amlc

sch1.py

for yo in range(128):
  for xo in range(128):
    C[yo*8:yo*8+8][xo*8:xo*8+8] = 0
    for ko in range(128):
      for yi in range(8):
        for xi in range(8):
          for ki in range(8):
            C[yo*8+yi][xo*8+xi] += 
              A[ko*8+ki][yo*8+yi] * B[ko*8+ki][xo*8+xi]

sch2.py

for yo in range(64):
  for xo in range(64):
    C[yo*16:yo*16+16][xo*16:xo*16+16] = 0
    for ko in range(64):
      for yi in range(16):
        for xi in range(16):
          for ki in range(16):
            C[yo*16+yi][xo*16+xi] += 
              A[ko*16+ki][yo*16+yi] * B[ko*16+ki][xo*16+xi]

sch3.py

for xo in range(64):
  for yo in range(64):
    C[yo*16:yo*16+16][xo*16:xo*16+16] = 0
    for ko in range(64):
      for ki in range(16):
        for yi in range(16):
          for xi in range(16):
            C[yo*16+yi][xo*16+xi] += 
              A[ko*16+ki][yo*16+yi] * B[ko*16+ki][xo*16+xi]

vanilla.py

for y in range(1024):
  for x in range(1024):
    C[y][x] = 0
    for k in range(1024):
      C[y][x] += A[k][y] * B[k][x]