Factorio Forums

www.factorio.com
https://forums.factorio.com/

UPS optimized 12-beacon smelting - w/minimal circuits

https://forums.factorio.com/viewtopic.php?f=193&t=100763

Page 7 of 10

Re: UPS optimized 12-beacon smelting - w/minimal circuits

Posted: Sat Feb 26, 2022 5:34 pm
by Belter
Belter v6.90
I've added half belt feed to see if there is any benefit, a bit faster. Benchmarks to come later.

Re: UPS optimized 12-beacon smelting - w/minimal circuits

Posted: Sat Feb 26, 2022 5:39 pm
by Belter
Benchmarking with mines
- Per latest discussion: drills to produce ore, 500 lanes vertical, one controller circuit, no mods during benchmarking
Drilling Mine vs Express Loader&Inf chest: Drilling Mines won
- Drills gave better results: bigger difference between results for the same BPs
- Also more realistic: for megabases 2000% mining eff is realistic (1700% is enough w/3xSM3)
- As region cloner supports ore fields it is not that difficult to set it up
Saves
- I attach the saves - please run them 5 x 25.000 ticks. I'll add at least one of DaveMcW's BP here as well later
- I've corrected a save, TY for the comment @flame_Sla !
- All BPs are in this thread
Results
- I've excluded run #5, it got a 204 UPS result for Stevetrov 2f (its in the CSV attached)
- Will do another test w/8 runs
- Time graphs from this run here

Re: UPS optimized 12-beacon smelting - w/minimal circuits

Posted: Sat Feb 26, 2022 6:28 pm
by disentius
It is a relative test... it doesn't matter what you use, as long as the different setups in the same test use the same feed.

Re: UPS optimized 12-beacon smelting - w/minimal circuits

Posted: Sat Feb 26, 2022 6:31 pm
by flame_Sla
Belter wrote: ↑
Sat Feb 26, 2022 5:39 pm
Benchmarking with mines
Per latest discussion I'll publish results here. I attach the saves w/mining drills - please run them 5 x 25.000 ticks. I'd like at least one of DaveMcW's BP here as well.
mining drill has a different number of beacons
this may affect

B009-Stevetrov_v1_st 500 - mining drill has only one speed module instead of 3

Re: UPS optimized 12-beacon smelting - w/minimal circuits

Posted: Sat Feb 26, 2022 6:33 pm
by Belter
flame_Sla wrote: ↑
Sat Feb 26, 2022 6:31 pm
 B009-Stevetrov_v1_st 500 - mining drill has only one speed module instead of 3
aaah, correctied, TY! Also moving out of the drills from the beacon range just to be safe.

Re: UPS optimized 12-beacon smelting - w/minimal circuits

Posted: Sun Feb 27, 2022 1:37 am
by Belter
Times
- Using a verbose output and AVGing the times per <tick> mod <cycle time> we get a "fingerprint" of the BP
- Data from this benchmark, UPS/avg Time charts are there
- Verbose data attached here
- Lower is the better
Stevetrov 1f vs 2f
- The difference is the 2 belt feed, which gave a UPS gain
- Due to the "tricky" 9600 cycle and 67 increment I was not able to AVG all ticks - would need a 9600 pixel wide graph...
- Only a 500 tick window of the 9600 tick cycle shown
- Each data point is avg of 25.000 x 5 / 9600 = 13 points (but the others looks the same)
- Due to buffering (last smelter) and odd cycle this BP is less consistent
B009-TimesVS-Stevetrov v1 1f vs 2f.gif
B009-TimesVS-Stevetrov v1 1f vs 2f.gif (153.74 KiB) Viewed 4628 times
Belter v6.8 vs v6.9
- The difference is the 2 belt feed, which gave a UPS gain
- Due to the 144 cycle time its the easiest to show. Triggering inserters in the cycle: 0,32,64,96,128 - graph off by 1 tick sry
- Clearly visible reduced transport and entity updates due to easier picking up ore from a 1 sided belt (v6.9)
- Each data point is avg of 25.000 x 5 / 144 = 868
B009-TimesVS-Belter v6.8 vs v6.9.gif
B009-TimesVS-Belter v6.8 vs v6.9.gif (108.81 KiB) Viewed 4628 times
Flomz
- Put here for comparison, also using 144 cycle time, but different triggering points (1,46,92,95 and 141 - graph off by 1 tick sry)
- Each data point is avg of 25.000 x 5 / 144 = 868
- Circuit times are better but only a little bit compared to Belter: separat wires (Flomz) vs separate signals on the smae wire (Belter)
B009-Times-Flomz v1.1-144.png
B009-Times-Flomz v1.1-144.png (73.15 KiB) Viewed 4628 times

Re: UPS optimized 12-beacon smelting - w/minimal circuits

Posted: Sun Feb 27, 2022 9:19 am
by flomz
V2.1, should be significantly faster:
  • Managed to reduce number of separate CNs to 3
  • Now using a single belt and 9 furnaces

Re: UPS optimized 12-beacon smelting - w/minimal circuits

Posted: Sun Feb 27, 2022 9:07 pm
by Belter
flomz wrote: ↑
Sun Feb 27, 2022 9:19 am
 V2.1, should be significantly faster:
Looks good! Also some parts looks familiar :) I'll measure it, created a prepared savegame. Can you prep it next time?

Also adding DaveMcW's v4 w/mines.

My plan to benchmark:
- Flamz v2.1 (using the save attached here)
- Belter v6.9 (re-using the save there)
- Stevetrov v1-2f (re-using the save there)
- DaveMcW v4 (using the save attached here)

Re: UPS optimized 12-beacon smelting - w/minimal circuits

Posted: Mon Feb 28, 2022 5:14 am
by flomz
Belter wrote: ↑
Tue Jan 20, 1970 1:13 am
 Looks good! Also some parts looks familiar :)
Yeah, output belt layout was inspired by your version)
Although inserter placement is changed to always interact with belt on underground input part, closer to the beginning of a transport line.
Belter wrote: ↑
Tue Jan 20, 1970 1:13 am
 Can you prep it next time?
Sure, is there an instruction somewhere?

Re: UPS optimized 12-beacon smelting - w/minimal circuits

Posted: Mon Feb 28, 2022 7:36 am
by Belter
Benchmark
- Vanilla, all mods disabled
- 500 lanes for smelting vertically after each other
- 1 circuit controller/s for all
- Ore: from mining drills w/2000% bonus, outside of the beacon smelting area range via belts
- Plates: to infinity chests via belts
- 25.000 ticks 10 runs
- All overproducing smelters are full
Results
- Close, but Flomz v2.1 is the fastest
- Removed Run 1 (black) from the averages - as ususal forgot to close browser/etc and did it during

Re: UPS optimized 12-beacon smelting - w/minimal circuits

Posted: Mon Feb 28, 2022 7:40 am
by Belter
flomz wrote: ↑
Mon Feb 28, 2022 5:14 am
Belter wrote: ↑
Tue Jan 20, 1970 1:13 am
 Can you prep it next time?
Sure, is there an instruction somewhere?
See this great article:
https://www.reddit.com/r/factorio/comme ... &context=3

I attach the script I use - this is a modified version of the one mentioned there. It shows UPS after each tests and gives verbose data in 1 sheet for easier Times charts.
See here
As disucssed in this thread, in short:
/editor
Remove all entities
fill with lab tiles
/c game.player.force.research_all_technologies()
/c game.player.force.technologies["mining-productivity-4"].level = 201
/c game.player.force.technologies["worker-robots-speed-6"].level = 16
/cheat all
add your BP at 0,0
add the ore fields and drilling mines out of the beacon range
ran it to fill smelters if needed
clone propetly w/region cloner to 500 lanes south or north
run it a bit (I do at least 10 game minutes, 1hr would be the best)
save it
Tests must ran with vanilla, all mods disabled.

Saves and a good set of points for testing: see (up arrow goes to that post)
disentius wrote: ↑
Thu Feb 24, 2022 7:22 pm
 - made an "empty" save with a hazard concrete point at 0.0 on the map, so all builds can be lined up the same. Includes 2 mods: region cloner, and delete empty
Note: I switched to the experimental version of Factorio (1.1.55) as there was save with that version which I wanted to check and forgot to set it back... We'll go back to stable versions I hope for future. You can get that verison by changing your Factorio setting in Steam.

Re: UPS optimized 12-beacon smelting - w/minimal circuits

Posted: Mon Feb 28, 2022 8:01 am
by Belter
flomz wrote: ↑
Mon Feb 28, 2022 5:14 am
Belter wrote: ↑
Tue Jan 20, 1970 1:13 am
 Looks good! Also some parts looks familiar :)
Yeah, output belt layout was inspired by your version)
Although inserter placement is changed to always interact with belt on underground input part, closer to the beginning of a transport line.

Please don't misunderstood my comment, we all look around and implement ideas from each other, refine our BPs and came to the same solutions.

Just found that there were UPS Wars 3 2 and 1 - I perosnally invented many ideas myself discussed there and other places years ago. Also re-used ideas for the great people in this thread :D - lot of useful information here.

Re: UPS optimized 12-beacon smelting - w/minimal circuits

Posted: Mon Feb 28, 2022 11:27 am
by disentius
empty testmap with region cloner and delete empty chunks, marker pointing to 0,0 factorio version 1.55

Re: UPS optimized 12-beacon smelting - w/minimal circuits

Posted: Mon Feb 28, 2022 12:14 pm
by DaveMcW
I removed a beacon from Flomz v2.1



less-beacon.jpg
less-beacon.jpg (766.81 KiB) Viewed 4384 times

Re: UPS optimized 12-beacon smelting - w/minimal circuits

Posted: Mon Feb 28, 2022 2:08 pm
by flomz
DaveMcW wrote: ↑
Mon Feb 28, 2022 12:14 pm
 I removed a beacon from Flomz v2.1
Nice! Transport line performance should improve as well.

Re: UPS optimized 12-beacon smelting - w/minimal circuits

Posted: Thu Mar 03, 2022 12:53 am
by Belter
flomz wrote: ↑
Sun Feb 27, 2022 9:19 am
  • Managed to reduce number of separate CNs to 3
Why 3 if 2 also possible? :-)
Belter v6.95
- Based on this BP separate signalling is possible, using green and red wires as only 2 needed. I'll do that a on the weekend and measure both
- this BP might be optimized a bit... just make sure the timed insterters stay the same position relative to each other
- I did the calculations actually to see if its possible to use only 2 signals...
...here is how

Re: UPS optimized 12-beacon smelting - w/minimal circuits

Posted: Thu Mar 03, 2022 2:21 am
by DaveMcW
Belter wrote: ↑
Thu Mar 03, 2022 12:53 am
- this BP might be optimized a bit... just make sure the timed insterters stay the same position relative to each other
You can remove a beacon from the right side by using consistent 4-tile spacing between furnaces and a beacon gap at each furnace.

Unfortunately the left side requires a 12-tile gap between furnace #3 and #5, which is not compatible with minimal beacons.

Re: UPS optimized 12-beacon smelting - w/minimal circuits

Posted: Thu Mar 03, 2022 3:05 am
by DaveMcW
Here is a 2-signal version with 60 beacons. It is almost perfect, it only wastes 3 ticks of activity on the middle inserter.



2-signals.jpg
2-signals.jpg (782.54 KiB) Viewed 4137 times

Re: UPS optimized 12-beacon smelting - w/minimal circuits

Posted: Thu Mar 03, 2022 5:02 pm
by disentius
Nice dave.
Tested 3 latest saves:
Belter v6.9
Stevetrov v2
DaveMcw v6

- 500 belts vertical saves
- Fed by miners, mining prod set to 2000, no modules needed ( I changed belter's save in that regard)

Results: (dave has the fastest iteration atm)
2022-03-03 17_54_28-bmresults.ods - LibreOffice Calc.png
2022-03-03 17_54_28-bmresults.ods - LibreOffice Calc.png (62.5 KiB) Viewed 4073 times

Re: UPS optimized 12-beacon smelting - w/minimal circuits

Posted: Fri Mar 04, 2022 7:51 am
by Belter
Dave's is the best so far, again... I'd replace the filter inserters to normal ones - same from UPS point of view.
disentius wrote: ↑
Thu Mar 03, 2022 5:02 pm
 - Fed by miners, mining prod set to 2000, no modules needed ( I changed belter's save in that regard)
Good idea, one less problem to care about (e.g. forgotten modules in the miners :D ...).

In the saves miners are in the range of the beacons, should be moved 2 tiles left for proper testing.
All times are UTC
Page 7 of 10

Powered by phpBB® Forum Software © phpBB Limited