[0.13.0] Assembling machine 3 - production speed ratios.

[matrix47]

I've reported this a while back, but in the wrong forum: viewtopic.php?f=18&t=20334
The bug is that for AM3 one copper cable assembler is not enough for eight advanced circuit assemblers. For AM1 or AM2 this ratio is perfect (all assemblers are working constantly).
You can check the discussion there, but seems like for AM3 production takes 1 tick longer than expected.
Here's youtube video demonstrating the issue (it was in 0.12, but I verified that 0.13 behavior is the same). https://www.youtube.com/watch?v=bnFqaZrs5Zk
When 0.13 came out, I checked that this is still not working as I expect it to. Thanks for the sandbox mode, things like this are now quite easy to test.

[kovarex]

Not a bug.

[matrix47]

could you please explain?

[DaveMcW]

It's clearly a floating point bug.

60 * 0.5 / 0.5 = 60 good!
60 * 0.5 / 0.75 = 40 good!
60 * 0.5 / 1.25 = 24.000001 = 25 boo!

But maybe this thread is supposed to go in "won't fix".

[matrix47]

It's clearly a floating point bug.

60 * 0.5 / 0.5 = 60 good!
60 * 0.5 / 0.75 = 40 good!
60 * 0.5 / 1.25 = 24.000001 = 25 boo!

But maybe this thread is supposed to go in "won't fix".

Yeah, this doesn't look to me like a "Bug that's actually a feature".
That's why I'd love an explanation, or a guide, how to figure out perfect ratios (in this particular case, it seems to be 10:77).
It also feels somewhat disturbing to me, that after upgrading a setup to a higher level assemblers, some of them stop working at a 100%. Though I understand it is inevitable if we have discrete time.
And in this case I was hoping that rounding error can be fixed. After all, almost 4% of production speed is sacrificed here for nothing.

[matrix47]

apparently, this has been fixed in 0.13.5
viewtopic.php?f=11&t=28128