Thanks Optera:)
In /editor mode all hidden items are already available
UPS optimized 12-beacon smelting - w/minimal circuits
Re: UPS optimized 12-beacon smelting - w/minimal circuits
I'm using the script w/my modificaitons from Factorio-Benchmark-Powershell
BTW a nice doc on benchmarking in reddit: /r/technicalfactorio/.../guide_how_to_benchmark_in_factorio/
BTW a nice doc on benchmarking in reddit: /r/technicalfactorio/.../guide_how_to_benchmark_in_factorio/
Re: UPS optimized 12-beacon smelting - w/minimal circuits
I've did an analysis on that, focusing on Editor Extensions. I'll switch to vanilla, not beacuse I've found better numbers, but to keep things simpler and clean.
See viewtopic.php?f=193&t=101673
Re: UPS optimized 12-beacon smelting - w/minimal circuits
I'd recommend 5 x 50.000 tick runs for a little bit better results. But 10.000 or 25.000 is also as good - but have at least 5 runs please.
Data and saves are in the same post above: viewtopic.php?f=193&t=101673&p=562072#p562072
There are graphs there related. Another graph (not in that post only here):
Data and saves are in the same post above: viewtopic.php?f=193&t=101673&p=562072#p562072
There are graphs there related. Another graph (not in that post only here):
- Attachments
-
- UPSvsTicks-ZMEE-10-10-6-runs.png (86.08 KiB) Viewed 5486 times
Re: UPS optimized 12-beacon smelting - w/minimal circuits
what is the build intended for?
for city blocks?
for city blocks?
Re: UPS optimized 12-beacon smelting - w/minimal circuits
Belter v6.80
- Applied the signal separaiton idea from this post - inspired by the BP from Blomz- Looks faster, benchmark to come later
- Attachments
-
- Belter v6.80.png (391.88 KiB) Viewed 5408 times
Re: UPS optimized 12-beacon smelting - w/minimal circuits
Benchmark - 4 x 25.000 ticks
- running vanilla Factorio, mods disabled- did my best to make sure all smelters which overproducing are full
- saves attached, I've created 5x500 lanes with express loader/infinity chests to be able to run w/o mods
- one controller for the 500 lanes
edit: see the next post for more results merged with this one
BPs tested
- Belter-v6.80: see BP in this thread here- DaveMcW-v4: see BP in this thread here
- Stevetrov_1feed: see savegame in this thread here
- DaveMcW-v5: see savegame in this thread here
- DaveMcW-v3: see BP in this thread here
- Flomz-v1.1: see BP in this thread here
- Stevetrov_2feed: see savegame in this thread here
Results
- I left my computer run alone for Run 1,2,3,4 - those are on the chart- No idea what happened for Flomz's run 2
- Run 5 removed from the chart: I've started browsing and listening to music, caused more randomness
- Would really appreciate another 5 x 25.000 run by someone else
- See timing numbers!
Fruther ideas, notes
- Apply signal separation to DaveMcW's BPs- I attach the script I've extended. It shows the UPS during measurements and contains some other tweaks to my taste
- Attachments
-
- benchmark4.zip
- (3.9 KiB) Downloaded 304 times
-
- B007-UPS.png (80.77 KiB) Viewed 5401 times
-
- B007-UPSruns.png (85.85 KiB) Viewed 5401 times
-
- B007-Times.png (73.38 KiB) Viewed 5401 times
-
- saves.zip
- (27.86 MiB) Downloaded 217 times
-
- B007 results.csv
- (3.94 KiB) Downloaded 185 times
Last edited by Belter on Thu Feb 24, 2022 7:58 am, edited 1 time in total.
Re: UPS optimized 12-beacon smelting - w/minimal circuits
2nd run
- I've run the same saves again 5x- got consistent results
- 2nd set attached
- My PC was running unattended, other apps closed
Results merged and filtered
- merged w/prev results- filtering out run #2 (Flomz data was an outlier)
- filtering out run #5 (All data was inconsistent as I've did some browsing during the tests)
- I've renamed the 2nd results to from 1..5 to 6..10
- B007-UPS-MergedFiltered.png (78.56 KiB) Viewed 5373 times
- B007-UPS-Runs-MergedFiltered.png (102.93 KiB) Viewed 5373 times
Results merged and unfiltered
Unfiltered results
- Attachments
-
- B007 results.csv
- (3.93 KiB) Downloaded 200 times
Re: UPS optimized 12-beacon smelting - w/minimal circuits
It seems that number of express loaders and chests considerably affects UPS (notice the difference between Stevetrov_1feed and Stevetrov_2feed).
We should probably try to elimitate this factor since there are no such entities in real megabases, and creating a half-belt in a real game often doesn't have this overhead.
Suggestions:
We should probably try to elimitate this factor since there are no such entities in real megabases, and creating a half-belt in a real game often doesn't have this overhead.
Suggestions:
- Use inserters to fill input belts, while making sure that number of infinity chests remains the same between versions
- Use electric drills with insane productivity/speed bonuses
Re: UPS optimized 12-beacon smelting - w/minimal circuits
EE underground belt has a much bigger impact... see here: 13%
I'd prefer the 2nd option in case it works in vanilla.
Adding more inserters would ruin our efforts to get the highest UPS and the lowest # of inserters active/swinging.
Re: UPS optimized 12-beacon smelting - w/minimal circuits
@Belter:
Your saves are not quite right. the point of 2 belts is to use one lane on each belt, that reduces computations for inserters a lot. also, you used 2 infinitychests per input. this gives 500 more active antities to deal with.
(If you want to know more, find "UPS wars" on this forum)
Notes:
- I prefer copying all designs down(south) 498 times, instead of 5x 100. I tried with Flomz' design, but got caught in the wires...
Belters design copied straight down gives a different performance (better) than 5x100
- we have 3 different timer designs atm: Belter/daveMcW, Stevetrov, and Flomz. I propose to put the timers in the name of the saves, I did that for this run. (dt,st,ft)
- agreed with Belter about runtimes of the saves.
- 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 chunks.
- the script I use makes sure no mods are loaded when running the test. (check in the logs afterwards to be sure)
- for quick checks you can use this lua snippet on the command line to count entities. fill in the correct prototype name for each one. (this one gives the stack inserter count)
Your saves are not quite right. the point of 2 belts is to use one lane on each belt, that reduces computations for inserters a lot. also, you used 2 infinitychests per input. this gives 500 more active antities to deal with.
(If you want to know more, find "UPS wars" on this forum)
Notes:
- I prefer copying all designs down(south) 498 times, instead of 5x 100. I tried with Flomz' design, but got caught in the wires...
Belters design copied straight down gives a different performance (better) than 5x100
- we have 3 different timer designs atm: Belter/daveMcW, Stevetrov, and Flomz. I propose to put the timers in the name of the saves, I did that for this run. (dt,st,ft)
- agreed with Belter about runtimes of the saves.
- 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 chunks.
- the script I use makes sure no mods are loaded when running the test. (check in the logs afterwards to be sure)
- for quick checks you can use this lua snippet on the command line to count entities. fill in the correct prototype name for each one. (this one gives the stack inserter count)
Code: Select all
/c local entity
entity = game.player.force.get_entity_count("stack-inserter")
game.player.print(entity)Stevetrov's build beats every other one, when properly setup and tested.
Test results:- 2022-02-24 20_18_14-testresults.ods - LibreOffice Calc.png (54.03 KiB) Viewed 5306 times
- Attachments
-
- Belter-v6.8_ft 500.zip
- (2.99 MiB) Downloaded 159 times
-
- Flomz-v1.1_ft 500.zip
- (3.7 MiB) Downloaded 146 times
-
- Stevetrov_v1_st 500.zip
- (3.26 MiB) Downloaded 112 times
-
- empty.zip
- (4.02 MiB) Downloaded 112 times
-
- results.csv
- (1.49 KiB) Downloaded 93 times
-
- Results.zip
- (6.95 MiB) Downloaded 95 times
Re: UPS optimized 12-beacon smelting - w/minimal circuits
How stable are the timing numbers?
- as using a savegame with full smelters, it should be consistent withing the cycle (144 ticks)- so how much is the noise?
1st .. 25th cycle
- I took the verbose timing data of my BP (Belter v6.80) and using the first 25 cycles created an animation- Using avg of 5 runs, same 1st ... 25th 144 cycles (so 25*144 = 3600 ticks)
- ignore the title left there by accident
- Time144-Anim.gif (1.91 MiB) Viewed 5295 times
1st .. 25th cycle averaged
- Another animation: lets average it, not just show- 1st frame is identical, last 25th frame is the avg of the 25
- ignore the title left there by accident
- Time144Avg-Anim.gif (1.75 MiB) Viewed 5295 times
Compared ot the big avg
- The 5 x 25 avg below is identical to the "big" avg (all runs, all cycles, 25.000 ticks):- title is proper
- Time144-AllAvg.png (99.58 KiB) Viewed 5295 times
- Attachments
-
- B007 verbose.zip
- (77.68 MiB) Downloaded 167 times
-
- B007 verbose-BelterOnly-1-c25only.zip
- (21.1 MiB) Downloaded 147 times
Last edited by Belter on Fri Feb 25, 2022 6:51 am, edited 1 time in total.
Re: UPS optimized 12-beacon smelting - w/minimal circuits
Well, looks there is some history here, let me check it out... I appreciate your comments as going to the same direction I'd like to go.disentius wrote: ↑Thu Feb 24, 2022 7:22 pm @Belter:
The point of 2 belts is to use one lane on each belt, that reduces computations for inserters a lot. also, you used 2 infinitychests per input. this gives 500 more active antities to deal with.
(If you want to know more, find "UPS wars" on this forum)
Re: UPS optimized 12-beacon smelting - w/minimal circuits
maybe someone will be interested
https://www.reddit.com/r/factorio/comme ... &context=3
https://www.reddit.com/r/factorio/comme ... &context=3
Re: UPS optimized 12-beacon smelting - w/minimal circuits
@flame_Sla:
1. /editor
2. Remove all entities
3. fill with lab tiles
2. /Cheat all
1. /editor
2. Remove all entities
3. fill with lab tiles
2. /Cheat all
Re: UPS optimized 12-beacon smelting - w/minimal circuits
Ran your saves, Stevetrov is +1.5% in this setup on my computer, will add DaveMcW's BP on the weekend.
Results:
- Attachments
-
- B008-UPS.png (80.91 KiB) Viewed 5218 times
-
- B008-Times.png (55.65 KiB) Viewed 5218 times
Re: UPS optimized 12-beacon smelting - w/minimal circuits
use mining drills, this is the best option
Re: UPS optimized 12-beacon smelting - w/minimal circuits
Well, switching to mining drills gives the opportunity to DI drill->smelter. Let's agree that for here at least the ore should come in belts, outside of the smelting area, produced by whatever # or drills w/2000% prod bonus (so one drill can fill half belt, matching late game setups).