optimization - How to make my Fortran loop faster with BLAS - Matrix vector multiplication -
i optimize speed following fortran code
do ii = 1, n (:,:) = (:,:) + c (ii) * b (:,:, ii ) enddo
with a(m,m)
dimension , b(m,m)
dimension.
i thinking of use blas
do jj=1,m call zgemm('n', 'n', 1, m, n, cone, c(:), cone, b (jj,:, : ),& n, czero, a(:,:), cone ) enddo
but not efficient still have loop. possible use increment , how?
in case n > m
Comments
Post a Comment