Currently we do something really ugly and lazy to prevent it from using up all memory. We just clear the entire dp once it hits its memory limit (of around 1GB) and start filling it up again.
This causes the next AI move (after clearing the dp) to take longer to calculate:
Currently we do something really ugly and lazy to prevent it from using up all memory. We just clear the entire
dponce it hits its memory limit (of around 1GB) and start filling it up again.This causes the next AI move (after clearing the
dp) to take longer to calculate:We should think of using some Replacement Strategies instead: https://en.wikipedia.org/wiki/Transposition_table#Replacement_strategies