Page 1 of 1
[0.16.51] Uranium processing + productivity modules = lost items
Posted: Sun Sep 23, 2018 10:33 pm
by Merssedes
Discovered following today:
After 500 production cycles with 2 x PR3 modules i got only 598 items out of 600.
I suspect that game incorrectry processes situation when items, produced by AM and by modules in AM are different. For example, AM procuced U-238 and module produces U-235 and one of them gets lost.
To be specific, I found
this, wich marked as "resolved", but have similar actions by description with the exception that now recipes not changing.
Re: [0.16.51] Uranium processing + productivity modules = lost items
Posted: Sun Sep 23, 2018 11:01 pm
by Loewchen
I don't understand why you would consider this a bug, the results are produced by chance and your results are only slightly below expectation.
Unless you can show that this happens reliably I would assume that you were simply unlucky.
Re: [0.16.51] Uranium processing + productivity modules = lost items
Posted: Sun Sep 23, 2018 11:36 pm
by Merssedes
I'm about single U-235: it's really just chance based.
AM states that it produced 600 items, chest contenst only 598 total (nothing being taken from there).
About more consistent results: in vanilla there is one more occurance of changing outputs on-the-fly: mining drill placed on top of multiple resources.
I spawned 2 resource entities: iron ore and copper ore with amount of 1000.
With productivity 50% while mined separately, they produce 1499 items each = 2998 total.
But if we try to mine them together with single miner, we get 1495 + 1496 items (im my 4 tests 2 times there was more copper, 2 times -- iron, but all the times with the same amounts) = 2991 total.
Where 7 items go?
Re: [0.16.51] Uranium processing + productivity modules = lost items
Posted: Mon Sep 24, 2018 12:13 am
by quyxkh
"AM"? You mean the centrifuge? The count is of how many cycles it's run. You can see this by watching an assembler produce yellow belts, one finished "product" is whatever the recipe produces. Producing yellow belts the product is two yellow belts every time, 10 "products" is 20 yellow belts because one "product" is two yellow belts. Producing uranium the product is a 99.3(I think)% chance of U-238 and a 0.7% chance of U-235. 600 "products" (the 20% productivity bonus works out at exactly 100 extra products per 500 un-bonused products) is two random numbers, U-238 within a few of 596, U-235 within a few of 4.
Re: [0.16.51] Uranium processing + productivity modules = lost items
Posted: Mon Sep 24, 2018 5:35 am
by Merssedes
quyxkh wrote: ↑Mon Sep 24, 2018 12:13 am
"AM"? You mean the centrifuge? The count is of how many cycles it's run. You can see this by watching an assembler produce yellow belts, one finished "product" is whatever the recipe produces. Producing yellow belts the product is two yellow belts every time, 10 "products" is 20 yellow belts because one "product" is two yellow belts. Producing uranium the product is a 99.3(I think)% chance of U-238 and a 0.7% chance of U-235. 600 "products" (the 20% productivity bonus works out at exactly 100 extra products per 500 un-bonused products) is two random numbers, U-238 within a few of 596, U-235 within a few of 4.
Yes, i do. Just was not able to remember how it's named in english. Agree with other part.
Expected result: count of U-235 + count of U-238 = 600.
What i got: count of U-235 + count of U-238 = 598.
UPD:
I run 10 tests with centrifuge with 2 PR3 modules (productivity + 50%) and 6 beacons with 2 SP3 modules each (resulting in speed of 2.775) with total of 6000 uranium ore input each.
Expected behaviour: 720 total units of output in each test.
Results are...
Code: Select all
Run | U-238 | U-235 | sum
1 | 715 | 9 | 724
2 | 711 | 6 | 717
3 | 716 | 2 | 718
4 | 716 | 5 | 721
5 | 719 | 3 | 722
6 | 719 | 7 | 726
7 | 714 | 5 | 719
8 | 714 | 6 | 720
9 | 712 | 6 | 718
10 | 715 | 6 | 721
average:
| 715.1 | 5.5 | 720.6
Re: [0.16.51] Uranium processing + productivity modules = lost items
Posted: Mon Sep 24, 2018 6:22 am
by Klonan
The probability in the recipe is not dependant, which means in some cases you can end up with no items being produced.
I don't see anything here to indicate a bug
Re: [0.16.51] Uranium processing + productivity modules = lost items
Posted: Mon Sep 24, 2018 11:20 am
by Merssedes
Klonan wrote: ↑Mon Sep 24, 2018 6:22 am
The probability in the recipe is not dependant, which means in some cases you can end up with no items being produced.
<...>
How does it "end up with no items being produced"?
And even if it's a case, what about tests with more then expected output (1, 4, 5, 6 and 10)? I replaced centrifuge before each test to "cleanup" state.
Re: [0.16.51] Uranium processing + productivity modules = lost items
Posted: Mon Sep 24, 2018 1:44 pm
by eradicator
Merssedes wrote: ↑Mon Sep 24, 2018 11:20 am
Klonan wrote: ↑Mon Sep 24, 2018 6:22 am
The probability in the recipe is not dependant, which means in some cases you can end up with no items being produced.
<...>
How does it "end up with no items being produced"?
And even if it's a case, what about tests with more then expected output (1, 4, 5, 6 and 10)? I replaced centrifuge before each test to "cleanup" state.
The centrifuge produces one U238with a probability of 99.3%, so there's a 0.7% chance that it produces nothing.
It also has a 0.7% chance to produce U235, with a 99.3% chance of producing nothing.
These two steps are independant of each other, so some cycles produces 2*nothing, and some cycles produce 1*U238 AND 1*U235.
There's an
old topic requesting moddability for not-independant probabilities, but it never got any dev answers.
Re: [0.16.51] Uranium processing + productivity modules = lost items
Posted: Mon Sep 24, 2018 3:36 pm
by Merssedes
OK. Tests on centrifuge without productivity: 199, 200, 200, 200, 201 out of 200 expected (2k input).
Am I right that in case of mining drills with multiple ores it resets productivity bar as "change of recipe"?