Game FPS at End game

[bigyihsuan]

I've found that endgame FPS loss seems to be from revealing the map too much. As the amount of scanned area goes up, UPS goes down...

The amount of scanned area has no impact on game performance as virtually all of the scanned area is not active because you're not near it or pollution is not near it.

Might be because I have tons of pollution and tons of entities in the factory. I'll edit this post later to see if revealing a huge area actually affects fps and ups.

[wahming]

The amount of scanned area has no impact on game performance as virtually all of the scanned area is not active because you're not near it or pollution is not near it.

Does that mean that polluted chunks affect game performance?

[searker]

The amount of scanned area has no impact on game performance as virtually all of the scanned area is not active because you're not near it or pollution is not near it.

Does that mean that polluted chunks affect game performance?

Yes. Thats the case because polluted chunks stay active - At least to update the pollution-value if absolutely nothing else is happening in that chunk.
There is a debug-option to visualize all active chunks.

[MrGrim]

I think wee need some techs and entities that allow us reduce huge amount of belts/bots/inserters etc.
Like longer underground belts, higher stack bonus for bots and inserters, loader/unloader/hopper. All thinks that allow us reduce performance impact in big factories.
Rapid inserter in v0.13 is like that and will allow us to reduce amount of inserters used especially for buffed assemblers. Now I have to use 12 inserters to catch up assembler speed and if I could reduce it to 2-3 inserters for 1000 asseblers it would be huge performance benefit. Same with Bots because for buffed assembles we need thousands of them. I have bob logistic mod that add faster bots and still for 200 smelters with 10 speed I need 500 bots. I would need 2000 or more vanilla bots and it's only for smeltrs.

Ya, but that just kind of sucks all the fun out of it. Replacing the fun of working within and around limitations with super powered entities would just turn the game into a grind. That's why I don't play with Bob's full set of mods, and what I do play with I've heavily modified (e.g. I love the electronics chain, but I modified it to fit more with a vanilla assembling speed/module/tiering pace).

I do wish the bots were more intelligent, though. Every time I see one skip 3 roboports only to decide 1 tile away from its objective that, oh ya, it does actually need a charge or a group of 500 bots hovering around a single roboport waiting to charge when there are literally 5 other roboports with no line sitting right bloody next to the one they're waiting on... idiot bots..

[Ghoulish]

I think wee need some techs and entities that allow us reduce huge amount of belts/bots/inserters etc.
Like longer underground belts, higher stack bonus for bots and inserters, loader/unloader/hopper. All thinks that allow us reduce performance impact in big factories.
Rapid inserter in v0.13 is like that and will allow us to reduce amount of inserters used especially for buffed assemblers. Now I have to use 12 inserters to catch up assembler speed and if I could reduce it to 2-3 inserters for 1000 asseblers it would be huge performance benefit. Same with Bots because for buffed assembles we need thousands of them. I have bob logistic mod that add faster bots and still for 200 smelters with 10 speed I need 500 bots. I would need 2000 or more vanilla bots and it's only for smeltrs.

Ya, but that just kind of sucks all the fun out of it. Replacing the fun of working within and around limitations with super powered entities would just turn the game into a grind. That's why I don't play with Bob's full set of mods, and what I do play with I've heavily modified (e.g. I love the electronics chain, but I modified it to fit more with a vanilla assembling speed/module/tiering pace).

I do wish the bots were more intelligent, though. Every time I see one skip 3 roboports only to decide 1 tile away from its objective that, oh ya, it does actually need a charge or a group of 500 bots hovering around a single roboport waiting to charge when there are literally 5 other roboports with no line sitting right bloody next to the one they're waiting on... idiot bots..

Indeedy!

[Neotix]

Ya, but that just kind of sucks all the fun out of it. Replacing the fun of working within and around limitations with super powered entities would just turn the game into a grind.

Then why we have faster inserters, faster belts, faster assemblers, why we have speed modules? Why we have entities that move that limitations? If we move limits for one entities we should move limits for other entities or have alternative solution that will catch up. Stacks bonus, bots speed are the easiest solutions, not the best ones. Different inserters with unique abilities (like double stack bonus for direct insertion), packing items etc. could be better solution but in need much more work.


About bots and recharging. I agree that it need improvement.

[MrGrim]

Then why we have faster inserters, faster belts, faster assemblers, why we have speed modules? Why we have entities that move that limitations? If we move limits for one entities we should move limits for other entities or have alternative solution that will catch up. Stacks bonus, bots speed are the easiest solutions, not the best ones. Different inserters with unique abilities (like double stack bonus for direct insertion), packing items etc. could be better solution but in need much more work.


About bots and recharging. I agree that it need improvement.

You're not wrong. I just laugh every time I see a fully upgraded miner with god speed modules on a Bob's Mod base because it's so hysterically overdone. I think 3 tiers is a pretty good compromise before teching up just means "more of the same, but faster". i don't want to move 10k ore a minute with one super belt and a couple of super inserters.

Basically, at some point just making higher tiers of the same items gets tired. Different mechanics that change how things are done without taking away the design and optimization challenges of the game that result in more throughput for fewer compute cycles would be cool, though. You can even add more depth with how different mechanics interoperate.

[Neotix]

I'm not considering mods because they brings a lot of OP things and devs can't control it. It's player choice to install mod or not (witch all advantages and problems). I'm speaking about pure vanilla game where we have modules and beacon and we can buff machines to the extend that inserters and bots barely catch up and we need tones of them. That way big factories cause huge fps drop because of amount of belts, inserters, bots, polluted chunks etc.

We can avoid long belts by using train and this is great alternative (but we still need tones of inserters in loading/unloading stations). Bots are alternative for belts in small scale but they're so slow and carry so little that we need thousands of them to move items in buffed factory and for that thousands of bots we need hundreds of roboports because we don't have dedicated charge pads (they're in bob mod by we're still speaking about vanilla game).

I think that good alternative for belts and bots could be hybrid like bot on monorail (act like bot but move only on rail) that can carry entire stack and don't need recharge because it is powered by rail. It could move items between logistic chest (chest have to be placed near rails). In that way we get smart delivery system but we still have to manage it, place rails and chest, think how to avoid jams and utilize that feature in the best way.

[MrGrim]

I'm not considering mods because they brings a lot of OP things and devs can't control it. It's player choice to install mod or not (witch all advantages and problems). I'm speaking about pure vanilla game where we have modules and beacon and we can buff machines to the extend that inserters and bots barely catch up and we need tones of them. That way big factories cause huge fps drop because of amount of belts, inserters, bots, polluted chunks etc.

We can avoid long belts by using train and this is great alternative (but we still need tones of inserters in loading/unloading stations). Bots are alternative for belts in small scale but they're so slow and carry so little that we need thousands of them to move items in buffed factory and for that thousands of bots we need hundreds of roboports because we don't have dedicated charge pads (they're in bob mod by we're still speaking about vanilla game).

I think that good alternative for belts and bots could be hybrid like bot on monorail (act like bot but move only on rail) that can carry entire stack and don't need recharge because it is powered by rail. It could move items between logistic chest (chest have to be placed near rails). In that way we get smart delivery system but we still have to manage it, place rails and chest, think how to avoid jams and utilize that feature in the best way.

I was just using mods as an example of the same things but faster concept taken to an extreme to illustrate that "faster belts, inserters, and assemblers" can only go so far before it becomes repetitive and a little goofy. This mono rail idea is interesting. You're thinking like a miniature train network type thing for intra-factory distribution? Moves items more efficiently and more quickly than belts but more challenging to optimize and set up? I think I like it.

[GeniusIsme]

It is quite trivial to isolate every factory in separate threads (plus maybe the handling of robots in general), as long as the elements in the factory are not connected by belts. Robots do not count - updating the count in something like a (special) chest thread safe is quite trivial (atomic thread safe additions/substractions are a processor level feature).

Isolating factory by connectivity is neither trivial nor needed. Factories are dynamic and developers will need to keep track of isolation status, and what is much more difficult - redistribute tasks between treads accordingly.
You are right about compromises, though. And straightforward one is to consume more memory in order to decouple tasks. Basically, you keep clone of the game state, and update one copy while examining another. This allows to split update into any number of tasks without the need of synchronization (by the way, atomic counters being processor feature doesn't mean they are fast). One more benefit of this approach - no weird effects of updated being in particular order - when you get different results depending on how your factory is organized inside chunk.

[Swadius]

I think wee need some techs and entities that allow us reduce huge amount of belts/bots/inserters etc.
Like longer underground belts, higher stack bonus for bots and inserters, loader/unloader/hopper. All thinks that allow us reduce performance impact in big factories.
Rapid inserter in v0.13 is like that and will allow us to reduce amount of inserters used especially for buffed assemblers. Now I have to use 12 inserters to catch up assembler speed and if I could reduce it to 2-3 inserters for 1000 asseblers it would be huge performance benefit. Same with Bots because for buffed assembles we need thousands of them. I have bob logistic mod that add faster bots and still for 200 smelters with 10 speed I need 500 bots. I would need 2000 or more vanilla bots and it's only for smeltrs.

I made a mod for this purpose. It is the PipeEverything mod in WIP. As far as I have experienced, it does help with fps, although I would like more people testing it out to make sure. Not many people trying it out though since the mod requires building the entire factory over again new game or not.

It essentially converts all intermediates into pipable liquids so belts are essentially few and far between.

[Keeper]

Hello forum people! First post.
In my latest base, and from the outset it was designed to be ridiculously sized, I had some numbers in my head what i wanted to see on the production chart so objectives set and away I went.

About 400 hours later my base is 4 times bigger than what i ever planned, I've had to move entire portions of my base elsewhere and do train links. I have 85 active trains at the moment, all 1-8-1's running on mostly coal, all on one rail network with 2x2 rail lines. Couple of queues here and there lmao. Tens of thousands of inserters, furnaces, assemblers, belts, refineries etc. a massive bot area with 4500 log bots, and a measily 2500 con bots. all sharing 750 charging ports.
There can be millions of anything all being made at once if I kick the whole place into gear.
I have learned here, parts of my base are "buffed" as it needs bots to help the belts bring the stuff quick enough to keep the assembler/machine on full time.

What I noticed is when I first started the game it was at 60FPS. And thats what inspired my objectives - at what point will it become unplayable? We need to push it. I do this with most games, all building games anyway. its my fun.

e.g. I did it with openttd on my old computer. I was grinding my shoddy old pc to a halt I had thousands of vehicles. When i built my new pc, copied my save over and had it run without an hickup and a bazzilion fps then you know your new PC is good Makes you feel goooood with direct real comparisons

I didn't really realise it when I was building my factory but my FPS dropped eventually to what it is now- 19fps. Okay it could be worse, and i'm sure i've seen it drop to 10fps at times. But getting around was becoming tedious, and with expansions geting ever larger if something b0rked or i end up waiting for some materials then it could take unreasonable time.

Last week one evening I fell asleep on factorio *probably waiting for slow fps lag* then the next day after work I got home back to the precious, then I realised factorio was still running and the base had caught up. AND IT WAS 55FPS!!! There was no movement on the belts, all the buffers caught up, all the furnaces are full, and the trains full, all belts were full and backlogged of items etc etc etc.

I shoot out a couple of storage chests, and that kicks the whole factory into gear, then I can see max production numbers and slowly slowly as more things turn on the fps gets lower. I dont like it. So I began to investigate why the FPS was reducing.

Like alot of you I first checked task manager, and performance monitor, and noticed nothing out of the ordinary - factorio was running, no more than 20 - 25%(THIS IS OBVIOUSLY THE PROBLEM LOL), Factorio is running 32 threads, it dosn't appear to use more than 1 or 2 cores(6 OR 7 POTENTIAL PROBLEMS HERE), I have AMD 8 core, so I dont know if 2 cores are 1 etc so theres that(PROBLEM). Theres plently of memory free(KIND OF OPOSITE OF PROBLEM BUT COULD CALL IT A PROBLEM). And nothing is having a hard time. (EXCEPT FACTORIO < BIGGEST PROBLEM)

From this post slight improvements can be found by:
1. making belts undergound
2. apparently using more log bots?
3. using more trains.


But these are things within the game, I dont like underground belts because then you cant watch your things going from here to there. I'm Already using bots, and trains. I considered cutting all the log bots, as I thought their pathfinding was probably adding to the fps drop. Instead I was adding more chargers, so they would do their work quicker. But apparently cutting bots wont help much no?
I noticed when I expanded (meaning i deleted a huge part of my base and rebuilt it elsewhere) using trains my fps went up a bit, not sure due to using less belts or less items on belts, or more stack bonuses on the inserters not sure.

I considered the inserters as a culpret as every single one of those probably has a number of jobs:
1. check every item passing underneath it (faster belts mean more items and consiquently more checking) or chest is worse? < i bet thats the problem
2. check to see if it can move said items into the machine behind it at that particular moment
3. move to grab something
4. swing badda badda swing! and back again.
a
I'm tired thinking about it, I'm just saying even if the factory has the belts full of items FPS seems to pick up so long as they arn't moving. Is it the hitboxes? Is it the machines? Is it the inserters? I wish i knew.
But clearly the game needs to use more of my cpu. If theres one thing I know, is that factorio would be much happier if it used ALL of my CPUs. MAKE FACTORIO USE ALL OF MY CPU's DAMMIT I SHOULD BE ABLE TO USE CPUS FROM OTHER COMPUTERS TO FACTOR IN THE REQUIRED POWER TO POWER MY FACTORIO. HELL i WAS GOING TO RIG UP A DIESEL ENGINE TO THE PC SEE IF THAT ROCKS IN SOME POWA!

Sorry for essay, AND RANT mostly useless mumbo anyway GG
BEST GAME EVER

[NoriSilverrage]

Slightly relevant. Are people seeing performance increases on large bases with 0.13? I didn't see anything specifically related to performance, but the Devs made so many changes that I'm wondering if performance was potentially increased.

[SpeedDaemon]

MAKE FACTORIO USE ALL OF MY CPU's DAMMIT I SHOULD BE ABLE TO USE CPUS FROM OTHER COMPUTERS TO FACTOR IN THE REQUIRED POWER TO POWER MY FACTORIO. HELL i WAS GOING TO RIG UP A DIESEL ENGINE TO THE PC SEE IF THAT ROCKS IN SOME POWA!

I'll just leave this here...

[Malryn]

Wow this Post really took off. I thought it was going to die considering the lack of etiquette at the start

I want to point out that the post I made with regards to none of my CPUs were at 100% and that I had to be at resource limit which was why my game was slow and that Resource Monitor nor task manager could see Factorio using up a core to 100% because the OS was jumping the application between the cores too quickly. Well I have never seen that nor has anyone I spoke to has seen it either so if anyone has any recommendation on how to capture that data it can be dismissed. I am going to make the argument Factorio is still not using 100% of a CPU while FPS is dropping down to 6 FPS. If anyone can show how their computer is acting with a massive factory it would be appreciated!

Agree with Keeper's Essay except for the Diesel generators. Would be interesting to see a Factorio cluster to run a single game but I would just like to see Factorio use the resources that are already given to it -.-

[taiiat]

i'll just reiterate that 'not using 100% of my CPU' is not how computers work.

if it's bogging down, that means it's reached the limits of your Processor since Factorio is entirely CPU bound on not potato computers.
if you want more information i already explained it earlier.

[TheSkiGeek]

Wow this Post really took off. I thought it was going to die considering the lack of etiquette at the start

I want to point out that the post I made with regards to none of my CPUs were at 100% and that I had to be at resource limit which was why my game was slow and that Resource Monitor nor task manager could see Factorio using up a core to 100% because the OS was jumping the application between the cores too quickly. Well I have never seen that nor has anyone I spoke to has seen it either so if anyone has any recommendation on how to capture that data it can be dismissed. I am going to make the argument Factorio is still not using 100% of a CPU while FPS is dropping down to 6 FPS. If anyone can show how their computer is acting with a massive factory it would be appreciated!

Agree with Keeper's Essay except for the Diesel generators. Would be interesting to see a Factorio cluster to run a single game but I would just like to see Factorio use the resources that are already given to it -.-

Assuming you're on Windows, if you have X cores, a single threaded app will appear to max out at 1/X of "100%" CPU. So if you have 4 cores, a process pushing one core to 100% will be using "25%" of your CPU resources.

You should also be able to right-click the Factorio process and set its CPU affinity so it can only run on one CPU, and then you'll see that one CPU's utilization should go to 100% if it's CPU-bound.