Hi,

I have sparse matrix of MATMPIAIJ format that needs to be multiplied with a
series of vectors and the resultant set of vectors are to be stored into a
matrix. I can do that in two ways:

1. Perform series of MatMult operations and store the resultant vectors in
a MATMPIDENSE matrix; or
2. Store the vectors in a MATMPIDENSE matrix and perform a MatMatMult
operation.

I want to know which of these two approaches will be more efficient and if
there is a crossover in terms of the number of vectors for which one
approach starts performing better than the other.

Thanks,
Bikash

-- 
Bikash S. Kanungo
PhD Student
Computational Materials Physics Group
Mechanical Engineering
University of Michigan

Reply via email to