Usually programs of LIFE operate two 2-dimensional arrays of booleans calculating neighbours for every cell in every generation by walking through nearest 8 cells for each cell on map and modifying second array according to LIFE rules (alive cell will live if it has 2 or 3 alive neighbours, dead cell will be alive again if it has exactly 3 alive neighbours, otherwise cell must be dead) as a buffer to copy it to primary array later at the end of calculation for each generation. My algorithm uses precalculated number of neighbours for each cell walking through nearest 8 cell ONLY if something changed inside cell (transition from dead to alive or from alive to dead) to decrease or increase number of neighbours in 8 nearest cells respectively.