kwz optimisation idea

kwztile.py

# original code:
for tile_offset_y in range(0, 240, 128):
  for tile_offset_x in range(0, 320, 128):
    # each large tile is made of 8 * 8 small tiles
    for sub_tile_offset_y in range(0, 128, 8):
      y = tile_offset_y + sub_tile_offset_y
      # if the tile falls off the bottom of the frame, jump to the next large tile
      if y >= 240: break

      for sub_tile_offset_x in range(0, 128, 8):
        x = tile_offset_x + sub_tile_offset_x
        # if the tile falls off the right of the frame, jump to the next small tile row
        if x >= 320: 
          break
        # ... decode tile @ (x, y)
        
# precompute this as table 
KWZ_TILE_COORDS = []

for tile_offset_y in range(0, 240, 128):
  for tile_offset_x in range(0, 320, 128):
    for sub_tile_offset_y in range(0, 128, 8):
      y = tile_offset_y + sub_tile_offset_y
      if y >= 240: break
      for sub_tile_offset_x in range(0, 128, 8):
        x = tile_offset_x + sub_tile_offset_x
        if x >= 320: 
          break
        KWZ_TILE_OFFSETS.append((x, y))

# replace nested loop
for (x, y) in KWZ_TILE_COORDS:
  # ... decode tile @ (x, y)