Lollie, do you mind explaining why you removed the 16G solver? Just curious why 8G would be faster. I've got some Vega FEs and a Radeon Pro Duo I wanted to throw at this, remember you had the 16G solver back on version 0.8.3.
It was only removed for 0.9 but brought back for 0.9.1 - the reason was that the 8G 0.9 solver was quicker then 16G 0.8.8 (and earlier). 0.9.2 has a bug that may cause to load wrong kernel (8G instead of 16G) in Windows on FE - will be fixed with 0.9.3
Is there possible to implement Cuckaroo29 BFC algorithm?
I am redoing my C29 codes atm. Hope to have it flexible enough to add Cuckarood-29 (Current Grin-29), Cuckaroom-29 (next gen Grin from January), Cuckaroo-29 (BFC, Swap) and Cuckaroo-30 (Cortex) one by one... not yet sure about the order though ^^