This is my 4th real design iteration of sum solve after a few mediocre attempts, as is typically the case when trying to sum optimise a new puzzle. 

Design #2 found arm 4, but originally had it access dispersal + calcifier + bonder + dupe and so was unable to output, significantly slowing down the solve (or adding 20g). I then designed around that arm getting to output instead of accessing dispersal. 

Design #3 was a trackloop of all 3 different length arms handling dispersal. To get to my current design I essentially just used trial and error and bashed my head against the wall playing around with dispersal location since that glyph SUCKS ASS. My initial solves were all more fast and expensive - I could not get slow and cheap to work at all due to just how clunky the dispersal glyph was for me.

The 3 arms on this solve's trackloop end up each having 2 distinct things to do, which feels like a nice distribution of workload, although I did not design with this in mind at all - I just noticed it afterwards :P. 
(1: Dispersal Quint + Earth, 2: Air + Water, 3: Fire + Output Quint)

It feels like you could save the second calcifier at the bottom with some better routing/layout as it's only used once, and that alone gives sub-300 here. The rest of the solve feels pretty alright, although maybe a 16-cycle loop could be improved upon? There's a little bit of deadtime due to the constrictive nature of 3 arms on a 4 long trackloop. 

I would guess that the winner be around 290 sum; I certainly can't see it getting any lower than 280 unless I've missed some secret sauce.