Go backward to Memory Access Analysis Go up to Top Go forward to Memory Access

Block Matrix Multiplication

$$

A = A11 A12 A21 A22 B = B11 B12 B21 B22

C = C11 C12 C21 C22 = A11 B11 + A12 B21 A11 B12 + A12 B22 A21 B11 + A22 B21 A21 B12 + A22 B22

• Submatrices of size $k\; xk$.
• $p=(n/k)2$ processes.
• Per process $M=5k2$ memory area used.
• Per process $C=2k3$ computations.
• Ratio $C/M=2/5\; k=2/5\; n/p$.