[nick49488171]

With a non-sequential generative approach perhaps the RAM cache misses could be grouped together and swapped on a when available/when needed prioritized bases.