Page 1 of 1

UPS / FPS helper - Power switches

Posted: Fri Nov 10, 2017 7:02 pm
by dinodod
Just wanted to know if I were to wire a factory to be smart and power down idle / unneeded factories, would that help improve UPS / FPS at all or would the power outage cause more issues on the server? I am trying to work on a SP map atm where I am using power switches to control which parts of my base gets turned on at different times using various logic builds.

I just wish there was a cleaner way to manage all the belts as well. That would be sweet.

Re: UPS / FPS helper - Power switches

Posted: Fri Nov 10, 2017 10:03 pm
by Jap2.0
Probably, though I doubt it would be much (if any) different than if that part of your factory wasn't doing any production (for example if it was out of resources or something), and I doubt it would completely remove the performance overhead. In short, I have no idea :).

Re: UPS / FPS helper - Power switches

Posted: Sat Nov 11, 2017 8:53 am
by Optera
I once got a dev reply stating powered down inserters stuck mid move take more computing than powered inserters idling (sleeping).
So you may want to make sure to add an off delay timer allowing all inserters to run out of items to move before powering them off.

Re: UPS / FPS helper - Power switches

Posted: Sat Nov 11, 2017 9:53 am
by DaveMcW
Fully powered entities always have better UPS.

You can use power switches to reduce power consumption and pollution, but don't do it for the UPS.

Re: UPS / FPS helper - Power switches

Posted: Sun Nov 12, 2017 12:16 am
by leitk
Do unpowered (0%) entities harm UPS?

Re: UPS / FPS helper - Power switches

Posted: Sat Nov 18, 2017 1:08 pm
by Mimos
My experience with parts of the factory that are in a standstill for reasons that are complytely not my fault (bold lie) the UPS go up a lot. I have never tried if powering them down safes additional ups or not. At least the slight additional reduction in power consumption could maybe reduce burner-based power production and thus increase UPS a tiny bit further.

Re: UPS / FPS helper - Power switches

Posted: Sat Nov 18, 2017 1:23 pm
by Zavian
Anytime you have large sections of your factory inactive, ups will go up (assuming it was below 60 to start with). I don't think it matters whether you used a power switch, or stopped the input or output. Totally inactive entities (because they are input starved or their output is blocked, or they are without power) use less cpu time than active entities.

Re: UPS / FPS helper - Power switches

Posted: Sat Nov 18, 2017 3:25 pm
by Bilka
Zavian wrote:Anytime you have large sections of your factory inactive, ups will go up (assuming it was below 60 to start with). I don't think it matters whether you used a power switch, or stopped the input or output. Totally inactive entities (because they are input starved or their output is blocked, or they are without power) use less cpu time than active entities.
It matters. If you unpower an entity it has to check each tick whether it receives energy again, so it can't go inactive.

Re: UPS / FPS helper - Power switches

Posted: Mon Nov 20, 2017 10:28 am
by dinodod
Optera wrote:I once got a dev reply stating powered down inserters stuck mid move take more computing than powered inserters idling (sleeping).
So you may want to make sure to add an off delay timer allowing all inserters to run out of items to move before powering them off.
Thanks for the tip... That makes redesigning the factories a challenge now... Mmmmm....

Re: UPS / FPS helper - Power switches

Posted: Mon Nov 20, 2017 10:34 am
by dinodod
DaveMcW wrote:Fully powered entities always have better UPS.

You can use power switches to reduce power consumption and pollution, but don't do it for the UPS.
Can you explain why not? The idea is to make a mega base and have each factory work in serial instead of in parallel. Thus instead of 50 factories working at once, 1 would be active and 49 would be idle until triggered by a logic condition.

I would think that as the factories are not powered, they don't have any requests for UPS except for checking to see if power is applied or not. I was thinking of applying this logic to the belts as well so they would be disabled and thus lower UPS demand. It would be circuit logic hell but belts not moving should in my theory help a little with UPS as well???

Re: UPS / FPS helper - Power switches

Posted: Mon Nov 20, 2017 10:38 am
by dinodod
Zavian wrote:Anytime you have large sections of your factory inactive, ups will go up (assuming it was below 60 to start with). I don't think it matters whether you used a power switch, or stopped the input or output. Totally inactive entities (because they are input starved or their output is blocked, or they are without power) use less cpu time than active entities.
That's what I thought as well, but wouldn't a powered inserter / factory still perform more UPS to check for validation of an item to grab / insert / etc whereas a unpowered inserter/factory only checks for power conditions? In a mega base, this leads to thousands of updates that are not needed and thus UPS is saved.

Re: UPS / FPS helper - Power switches

Posted: Mon Nov 20, 2017 1:16 pm
by mrvn
dinodod wrote:
Zavian wrote:Anytime you have large sections of your factory inactive, ups will go up (assuming it was below 60 to start with). I don't think it matters whether you used a power switch, or stopped the input or output. Totally inactive entities (because they are input starved or their output is blocked, or they are without power) use less cpu time than active entities.
That's what I thought as well, but wouldn't a powered inserter / factory still perform more UPS to check for validation of an item to grab / insert / etc whereas a unpowered inserter/factory only checks for power conditions? In a mega base, this leads to thousands of updates that are not needed and thus UPS is saved.
An assembler has a waiting queue for inserters. The inserter gets registered there and goes to sleep when the assembler is blocked. So the inserter doesn't have to check each tick, it gets told when to wake up again.

I'm not sure about belts and waiting queues. Does an assembler waiting on an empty belt check the belt every tick? Or an assembler with stuff in its hand and a full belt?

I'm pretty sure inserters with circuit conditions get an event every time the signals change but can sleep in between. So you might want to turn the inserters off by circuit to save UPS. But better test that.