Speed Modules, What are they good for?

[stm]

I have a question about the use of speed modules (All sizes):
Under the assumption, that we can neglect setup costs (That is the cost of infrastructure etc.), why would I want to use speed modules anywhere except at oil-wells?
Isn't it just better to just build more assemblers etc. instead? This even has the advantage, that you won't run into problems with insertes, and space is cheap, while energy -- though not exactly problematic -- does cost (except if you want to go solar, but then that takes the whole fun out of it anyway).
The only item, where this would not be possible are pump-jacks, which are limmited in number and thus a combination of efficiency and speed modules wins at some point.
But anywhere else (Not talking about ingredience in recipies)?
Stm

[MeduSalem]

Speed Modules are the only Modules worth using in Beacons:

viewtopic.php?f=5&t=53478

The thread already lists various benefits of using them.

[Kyralessa]

You could ask the same about computer processors: Why bother increasing the speed, if you can just add multiple processor or multiple cores?

The answer is that parallelism is more complex than just ratcheting up the speed.

With enough speed modules in an assembly machine 3, as well as a lot of surrounding beacons with their own speed modules, you can have one assembly machine cranking out items at a rate that would otherwise take many machines and quite a lot of real-estate once you factor in belts, inserters, etc.

If you're trying to keep your base reasonably compact, it's a good option.

If you're putting trains everywhere and delivering with them, then you may have tons of real estate such that the parallelism doesn't bother you, and the speed modules may not be that important.

[stm]

Speed Modules are the only Modules worth using in Beacons:

viewtopic.php?f=5&t=53478

The thread already lists various benefits of using them.

Not really
It is obvious, that the speed beacons improve the speed of the assemblers. But you can also do that by building more assemblers. The setup costs might be higher, granted, but I explicitly want to take those out of the equation, since at the point that would be relevant, one propably has enough resources anyway.

You could ask the same about computer processors: Why bother increasing the speed, if you can just add multiple processor or multiple cores?
The answer is that parallelism is more complex than just ratcheting up the speed.

Well, that is a question which is asked quite often and any programmer worth his code knows, that some things can be parallelized betzter than others. There are Problems (like sorting) which do benefit from a Processor which is faster, but are slower even when using 10 ore 100 cores instead. It always depends.

In the context of factorio this does not hold true in my opinion, since two assemblers will produce at least twice as much as one if supplied with enough raw material, and it is actually easier to supply two than it is to supply one with the same amount.

With enough speed modules in an assembly machine 3, as well as a lot of surrounding beacons with their own speed modules, you can have one assembly machine cranking out items at a rate that would otherwise take many machines and quite a lot of real-estate once you factor in belts, inserters, etc.

If you're trying to keep your base reasonably compact, it's a good option.

If you're putting trains everywhere and delivering with them, then you may have tons of real estate such that the parallelism doesn't bother you, and the speed modules may not be that important.

On the other hand: with tons of speed modules you need "megatons" of electric power. If that does not bother you...
Even without trains I do not really see a problem, since doubling your speed via modules not only doubles your energy (and pollution) but has a factor of 2.4 (20% more).
If you use beacons this goes up even more (and also increases footprint about the same a if you would using an additional assembler (per beacon)

[MeduSalem]

Not really
It is obvious, that the speed beacons improve the speed of the assemblers. But you can also do that by building more assemblers. The setup costs might be higher, granted, but I explicitly want to take those out of the equation, since at the point that would be relevant, one propably has enough resources anyway.

The space required for more machines is nasty but more about that further below.

UPS will also be a problem the more machines/infrastructure there are and the more chunks have to be active.

On the other hand: with tons of speed modules you need "megatons" of electric power. If that does not bother you...

Electricity is cheap ever since Nuclear Power. A multi-Gigawatt power plant easily fits within 8-10 Roboport areas.

If you use beacons this goes up even more (and also increases footprint about the same a if you would using an additional assembler (per beacon)

Well good with Beacons you need twice the area per assembler due to the beacon per assembler but you also get a speed up by 340% and 40% extra items (if you use productivity modules which are the only ones worth using in assemblers imho)... and the 40% extra items also means you need 40% less resources and infrastructure in the previous crafting stage... another saving that stacks multiplicatively over the crafting stages.

So the speed beacon setup ends up being much more compact than just using more assemblers. Speed Beacons basically improve the "produced items per area"-ratio a lot.

A speed beacon setup for assemblers usually requires only 1/3 (or less considering the crafting cascade) of the space of a setup without using beacons.

[Zavian]

Once you start using productivity modules, then a setup with max productivity modules and speed beacons can actually use less power per item produced then the same setup with only productivity modules. Once you factor in the reduced raw material cost and hence the reduced amount of power need for making the raw materials, then the productivity modules + speed beacons can also be more power efficient (per item produced) than straight assemblers without beacons.

One last use case is for megabases. Eventually you build large enough that adding more assemblers is adding more work for your cpu to do each tick and reducing your ups. Adding speed beacons is more ups efficient than adding more assemblers.

[leitk]

I think you're asking what is the point of Productivity module 3's (PM3s) in assemblers and speed module 3's (SM3s) in beacons vs just making more assemblers with PM3s.

The benefits of beacons are fewer modules total needed, less power (while running), less pollution, and less space required. This also means a lot less other infrastructure.

I see that you're OK with ignoring the cost, and I agree that power, pollution, and space are (mostly) irrelevant.

The real benefit is if the cost is lower, then you can build more sooner.
Since modules are not trivial in cost, saving 20 means you can set up another line of production much faster.

You can get the same results from lots of assemblers, or use beacons and save a ton on your build cost and time.

[impetus maximus]

speed modules in beacons can let you save electricity (and UPS) by using fewer assembly machines.

[Koub]

An assembling machine does not use power per item produced, but per time spent crafting.
If you speed-up the assembling machine, you mechanically reduce the power consumed per item (this does not account for the augmentation of consumption that comes with the modules). Usually, speed modules are considered optimal in beacons, around machines equipped with productivity modules (that lower crafting speed significantly, thus de-multiplying the consumption per item).

[stm]

speed modules in beacons can let you save electricity (and UPS) by using fewer assembly machines.

And how is that possible?
A Speed Module increses the powerconsumption more than the speed, so it should be more efficient to use several assemblers than using speed modules?!
And this does not even take into account the power drain of the beacons itself.
Stm

[impetus maximus]

And how is that possible?
A Speed Module increses the powerconsumption more than the speed, so it should be more efficient to use several assemblers than using speed modules?!
And this does not even take into account the power drain of the beacons itself.
Stm

assembly machines with productivity modules. i've tested the use of more assembly machines with lvl 1 efficiency modules, vs speed beacons, and productivity modules in less assemblers.
the speed becon/prod modules assemblers use less energy.

[leitk]

speed modules in beacons can let you save electricity (and UPS) by using fewer assembly machines.

And how is that possible?

If you compare for a fixed total output (and input) to the assemblers, speed modules in beacons are cheaper in terms of build cost, power (while running), pollution, and size.

This is all assuming you are using productivity module 3s in the assemblers.

You are correct that if you take, for example, 4 assemblers and tack on 14 beacons your power use goes up, probably by a factor of 4 or so.
But your productivity per second also goes up by a factor of 11 (assuming you can keep them fed, not a trivial problem).
In this example you need to compare the costs of 4 assemblers and 14 beacons vs 44! assemblers without beacons, because they have the same output per second.
You could also compare them by removing 3 of the assemblers and adding 4 beacons, so you have 1 assembler and 4 beacons, it's still cheaper than 4 assemblers and is in fact more productive per second.

[Cribbit]

Speed beacons are crucial when using productivity modules.

If you're not using productivity modules the cost of speed modules/beacons is ludicrous compared to just building more assemblers. (with rare cases for perfecting a ratio, eg recipe needs 3.2 assemblers so you have 3 asm3's, one with a spd 1, which is cheaper than a 4th asm3)

Speed bonuses are additive not multiplicative, so an asm3 with 4 prod3's is 11 times faster when you have 8 speed beacons around it. Those 8 beacons can cover an additional 7 asm3's, so you're only really paying for 1 beacon per asm3. The modules are the vast majority of the cost, so you're getting 11x production for 1.5x cost.

Prod3 modules are crucial for stretching your materials as far as possible. While the initial setup cost is high it is a one time cost and it can quickly pay for itself.

[Frightning]

I have a question about the use of speed modules (All sizes):
Under the assumption, that we can neglect setup costs (That is the cost of infrastructure etc.), why would I want to use speed modules anywhere except at oil-wells?
Isn't it just better to just build more assemblers etc. instead? This even has the advantage, that you won't run into problems with insertes, and space is cheap, while energy -- though not exactly problematic -- does cost (except if you want to go solar, but then that takes the whole fun out of it anyway).
The only item, where this would not be possible are pump-jacks, which are limmited in number and thus a combination of efficiency and speed modules wins at some point.
But anywhere else (Not talking about ingredience in recipies)?
Stm

Setup costs aren't neglectable if you're using higher tier modules (tier 1 are pretty cheap, but tier 2 are costly, and tier 3 are fiendishly expensive). The obvious use cases for Speed modules are Pumpjacks (because productivity is pointless on an inexhaustible resource, and speed gives the fastest income rate per oil field) (There is an exception, a 12-beacon surrounded Pumpjack actually gives a faster rate if the Pumpjack itself has Prod3 rather than Spd3, assuming all 12 Beacons have Spd3 as well). The other obvious use case is in Beacons particularly because the Speed boost is additive against the Speed penalty of Productivity modules which is a much better interaction than if it were multiplicative. Besides these two obvious use cases here are some other things that make Speed modules good: Faster machines=more compact setups, interaction with Efficiency modules can lead to fast, and very energy efficient setups (if energy conservation is a thing you enjoy, it's not like producing energy is hard in this game though). As an example of the latter (and of an actual use for Eff3 modules), you can use Assem3s with 3xEff3 and 1xSpd3 to get 150% speed at 20% energy consumption, which means your energy cost per item produced is just 13.(3)% of the usual cost from an Assem3 with no modules.

[stm]

particularly because the Speed boost is additive against the Speed penalty of Productivity modules which is a much better interaction than if it were multiplicative.

Thanks, that is actually the answer I was looking for. Once again Factorio "math" is what makes something counter intuitive work out.
I forgot to take into account the way the different modules interact with each other. They should really change the text from "+X%" to "+X" or "+X% of base".
Stm

[EpicFail1403]

A speed module increase an assembler's working speed by 50%, while increasing it's power usage by 70%.
It seems not worthy.
But
but anything before 'but' is bull sh*t
When used with productivity modules:
An assembler with 4 productivity module 3, is working at 100%-60%=40% speed and 100%+320%=420% power usage.
BECAUSE MODULE EFFECTS ON ASS MACHINES ARE ADDITIVE,
When affected by a beacon with 2 speed module 3, such previously mentioned assembler receives +50% speed bonus which is 125% faster than 40%, while power usage increased by 70%, which is only 18% more compared to the previous 420%.
Compared to an assembler with 4 productivity module 3 with no beacons, each beacon added can actually reduce energy cost per product AND boost production speed.
Yes, power usage is drastically increased, but you get 40% more product.

[Patashu]

There are also a few non-vanilla reasons why you might want to use a speed module:

1) A mod, like Building Platforms, restricts how easily you can scale up industry.
2) The production building is more expensive than filling it with speed modules (not sure what mods this happens in, but it COULD happen)

[BlakeMW]

A speed module increase an assembler's working speed by 50%, while increasing it's power usage by 70%
It seems not worthy.
But
...

Even putting aside synergy with productivity modules, taken by itself a speed bonus means that each item requires less energy to produce because it requires less machine time. A pure +100% speed bonus would halve energy per item produced. Speed modules have a higher consumption penalty than speed bonus (except in Bob's OP Modules where you can get pure speed modules) so they do increase energy consumption per item, but not nearly as much as it appears at first glance:

Code: Select all

1x Speed1: 1.5 / 1.2 = +25% per item
2x Speed1: 2.0 / 1.4 = +42% per item

1x Speed3: 1.7 / 1.5 = +13% per item
4x Speed3: 3.8 / 3.0 = +27% per item

So for instance while 4x Speed3 Modules lists a +280% energy consumption increase, in fact it's only +27% energy per item produced thanks to each item being produced in much less time, the assembler running triple-time is only using 27% more energy than 3 Assemblers without modules.

Speed modules are still best used for their synergy with productivity modules (putting aside this synergy, building more assemblers tends to be better than +speed), but chucking a bunch of speed modules into an assembler does not result in dramatic energy use increases. Speed beacons when not boosting productivity assemblers do carry more of an energy consumption hit thanks to the 480kW drain - although the direct energy use of the assembler tends to dominate the energy consumption: a beaconed assembler can draw around 2MW, so in alternating rows the beacons only end up accounting for 20% of the power draw.

[Bauer]

I get the point re speed modules.
I have a problem with efficency modules. I would love to use them, but I don't see the point.

Let me compare 2 typical scenarios (all modules tier 3 for an infinitly long game)

1. Tier 3 assembly machine with 4 prod. modules + 8 beacons with speed
Production: 1.25 x 1.4 x 4.4 = 7.7 items/sec
Energy consumption: 210 kW x (1 + 4x0.8 + 8x0.7) + 480 kW = 2538 kW
(the 480 kW for an average of 1 beacon per assembly machine)
--> Total energy cost is 329.6 kW/item

2. Tier 3 assembly machine with 4 prod modules + 8 beacons with 50/50 speed and efficiency modules
Production: 1.25 x 1.4 x 2.4 = 4.2 items/sec
Energy consumption: 210 kW x (1 + 4x0.8 + 4x0.7 - 4x0.5) + 480 kW = 1530 kW
--> Total energy cost is 364.3 kW/item

==> 54.5% output at 110.5% energy cost

Ok, in both scenarios energy cost is higher than the unmodified T3 assembly machine (168 kW/item). I'm ok with that.
But should the energy cost be higher when using efficiency modules?????
Is my math right? BTW, it gets even worst if you have more than 1 beacon per assembly machine on average.

Is there a scenario in which those efficiency modules make sense?
Especially if you turn off pollution!?

[Bauer]

If you do NOT use productivity modules, efficiency modules might be an advantage.

1. T3 assembly machine with 4 speed modules, 8 beacons with only speed modules
Producion: 8.75 items/s
Energy consumption: 2454 kW/s
--> 280.5 kW/item

2. T3 assembly machine with 4 speed modules, 8 beacons with 50/50 speed and efficiency modules
Production: 6.25 items/s
Energy consumption: 1446 kW/s
--> 231.4 kW/item

==> 71% output at 83% energy cost
(I have difficulties to get excited)


3. T3 assemby machine with 4 speed modules, no beacons
Production: 3.75 items/s
Energy: 212.8 kW/item

4. T3 assembly machine with 2 speed, 2 efficiency, no beacons
Production: 2.5 items/s
Energy: 117.6 kW/item

==> 66.7% ouput at 55% energy cost


Conclusion: There is some limited use for efficiency modules in case of receipts that do not use productivity modules. However, if you generate the energy with solar, the impact on UPS makes me use speed modules only :-/