UPS Optimization - Red Science (editor)

[Belter]

Current results w/100 belts

1. 616 UPS - guitars11 v4.2.2 - FactorioBox
2. 612 UPS - guitars11 v4.2.7 - FactorioBox
3. 610 UPS - guitars11 v4.5 - FactorioBox
4. 607 UPS - guitars11 v4.2.6 - FactorioBox
5. 606 UPS - guitars11 v4.5.1 - FactorioBox
6. 602 UPS - guitars11 v4.2 - FactorioBox
7. 592 UPS - guitars11 v4.2.2-8t - FactorioBox
8. 590 UPS - Stevetrov v1 - FactorioBox
9. 585 UPS - guitars11 v4.2.11 - FactorioBox
10. 582 UPS - guitars11 v4.2.12 - FactorioBox
11. 577 UPS - guitars11 v4.2.10 - FactorioBox
12. 560 UPS - guitars11 v4.1.1 - FactorioBox
13. 558 UPS - guitars11 v4 - FactorioBox
14. 492 UPS - Belter v1.03 - FactorioBox
15. 465 UPS - guitars11 v3 - FactorioBox
16. 320 UPS - Belter v0.53 - FactorioBox

For reference: Stevetrov's 10k belt megabase (1.1) gives 93 UPS, flame_Sla 10k - 10x1000spm Belt Module 116 UPS on my potato i5 4 core CPU.

Edit 16th Apr: on my system turning off AV, Google Drive and Chrome during benchmarking gives 5+% difference, so had to redo all measurements. At the same time consistency is better when those are off.

Edit 22th Apr: added all as of writing. 1st updated from 620 -> 616 based on multiple runs at FactorioBox. Factoriobox links now goes to measurement there

Target: 1 full blue belt of Red Science pack produced, best UPS wins

For ppl interested in creating the ultimate UPS optimized Red Science factory
A very late game scenario, when all tech is available and care about optimization only
Output: 1 full blue belt (2700 item/minute) no gaps
Success criteria: best UPS. No questions asked beyond that. DI/Trains/Spaghetti base: just make it fast!
If there is interest I'll move fwd to the other science packs

Conditions: end-game scenario, vanilla only, editor allowed

Input: wherever ore you add wtih the editor. You must mine the ore.
Minig Prod: +1990% (so an electric mine can produce a full lane w/3 SMs)
Mods: none, vanilla. For editing you can use whatever you want, but measurement will be done w/o any mods
Extra items allowed from editor per belt:
- 1 x Electic energy interface only to produce electricity - pls make a clear power entry, don't put this in the middle
- 1 x Express loader + 1 x Infinity box only to consume the RS packs
Specify ramp-up time (not more than 10 mins please)
Must be able to produce a full blue belt w/o any gaps for 1 hour
TBD: if someone is interesting in roboports, let's agree on a late game upgrade level

To submit here: 1 belt savegame, 100 belt save nice to have

Give save game for 1 full blue belt preferred savegame name: Red Science UPS Editor - <your name> v<version>.zip
Adding a picture and some description is welcome
Adding a BP is also a nice to have (but due to editor mode ore patches are not in the BP)
You might attach a savegame of 100 belts optionally to make my life easier.
You might upload it also to FactorioBox for quick testing for everyone

Measurement: 100 belts

Will measure w/a 100 region cloned map. If possible allow cloning down to south.
Merging signalling for the 100 belt is OK, then give a map which supports it.
Adding storage boxes not preffered as might introduce huge buffers and ramp-up times / inconsistency. Use them wisely.
I plan to post my benchmark results weekly here.
If a gap is detected after the ramp-up time the submission is disqualified.

Editor tips

details

Set mining prod to 2000%:

Code: Select all

/c game.player.force.technologies["mining-productivity-4"].level = 200

List all entities - good to find bad modding mistakes:

Code: Select all

/c  stats = {} entitycount = 0 for _, entity in pairs(game.player.surface.find_entities()) do   entitycount = entitycount + 1   inv = entity.get_module_inventory()   key = ""   if inv ~= nil then     pm3count  = inv.get_item_count("productivity-module-3")     sm3count  = inv.get_item_count("speed-module-3")     key = entity.name .. " pm3: " .. pm3count .. " sm3: " .. sm3count   else     key = entity.name   end   if stats[key] == nil then stats[key] = 1 else stats[key] = stats[key] + 1 end end game.print("Total: " .. entitycount) for i,k in pairs(stats) do    game.print(k .. " " .. i)  end  

Mods for editing:
- Rate Calculator by raiguard - most useful, calculates throghput
- Calculator UI by SoHWichu - in game calculator
- Delete Empty Chunks by Strath
- Editor Extensions by raiguard - don't leave these entities in the final savegame
- Power Grid Comb by shanemadden - clean up messy wires
- Region Cloner by mulark

Links relevant/might be interesting

Best UPS smelting: DaveMcW's in UPS optimized 12-beacon smelting - w/minimal circuits
Factoriolab for simple calculations
Factorio wiki: Belt transport system
Factorio wiki: Transport belts Physics
Factorio wiki: Insertion_limits
Prev thread in 2022, but not edit mode: UPS optimized Ore->Red Science 2700 spm
Relevant Reddit threads in /technicalfactorio
- Test Case: Red Science
- UPS Wars 6: Labs
- advertising this topic here
factoriobox
- Stevetrov's 10k belt megabase (1.1)

Comments/suggestions welcome!

[Belter]

Belter v1.0

edit: see Belter v1.03
- baseline/example for 8 beaconed direct insertion / direct mining production
- not timed (yet)
- not nice

[Belter]

Theoretical calcs for the Assembly Machine to produce the Red Science pack

- a full lane (half belt) needs 1350 / min to produced and put
- AM3 has an 1.25x multiplier
- Build time (1x) of Red Science Pack: 5 seconds
- assuming a 100% feed (so the assembly machine never runs out of copper plates or gears)

The followig # of beacons required (with 2 x Speed Module #3) for a full lane (half belt):

Red Science AM3 Calc.png (5.74 KiB) Viewed 5925 times

A detailed graph showing the numbers behind this table:

Red Science AM3 Calc for full lane.png (84.79 KiB) Viewed 5925 times

• AM3: Assembly Machine 3

• SM: Speed Module (Level 3)

• PM: Production Module (Level 3)

[Belter]

Updated the first post a bit, hope it is more readable now.

[guitars11]

Here is my red science. It is already 100 belts and should be ready to go for benchmarking.

Screenshot 2023-04-08 133830.png (3.15 MiB) Viewed 5650 times

[Belter]

Belter v0.52

edit: see Belter v0.53 instead
- finally managed to feed a 12 bacon setup - required timing of both input and output
- might be slower, not checked yet
- gear production not timed

Red Science UPS Edit - Belter v0.52.png (932.12 KiB) Viewed 5576 times

[guitars11]

This is a significant upgrade to what I posted earlier. According to my benchmark, my previous post was getting around 1.550 ms avg and this is getting around 1.180 ms avg.

Screenshot 2023-04-08 211806.png (2.98 MiB) Viewed 5574 times

[Belter]

Belter v0.53

- just cosmetic changes for easier 100 lane clone
- savegame is x100 belts
- numbers show that DI is the way to go... Added for reference mainly for a fully beaconed setup
- smelting is done via DaveMcW's UPS winner setup
- could be improved w/gears machinery optimization, but so far UPS-wise from the others...

Factoriobox: 295 UPS on my potato CPU

[Belter]

Belter v1.03

- timed 8 beacon DI setup
- horizontal layout
- savegame is x100 belts
- possible improvements: lane change with bi-drirectional setup
- possible improvements: timing plate insertion
- possible improvements: turn on inserters for a shorter period of time

https://factoriobox.1au.us/map/info/310 ... 8f97b373cc / 454 UPS on my potato CPU

[Belter]

This is a significant upgrade to what I posted earlier. According to my benchmark, my previous post was getting around 1.550 ms avg and this is getting around 1.180 ms avg.

This rocks, nice job!!

I'll do a local benchmark with more runs and with some analysis.
https://factoriobox.1au.us/map/info/ed1 ... cfe29d2004 / 518 UPS on my potato CPU

[guitars11]

Here is another improvement by about 7% from v4. avg time is now about 1.10 ms compared to 1.18 ms. Made all of the inserters that output red science go onto stubs first and also made it tileable.

Screenshot 2023-04-09 214134.png (2.16 MiB) Viewed 5431 times

[Belter]

Here is another improvement by about 7% from v4. avg time is now about 1.10 ms compared to 1.18 ms.

The same file (did a binary compare) uploaded to FactorioBox Red Science UPS Editor - guitars11 v4.1.1 - assume it was you, TY.

530 UPS on my computer, nice!

Would be nice to get UPS numbers from others as well. In case you guys trust this site - as the script downloads another script and and savegames as well. Then uploads data from your system there.

I gather the current versions and results on my system in the first post.

[flame_Sla]

1.112 ms - 0899.049 ups - Red Science UPS Edit - Belter v1.03 x100.zip
1.141 ms - 0876.720 ups - Red Science UPS Editor - guitars11 v3.zip
0.911 ms - 1098.055 ups - Red Science UPS Editor - guitars11 v4.zip
0.899 ms - 1112.395 ups - Red Science UPS Editor - guitars11 v4.1.1.zip

[ 10600k - ddr4-3600 ]

[guitars11]

Here is another improvement by about 7% from v4. avg time is now about 1.10 ms compared to 1.18 ms.

The same file (did a binary compare) uploaded to FactorioBox Red Science UPS Editor - guitars11 v4.1.1 - assume it was you, TY.

530 UPS on my computer, nice!

Would be nice to get UPS numbers from others as well. In case you guys trust this site - as the script downloads another script and and savegames as well. Then uploads data from your system there.

I gather the current versions and results on my system in the first post.

Thanks! Yes I did upload the map and have run a couple benchmarks on v4 and v4.1.1, getting around 850 and 890 respectively. Just wanted to say thanks for setting up this challenge. I like optimizing and such

[guitars11]

1.112 ms - 0899.049 ups - Red Science UPS Edit - Belter v1.03 x100.zip
1.141 ms - 0876.720 ups - Red Science UPS Editor - guitars11 v3.zip
0.911 ms - 1098.055 ups - Red Science UPS Editor - guitars11 v4.zip
0.899 ms - 1112.395 ups - Red Science UPS Editor - guitars11 v4.1.1.zip

[ 10600k - ddr4-3600 ]

Awesome! Thanks flame for benchmarking. Looks like 4.1.1 wasn't a 7% uplift from 4, seems like it's more 2-4% depending on cpu

[Belter]

Here is my red science. It is already 100 belts and should be ready to go for benchmarking.

Add v3 to FactorioBox as well: Red Science UPS Editor - guitars11 v3 405 UPS

[guitars11]

Here is another about 5% improvement from v4.1.1. Made one long build instead of combining 2 builds together which reduces the amount of transport sections. I'm not entirely sure what to call it but essentially these sections are created whenever an inserter is placing/picking things on the belt as well as merging belts together.

Screenshot 2023-04-11 182211.png (1.46 MiB) Viewed 5196 times

(v4.2). These arrows on the belts determine the 'sections' for a belt and making the build longer reduces the sections by 1 since we don't have to combine the top build to the bottom build.

Screenshot 2023-04-11 182459.png (1.24 MiB) Viewed 5196 times

(v4.1.1).

here is the factoriobox https://factoriobox.1au.us/map/info/e54 ... 114218dcb5
factoriobox v4.1.1 and v4.2 is about 11% improvement

[Belter]

Here is another about 5% improvement from v4.1.1. [...] v4.2 is about 11% improvement

Added to the list (to the top:))

I'm thinking about to measure this setup with some changes. Currently these are in theory only, maybe on the weekend:
- a better/perfect timing but this PM mod is tricky. In perfect timing case the inserter should pick 11 or 12 items, after every 8th cycle. Getting rid of the buffering should improve UPS
- also it worth checking if using 1 or 2 speed modules is better for UPS (would require only 2x14 machines not 2x15 but then more plates/gears needed)
- timing plate/gear inserters

[guitars11]

About a 2% improvement from v4.2. The main difference is moved around the beacons a little bit where the iron is which allowed 1 more beacon to affect the gears and copper. This increased downtime which seems to have helped just slightly. Even though the overflow belts aren't connected anymore, which increases the transport sections by 1, it suggests that downtime is more crucial.

Increasing the speed of the miners actually hurt performance but I'm not entirely sure why. It would suggest that adding speed modules to miners would make them be inactive for longer therefore increasing ups, but this wasn't the case.

v4.2

Screenshot 2023-04-12 094219.png (1.9 MiB) Viewed 5137 times

v4.2.2

Screenshot 2023-04-12 093830.png (1.39 MiB) Viewed 5137 times

[guitars11]

Here is another about 5% improvement from v4.1.1. [...] v4.2 is about 11% improvement

Added to the list (to the top:))

I'm thinking about to measure this setup with some changes. Currently these are in theory only, maybe on the weekend:
- a better/perfect timing but this PM mod is tricky. In perfect timing case the inserter should pick 11 or 12 items, after every 8th cycle. Getting rid of the buffering should improve UPS
- also it worth checking if using 1 or 2 speed modules is better for UPS (would require only 2x14 machines not 2x15 but then more plates/gears needed)
- timing plate/gear inserters

Yeah some good ideas.
- I think this is something to look into because right now the inserters are clocked to pick up 8.4 items/sec. I'm not entirely sure where the community found this number, if it is just from trial and error or if there is some math behind it but would be interested to see perfect timings. Is the PM mod a mod which calculates timings?
- I have tried 2 things so far. The first one is just speeding the furnaces which seems to hurt performance by about 2% I think. This also increases mining activity time since we are no longer using prod in the furnaces. The second was trying to figure out using 2x14 machines like you suggested. It seems having 1 speed and 3 prod for the science can be done using 2x14 machines but am unable to find the correct clock cycle for the inserters. So figuring out how clocks are calculated for science would be a big help.
- I don't think this would be beneficial because the inserters are only active when the science needs those items so the inserters would handle their own sleep/wake cycle. This would also add more circuits, increasing ups. I would say try it and see if it proves your theory