Most performant item teleportation?
Looking at a megafactory-scale base - which mod(s) are the most performance-friendly for teleporting items at scale? E.g. teleporting all raw ores in the game, all finished products, etc.
Looking around I see a few possible candidates (Digital Storage, Storage Teleporter, Depot Downloader, FaxIt, etc.) however very little discussion around the performance impacts of those mods in large bases.
16 Replies
There are many variables, so I don't think it's an easy task to measure/compare their performance
You're not wrong, a better way to phrase the question may have been has anyone used these mods (or similar) in 1.0 with a large-scale factory and if so, was there any noticable performance impact
it would be great if you could do some testing, but I know that's a time commitment
the most performant one would probably be written using both C++ and blueprint code (as opposed to just blueprint) and one that operates infrequently on large batches at a time (like entire stacks). not sure what fits those criteria
Thinking you might be right, if there's no data out there then I'll have to go create some. Any particular tools/commands that would be good for measuring/comparing?
May have to design my factory with refactoring in mind so limiting any item teleportation to inputs and final outputs while leaving all the intermediate steps between the machines on belts so it's easier to refactor for testing and/or if there's a big performance issue
I've spun up a test save with all the popular mods for this I could find but they all seem to move items individually (albeit at a high speed) rather than stacks unfortunately
Looking a bit closer - the Dimensional Downloader mod is grabbing "stacks" (although sometimes not full ones? sometimes it grabs 200 plates, sometimes it grabs 50-100) into an internal buffer and then spitting those out which seems promising if paired with Instant Upload Speed to make it actually usable (even if the upload speed is significantly slower than the 95k/minute it shows in the UI lol)
It's a lot more limited than the likes of Digital Storage, especially given that mod can handle fluids/gases as well, but then that might be a good thing from a performance perspective
I use a mixture of mlb, magic splitters, back to production line, dimensional downloader. Only one I don’t use at the moment is digital storage.
There is little to no impact on performance, at least for me.
The unreal console command "stat fps" can give you frame times, but there's lot of other things that could get in the way of that. There must be something better
@Mrhid6 what do y'all use to test performance when testing DS?
We generally use VS cpu profiler
Is that something an end user can use by attaching a debugger or do they need source as well?
They would need to source @Rex
Rex helped me optimise DS so maybe they have some more tools
I just used VS' CPU profiler, even without source you can at least see which functions use up most of CPU time I believe
As a note we had a massive save that used DS massively and was running at something like 9fps and we managed to get it to around 40fps with the latest DS update
By we I mean @Rex [they/them] 😜
Nice. Not sure how far down the rabbit hole you guys went with testing, I've been wondering if the mod would perform even better if you only used Storage Connections for initial inputs and final outputs i.e. have chests at the beginning of the production line which export the raw materials from DS and chests at the end which import the results into the network. Leave all the machines doing the processing connected via belts
I haven't been able to find much detail on the performance impacts of say Factory/Crafting connections vs belts especially after the 1.2.1 performance update
If there is no data on that though then happy to take a look at VS CPU Profiler and see if I can gather some data on it. May take a few weeks to build the factory needed for testing at scale though :dogekek:
There is definitely some benefit to using factory adapter as you don’t have the overhead of items moving along belts/pipes but would be interested in both results
Will let you know how I get on, I couldn't find anything to say if the overhead of belts was expected to be more or less than items going in/out of DS via adapters. I assumed belts would be less overhead since CS have no doubt spent a lot of time optimizing belts but your comment implies the opposite so that's really interesting.
Appreciate the input from everyone (:
Yeah it’s only the rendering and the movement of the items as that might be tiny but something to consider
Micromanage, when it gets finally fixed. Or SCIM manual edit.
Game does not care if you move a belt halfway across the map, as long as its logically connected. Items will just 'teleport' across it.
Can be done with trains too.