We now show how the parallel matrix-vector multiplication
and rank-1 update can be used to implement matrix-matrix multiplication
and other matrix-matrix operations.
For simplicity, we concentrate on the case
C = A B , where all three matrices are 
.
In all our explanations, we will use the following partitionings:
with 
and 
where 
represents the j th column of matrix X .
Also,
where 
represents the i th row of matrix X .