User talk:Evil4Zerggin/Advanced cannon notes

Here's some 5-clip setups after i've figured out you can attach autoloaders from all directions(!)(and a 2 clip, 1 feeder setup because it stacks somewhat nicely assuming you have a central shaft):

2 clips, 1 feeder
Sustains full rate of fire at 64 or more autoloaders. Good 3x3 packing assuming you have a central shaft to attach the autoloaders to, but sustaining a full fire rate through it is difficult. $$\frac{\sqrt{2}}{4^{0.75}} = 0.5$$

5 clips, 2 feeder
Sustains full rate of fire at 25 or more autoloaders. Better theoretical efficiency than a 4 clip, 2 feeder setup, but even worse packing.

$$\frac{\sqrt{5}}{7^{0.75}} \approx 0.470$$

5 clips, 4 feeder
Sustains full rate of fire at 2 or more autoloaders. Somewhat lousy efficiency for its packing, but it still performs better than 2 clips, 2 feeders.

$$\frac{\sqrt{5}}{9^{0.75}} \approx 0.430$$

Overall, per-block efficiency seems to increase on belt-fed setups while it decreases on conventional autoloader setups based on how many ammo input feeders you have. Perhaps a better approach would be using the setup granting sustained fire for the least number of feeders?

the 2 clip, 1 feeeder setup achieves break-even with 0-clip autoloaders at 256 blocks - exactly the point at which full fire rate is achieved. 3/4/5 clips, 1 feeder has an efficiency of ~0.518, ~0.521 and ~0.519 - note the 5 clip setup actually is worse than the 4 clip, 1 feeder setup! RA2lover (talk) 13:06, 25 November 2015 (UTC)


 * Per block efficiency eventually starts decreasing on belt-fed setups past 7 feeders IIRC, but by then you've already run into packing issues and a lack of available faces anyways. I'll add 2 + 1 to the main page. --Evil4Zerggin (talk) 07:43, 26 November 2015 (UTC)

It also turns out 2+1 autoloader setups pack very well on 2x2 and 4x1 spaces. it could be a better overall setup than 3+2 assuming you don't need an indefinitely constant rate of fire or extremely long turrets with odd symmetry. RA2lover (talk) 17:17, 27 November 2015 (UTC)