, computers.
Would be really nice to read about this flood control, do you have some links?
I'm learning Lua now... Listing all entities, checking modules SM3 PM3 - 1 beacon with only 1 module as a test:
code
UPS optimized 12-beacon smelting - w/minimal circuits
Re: UPS optimized 12-beacon smelting - w/minimal circuits
I'm learning Lua now... Listing all entities, checking modules SM3 PM3 - 1 beacon with only 1 module as a test:
- Attachments
-
- Lua-ListAll-out.png (230.41 KiB) Viewed 8872 times
Re: UPS optimized 12-beacon smelting - w/minimal circuits
The console simple hides any duplicate messages in the past 60 ticks.
Here is a 11 beacon direct mining blueprint.
Here is a 11 beacon direct mining blueprint.
- 11-beacon-direct-mining.jpg (276.05 KiB) Viewed 8828 times
Last edited by DaveMcW on Mon Mar 14, 2022 8:48 am, edited 1 time in total.
Re: UPS optimized 12-beacon smelting - w/minimal circuits
Here is a 10.8 beacon blueprint that can fit on vanilla resource patches.
You can save 2 more beacons by joining the two rows together (also requires more signals).
You can save 2 more beacons by joining the two rows together (also requires more signals).
- 10.8-beacon-direct-mining.jpg (974.24 KiB) Viewed 8828 times
Last edited by DaveMcW on Mon Mar 14, 2022 8:49 am, edited 3 times in total.
Re: UPS optimized 12-beacon smelting - w/minimal circuits
A 10.6 beacon blueprint is possible, but that runs furnaces at 100% of capacity and requires Antikythera level clocking. I ran some calculations on what would be required.
Inserters set to stack size 11, 11, 12, 12, 12.
154.667 tick swing cycle.
23200 tick (150 swing) master cycle.
6 swings out of 150 where an 11 and 12 stack size are swapped (one for each permutation).
Even if the combinators to do this are essentially free, I am pretty sure setting stack size from the circuit network consumes too much UPS.
Inserters set to stack size 11, 11, 12, 12, 12.
154.667 tick swing cycle.
23200 tick (150 swing) master cycle.
6 swings out of 150 where an 11 and 12 stack size are swapped (one for each permutation).
Even if the combinators to do this are essentially free, I am pretty sure setting stack size from the circuit network consumes too much UPS.
Re: UPS optimized 12-beacon smelting - w/minimal circuits
Direct insertion BPs - benchmark
- 500 lanes- mining at 5000% /c game.player.force.technologies["mining-productivity-4"].level = 501
- no mods
- saves attached
DaveMcW v7
- 11 beacon setup - 10 smelters- 160 cycle
- BP published here in this thread
- hit 300 UPS on my computer
- I've tried to enable/disable the mining drills (red wire, 1 tick on : 30 ticks off ) but that caused -15% UPS
DaveMcW v8
- 10.8 beacon setup (8x11 + 2x10) - 10 smelters- 472 / 3 cycle, 11 stack size fo the 2 smelters with 10 beacons
- BP published here in this thread
Belter v8.16
- 10 beacon setup - 11 smelters- 176 cycle
- BP published here in this thread
More graphs and saves
- Attachments
-
- B025 results.csv
- (1.75 KiB) Downloaded 135 times
-
- B025 UPS.png (75.46 KiB) Viewed 8773 times
Last edited by Belter on Tue Mar 15, 2022 12:33 pm, edited 3 times in total.
Re: UPS optimized 12-beacon smelting - w/minimal circuits
Looks 10.4 beacon setup is the theoretical limt here for 10 smelters (w/speed modules)...
Do you think a smelter running continously is the best for UPS?
- Attachments
-
- theory.png (86.69 KiB) Viewed 8760 times
Re: UPS optimized 12-beacon smelting - w/minimal circuits
So for 10.4 - this is just a theory, how to get out of the produced plates from the smelters:
Precise timing of the yelow ones would lead to 295,680 cycles (11 and 12 stack sizes)
The blue ones look more promsing but less efficient: 384 cycle time (7 and 10 stack sizes, bad for UPS)
Too bad inserter cannot grab 14, it has number 7 in it..
Precise timing of the yelow ones would lead to 295,680 cycles (11 and 12 stack sizes)
The blue ones look more promsing but less efficient: 384 cycle time (7 and 10 stack sizes, bad for UPS)
Too bad inserter cannot grab 14, it has number 7 in it..
- Attachments
-
- 10.4 calcs.png (23.18 KiB) Viewed 8746 times
Re: UPS optimized 12-beacon smelting - w/minimal circuits
So for 10.4 shooting for 12 stack size which is the best of UPS:
- 80640 / 490 = 164.571 ticks (exactly 12 * 192 / 14)
- 80640 / 525 = 153.6 ticks (exactly 12 * 192 / 15)
Benchmakring this with 25.000 ticks looks a joke.. ..also putting these two cycles to the belt w/o waiting is not obvious, lets say... As the smelters can hold only 100 items, their buffer is 8.3 full 12 swings. Not 525.
- 80640 / 490 = 164.571 ticks (exactly 12 * 192 / 14)
- 80640 / 525 = 153.6 ticks (exactly 12 * 192 / 15)
Benchmakring this with 25.000 ticks looks a joke.. ..also putting these two cycles to the belt w/o waiting is not obvious, lets say... As the smelters can hold only 100 items, their buffer is 8.3 full 12 swings. Not 525.
Re: UPS optimized 12-beacon smelting - w/minimal circuits
Good idea using speed modules to reach 10.4 beacons.
1 speed/1 prod gives a higher production rate than 2 speed modules. That should give enough flexibility to find a nice solution.
1 speed/1 prod gives a higher production rate than 2 speed modules. That should give enough flexibility to find a nice solution.
Re: UPS optimized 12-beacon smelting - w/minimal circuits
10.4 beacon blueprint. I made a wide version for slightly better UPS, but it can easily be compressed to fit in a vanilla ore patch.
Note that the fast furnaces need 290 hours to fill their buffer, so I recommend disconnecting the circuit network for a minute instead!
Note that the fast furnaces need 290 hours to fill their buffer, so I recommend disconnecting the circuit network for a minute instead!
- 10.4-beacon-direct-mining.jpg (245.09 KiB) Viewed 8686 times
Re: UPS optimized 12-beacon smelting - w/minimal circuits
I used to remove the chest at the end, fill everything at 64x speed to make sure all smelters are full for proper benchmarking. Then clone 499x.
Will test this on the weekend to see if any better UPS-wise. The risk here if somebody submits a BP which is underproducing but very slowly. 25.000 ticks is only 7 game minutes / 0.11 game hours. So if you need an extra plate every 5 seconds we won't notice...
Re: UPS optimized 12-beacon smelting - w/minimal circuits
test the output first with 1 smelter row. (I do) speaker attached to belt, set reading on hold, let speaker beep at anything < 8. let it run a while on speed 64.
Re: UPS optimized 12-beacon smelting - w/minimal circuits
Works! Honestly this timing is far from obvious...
DI Benchmark results
- 500 lanes, 25k x 10 runs- DaveMcW v9.0: 10.4 beacon setup from here
- DaveMcW v7.0, DaveMcW v8.0, Belter v8.16: same as here
- Saves also attached - see the spoiler seciton below
More grpahs and saves
- Attachments
-
- B026 UPS.png (81.9 KiB) Viewed 8221 times
-
- B026 Times.png (79.19 KiB) Viewed 8221 times
-
- B026 results.csv
- (4.4 KiB) Downloaded 124 times
Re: UPS optimized 12-beacon smelting - w/minimal circuits
Can we have a separate category for on-patch smelting? I understand that that is best objectively but it is not really applicable to games using settings closer to default.
Re: UPS optimized 12-beacon smelting - w/minimal circuits
This is the last, non-DI result and BPs: viewtopic.php?p=562815#p562815
Re: UPS optimized 12-beacon smelting - w/minimal circuits
Here is the 10.4 beacon blueprint modified to fit on a 23x29 ore patch.
I also made the combinator logic easier to understand.
I also made the combinator logic easier to understand.
Re: UPS optimized 12-beacon smelting - w/minimal circuits
This is a really cool BP! Due to the 10.4 setup which is ( 2x11 + 3x10 beacon ) x 2 lanes it was possible to make it really compact:
- DaveMcW v9.png (845.46 KiB) Viewed 7906 times
- There is a "normal" 160 tick cycle as at tick 129: "everything" is emitted. So all inserters controlled by signals #1,#2 and #3 triggered.
see it in action
- There is a 4128 tick clock which is 25.8 x 160. So every 25th 160 cycle is shorter (only 128 ticks), and there are 2 additional triggers for the inserters of the 11 beacon smelters: at 2, 194 and 96, 162. Also in this cycle the "normal" sign is not triggered as it would at 129. So instead of 2, those inserters triggered 3x.
Re: UPS optimized 12-beacon smelting - w/minimal circuits
Belter v9.2
- I was not able to beat the speed of Dave's last BP.- It might be still interesting to see this 8 Beacon DI BP as it takes less vertical space
- Demand = Supply exactly. Not over or underproducing. Note: for 8 Beacon setup 2xSM3 is the fastest
- Reqs 12 smelters ( vs 10.4 or 11 for faster DI BPs) -> more entities
- 1 signal is used for perfect insertion
- BP Belter v9.2.png (491.81 KiB) Viewed 7808 times
Comparison to DaveMcW v9.1
- 6% slower UPS: 275 vs 290- transport and entity updates showing more # of entities (worse)
- signalling time is proportional to the 5 : 6 insterter entity count (worse)
- electricity is slightly better due to the less amount of power entities (better)
Due to the perfect sync and single signalling the time graph is really cool: shows both plates produced and even inserted:
- Times Belter v9.2.png (128.12 KiB) Viewed 7808 times
Saves and more pics
Re: UPS optimized 12-beacon smelting - w/minimal circuits
You can compress the mining drills an additional 1 tile vertically and 3 tiles horizontally.
- compressed.jpg (879.41 KiB) Viewed 7794 times
Re: UPS optimized 12-beacon smelting - w/minimal circuits
This is cool, again! I've tried to simplify it but anyting I tried made it slower. At this level of UPS optimization a lane side change or side merge does not work, see my ideas.
I also attach a save for DaveMcW v9.2 if anybody want to benchmark w/500 lanes.
UPS: 277 for the original BP, 260 or lower for the alternates on the pics.
- Attachments
-
- DaveMcW v9.22.png (611.12 KiB) Viewed 7670 times
-
- B033 DaveMcW v9.2.zip
- (3.63 MiB) Downloaded 106 times
-
- DaveMcW v9.21.png (589.86 KiB) Viewed 7703 times