 4, Alder Lake, and Raptor Lake all run Factorio far better than any CPU from two years ago could, and I have never managed a big enough base to dip below 60 UPS, though I’m currently running a Ryzen 9 5950X (bad price to performance for this game), and my most recent base is by far the biggest I’ve ever made. Props to the developers for making this game so performant.

[@DaveHcontrarian]

late 2023 update?

[@memory_stick]

Ryzen 7 7800X3D is the Zen4 equivalent of the Ryzen 7 5800X3D (8Cores, 3D V-Cache). There's also the more expensive 7950X3D (16 cores) and 7900X3D(12 cores) but both of them don't offer any value in Factorio

[@ritinfek]

Where is the part 3 of the Masterclass focusing on Circuit Network ?

[@masterphillips]

Hey Nilaus, great work as always. I'm working on an Omnimatter and FE+ base at the moment, about to scale to megabase size, and I will definitely be using as many of these tips as possible! Thank you!

[@primarykeyz5527]

10:15 FYI: It's F4 for the debug menu not F5

[@damnthatisnice]

I‘ve seen a reddit post where a user self-programmed a factorio client with full multicore support and he gained ~3,8 times more performance with 4 cores versus the normal client. He said the devs said multicore is not possible but he proved that it is and actually one dev replied that they look into it and op should provide a link to its client. So maybe there is hope for insane bases with almost no ups issues. No multicore support in 2020 is actualy a shame but i guess since this game is rather old I‘d give it a slight pass.

[@wretlaw1203]

to be fair, that code he made doesn't incorporate half the game, and it's quite possible that the way Factorio is now would be impossible to retroactively make it multicore compatible, and would require a rewrite of major parts. I remember one of the arguments against multi-core Factorio was that multicore does not do deterministic well, due to asynchronous computation, but given a small overhead cost of say, 10%, for syncing cores, the addition of 300% for just three more cores would be a game changer (and many cpus these days have more than four cores, and counting).

[@WilliamSpano]

Mega base! I still need to get to Kilobase or Decabase. I'll start micro and factor 1.

[@DFPercush]

I don't see why it couldn't be done. Yes syncing is one of the hardest things to do in programming (if you have proper abstractions it's really not that bad), but the thing is, each frame depends on the state of the last frame, not the previous item you're processing. If every single machine depended on the last machine in the list, and they had to be in order, it wouldn't work. But the machines only depend on their own state and their surroundings like inserters from the previous frame. It might cost twice the memory depending on how it's currently done, but the order of things should only matter in terms of stages, like circuit networks, then train pathing, then inserter rotation, then crafting, etc. All entities of the same type should be able to be processed in parallel. I'd really like to see that happen.

[@okuno54]

@@DFPercush Remember that when working with high performance, increased memory consumption can also cause contention on L2 and L3 caches, so you might see performance degradation that cannot be solved by adding more RAM. This can be worked around by carefully arranging the layout of data in memory, but if the game wasn't designed like that from the ground up, the effort involved approaches a complete rewrite. Additionally, there may be (I'm not sure because I haven't gotten that low-level looking into it) mechanics based on sequentialized operations which would then either need to be changed, or hamper parallelization efforts; I wouldn't make any claims about how relatively easy it should be until I've understood the source code, which in this case I don't have.

[@serith102]

Multithreading and parallel processing are not trivial and would most likely result in a decrease in performance. Also I'll just call BS right now on 3 to 8 times performance increase with 4 cores, it's extremely rare to even get a 1 to 1 increase from adding a core on anything more complex than simple programs a 2x increase in performance on the same algorithm is nonsensical.

[@tschichpich]

30min intruduction and then comes the conclution :)

[@Tapecutter59]

I've been playing since v16 beta, I never play with bitters or pollution these days, only touch uranium if I have to. I've built a couple of spaggettified city block megabases, Came here to get some oranisational tips, was not dissapointed.

[@AnonyMous-gt8vq]

Maybe you could mine directly into furnaces and then load trains with bots to be even more efficent. Plates have a larger stack size as well.

[@E-102_Gamma]

That would take up a lot of precious real estate on top of the ore patches, and would be much, much slower because the furnaces wouldn't even begin to keep up with the mining drills when Mining Productivity is that high.

[@Snoogen11]

I've always wondered how many logistic robots your megafactories have near the end of your game, I've never had more than 3000 at once, and even then, only 1k of those were actually doing work.

[@Mecrom]

disabled and removed biters and went from ~20 to ~55 UPS. Bbbut my defense setup was so cool though... What am I gonna do with my 30k laser turrets? ;;;;(

[@tistelnilsson]

Whitelist laser turrets in your destructor. But first blueprint it.

[@dfarmer_]

Reverse factory mod to get your resources back!

[@Pro-Bert]

@nilaus Does the command kill all the biters you see on the map at this moment or will the command kill ALL biters so that new map areas are free of them? Thanks for your great Videos! 🙂

[@roxterat]

Thank you for a good explanatory video!

[@sillymonkey725]

Do idling entities hurt UPS too? I have inserters and assemblers in mind. It's easy to have inserters that will work 2% of the time, sometimes less. Like at old defensive outposts that never get attacked anymore, for example.

[@goffe2282]

I'm running on an M1 laptop.... so it may be that.... I have a 2.7k SPM base that I built with friends. The timing sits around 13-14..... but I only have around 21UPS. I have no idea what's going on.

[@RUNETHERABB1T]

oh god is that danish accent recognisable. From one danish person to another!

[@Dan-vq4pz]

IS THAT MINING PRODUCTIVITY 72 RESEARCH 😮

[@thewhitefalcon8539]

FYI, it's "0.6 milliseconds", not "0.6 updates"

[@qwerty9091000]

when I press F5 I get a different debug menu

[@dra6o0n]

I wonder if there's a way to utilize factorio as a CPU benchmarking tool... LOL!

[@Nortonius_]

66.6K subs...... WooOoOOOooo 🎃👻. This is an extremely useful one! Thank you!

[@darkpist]

how about to create a solar panels array with a power production of a sun power (sun power = 3.9 x 10^26 == 6500000000000000000 solar panels (one solar panel = 60Kw) ) i would like to see it :)

[@VRGamercz]

Hello, just finished my first rocket, do you think megabase is possible on 9900k and 32GB of 3600MHz DDR4? Dont think that GPU matters but if it does its OC'd ROG 2080 super

[@sanjikaneki6226]

how did you stop fire at will on arty ?

[@slowfudgeballs9517]

So like when will a 5600x run out of performance?

[@ingmarins]

youre wrong about smoke. it used to be bad, but was bugfixed

[@theDemolisher13]

sadly most of this if not all of it doesn't help me with my factory. As the Goal is to make a self expanding factory with that can handle biters and rough terrain and expansion all by itself. which means a lot of these systems can be used as robots in order to be useful like you show need to be on their own network per production block but for a S.E.F. they all have to be on one network else complications accrue. Furthermore solar is great but it will never keep up with a power demand that grows exponentially over time. but they do help lesson the demand on the nuclear plants while steam engines just can't keep up period. with all that said good work here fellow engineer.

[@nilsw8076]

I think my i7 7700k @4.8 ghz is bottlenecked before i get to megabes lvl. I must have build really inefficient.

[@Rick203]

Who dares to interrupt my Sekiro watching :-)

[@chrisloach9658]

25:40 What's the problem with power switches?

[@TheAechBomb]

they join and seperate electrical networks, which needs to process every entity on each network every time they activate (I believe)

[@Fetch049]

no section markers in this master class video (or the last one). Good content, but it seems the numbers don't support the conjecture that nulcear is so bad. 0.1 updates is pretty manageable. Now, I recognize 0.001 updates for solar is far less than 0.1 for nuclear, but there are better places to try and optimize overall game performance than removing nuclear. Pareto principle says to optimize the biggest problem first, which is entities. Since there's no visibility into what entities are doing what, we can't really optimize them very much. :-(

[@okuno54]

The Pareto Principle doesn't consider the difficulty of optimizing each part. Replacing nuclear with solar is a very simple change, but optimizing a rail network might mean a complete overhaul of the topology of your base, assuming the new network you are planning would actually improve UPS.

[@TheJacklikesvideos]

you have to scale up nuclear with your power demands. the solar processing is flat. if he were running his whole base on nuclear, it would not run at 60ups.

[@thorH.]

Is there a mod that makes nuclear efficient?

[@TheAechBomb]

factorio nuclear is incredibly efficient as-is, even a couple centrifuges with basic processing will supply several reactors (I believe it's like 5 miners per reactor)

[@thorH.]

@@TheAechBomb I mean computationally efficient. My game is running at 15 fps already and Im using solar right now. Nuclear is more convenient tho, not such a hustle to set up.

[@EKbnc]

my question is is it possible to build a 1000k science per minute base, without ever using a belt (meaning, everything is bots / trains), and still have 60 UPS

[@ariedge2435]

5 people who dislike this must play on vista Edit: now it is 17

[@thorH.]

Me and my 600 trains 🤔

[@TiuJaKas]

help i got 640 trains FPS IS DYING

[@silentsixteen]

Bruh, just use a 3070 and 5600x /s

[@Nilaus]

Factorio is CPU and RAM speed constrained so it will make very little difference with a better fraphics card

[@JohnDoe-tn5qp]

Disabling biters and pollution is cheating. Change my mind.

[@Nilaus]

Biters do nothing but eat processor power in the late game. Between Artillery, Nukes, Spidertrons and Personal lasers there is no challenge of biters even with max Deathworld settings. If you want to build big you have to remove biters in the late game

[@cumykaze2435]

Now that's a long introduction from 0:00 to 30:16 :D

[@Bigfatmasta]

Thx

[@VladTheImploded]

I absolutely love the normal master class videos, showing the how and why to create super optimized production lines. I had no idea I would like this video as much as I did. Really, this was a super cool and informative video. I'm now planning on making my own megabase for the first time, but I want to wait until after the train changes in 1.1.

[@DFPercush]

Trees are the real enemy, but they are not the final boss.

[@SirHackaL0t.]

Turning off the clouds helps as well. Great video and most timely as I was looking this evening at what to do to get performance up when in the map view. :)

[@stefanbergmann1884]

Well ok, all entities reduce UPS. So whats about concrete tiles ? And whats best? Belts or underground belts on large scale?

[@leonardschulz7050]

I think, entities on their own only use RAM, entity UPDATES are what matters, right? Since concrete does not need to calculate anything, it should not matter (like solar panels don't hurt performance, since they dont update, they just exist.) Belts VS Underground Belts has been tested extensively (maybe look through reddit), the conclusion was basically that underground belts do not increase the performance significantly enough to bother replacing your normal belts.

[@TheJacklikesvideos]

@@leonardschulz7050 floor prevents ground absorption, so if you have pollution on, concrete hurts UPS, worse so if you also have biters.

[@Weeohhweeohh]

I love your factorio content. I have probably watched well over 100 hours of it. That being said, I have never played it or have the intention of playing it. Not sure what's wrong with me hahahahahahhahaha

[@retlom779]

Nothing wrong :) there are tons of people who watch soccer or other sports without playing it so why not watch anything else you enjoy watching :)

[@AncalagonGames]

with the new ryzen we can go bigger base :p

[@tigry3623]

You know you are fast when someone hasn't put up a summary as pinned comment :)

[@viqq183]

I still don't get it why I get 20fps max on bigger bases.

[@davidboleen1906]

So what mods kill the performance? A list would be cool.

[@fang_xianfu]

You called out the Bottleneck mod for its performance impact - it's just worth highlighting that the newer Bottleneck Lite mod provides the same functionality with zero performance impact. It bakes the icons into the sprite, and the sprite already knows the state because it uses that info to know whether to animate or not, etc.

[@kanarie93]

AMD Ryzen 7950 X3D and your ups will skyrocket due to the big L3 cache :D Factorio is not so much cpu bound but its so much memory bandwith usage due to the calculations being the same over and over again. L2/L3 cache realy helps then. Its not an AI system doing all kinds of random calculations like normal games then L3 doesnt have a huge effect.

[@iR0CKt]

So I just gotta rent. Just saying yeah an i7 is enough is a f***ing nogo! I7 what first gen? R u serious? No. Throwing around terms tf?

[@wintermath3173]

What exactly makes steam storage consume UPS? Is it when the amount in a tank changes? If so, would the UPS impact of a steam accumulator that consumes exactly 0 steam when it’s not needed be much lower?

[@Bastelkorb3]

The liquid system is never in a resting position like Belts. Even when your tanks are full the game calculates flow between each of them.

[@rocketman2304]

why electric newtwork is so high ? i have 0,5 no sense

[@meixo9083]

im finally here for a reason ...

[@loganheinrichs2418]

How do loaders compare to inserters as far as UPS are concerned?

[@bekar_au]

Depends on which loaders you use. Some loader mods are programed as inserters (that run quickly, without the animation).

[@Naruedyoh]

So... When do you start your ESA internship?

[@gregkramer5588]

Nicely done. My base is running about 3800 SPM update is down to 50.

[@aoeuable]

Wait if you disable pollution how do you get rid of trees?

[@edison_ef7727]

i was waiting for this kind of video

[@strex2062]

I think that mining using robots could be a good option.

[@Delta-pb3tk]

introduction, conclusion. nothing unneeded

[@derpyburgerz6018]

The F5 button shows a grid for me.

[@kensavage7657]

My megabase was getting slower and slower - I only get around 20/25 UPS now. Followed your directions - see that I have "Entity Update" at a consistent value of 45. Yup, guess that's where my problem is. I, like many people, never planned to have a megabase - so never did any optimisations, really. My base just kept expanding, needing another area to produce a bit more plastic, circuits, anything... Then I couldn't keep up with space science demand, so needed a 2nd rocket area, which needed more resources. And so the cycle continued, until I ended up with the sprawling base that I love! No way to get that Entity update value down, without massively rejigging my base. Guess it's time to decide whether I can face starting a new base, from scratch......

[@JohnDoe-tn5qp]

Or you can backup a copy of your current base for the future, and delete/rebuild as much as you want. This way you will keep all of the research you have done, you'll have a good supply of resources to build big with, and you can always go back and replay your old version if you want.

[@Felinaro]

How about a inserters clocking? Is it possible to save noticeable ups by adding circuit networks to clock inserters to do more efficient swings?

[@nickwilson3499]

Once an inserters gets a pulse it does a full swing no matter what. So it wouldn’t make any more efficient swings.

[@Felinaro]

@@nickwilson3499 1. only if it grabs from box, grabbing from belt is a tricky thing; 2. main point of clocking, as I understand, is to make less swings with more items, e.g. not "10 swings with 1 item each", but "1 swing with 10 items". Im not really familiar with suck trick, but I heard about it, and in doubt - is it really worth it?

[@nickwilson3499]

@@Felinaro oh yeah, with stack inserters that makes sense. however, in hyper UPS optimized bases you wouldn't usually have much down time for inserters.

[@ChiefFalque]

If you want UPS stability on servers, you need multiple servers and clusterio or whatever it's called (you can split the world or connect multiple worlds between multiple servers, with the ability to send stuff back and forth between them.

[@kyngs8540]

This is the reason why every time I play Factorio, I am running Dedicated Server on my VPS because then it runs way smoother, the only downside is that riding in a car is even more horrible than normal.

[@Marduk401]

why are you explaining what fps is to a factorio player. if someone does not grasp the meaning of fps he will be unable to play the game up to that point that he has performance issues in the first place.

[@Nilaus]

what an elitist attitude :( Why even explain anything; if you don't know it already you will be unable to play properly

[@Marduk401]

@@Nilaus come on dude factorio is elitist by itself, it's clearly meant for above-average intelligence. I'm not elitist because its already an elit, people who play these kinds of games are smart.

[@Keykey70]

@@Marduk401 lmfao

[@BigKevSexyMan]

I have a question. For loading my mining trains, I was thinking of downloading a logistic cargo wagon mod, then have my miners store everything directly in provider chests, then use bots to load the trains from said chests. While I get that it wouldn't be as low impact as the direct mining setup, would it still be an acceptable idea performance-wise?

[@Nilaus]

sorry, can't help as I have never tried it

[@bobbooty]

FYI raiguard's Bottleneck Lite has no UPS penalty.