<k​ayabanerve:matrix.org> 64-input FCMPs being slower than individual FCMPs sounds wrong. They should have the same size MSM yet the former should resolve a bit faster as the MSM has complexity O(n / log(n)).

<a​rticmine:monero.social> When it comes to scaling large transactions the penalty goes as the square of the percentage block weight increase. So at the minimum penalty free zone of 1000000 bytes a TX with a weight of 160000 bytes pays 256x the penalty of a 10000 byte transaction and should be required to satisfy at least 256x the minimum node relay fee.

<a​rticmine:monero.social> Now with linear fee scaling as the long term median increases the fee differential would only be 16x if the long term median were 16000000 bytes (a 16x increase)

<a​rticmine:monero.social> The above alone favors say eight 10000 byte transactions over one 80000 byte transaction unless we have a corresponding increase in adoption