basically this is a period 4 solution that uses 3-arm trackloops at period 12 in two sections to save cycles by avoiding regrabs: disassembling the input, and turning silver to quicksilver.

im honestly not sure what the min cycles on this but i hope its dubious enough that all of top 50 isnt there. the quicksilver bonding feels inefficient, i'm sure there's some latency to drop there. the start is rough, finding a way to outputin the other direction would prob be the biggest individual cycles drop on this approach

i feel like i probably lost a cycle or two in the mirroring process