What a mess... ABC wants CTOR without proving how it improves perfomance. SV wants 128 MB without solving first bottlenecks. Yes, BU/XT is the way to go.
You don't need to solve bottlenecks to increase the cap. Miners will use the software as they see fit, and the released client doesn't need to be optimized to match the cap.
Focusing time and energy on anything but the bottleneck is exactly how many people waste their lifetime. Focusing on minor issues ignoring the real threats to their existence.