Skip to content

Increased performance on routine compute_mmn in pw2wannier90 program.

Nicola Spallanzani requested to merge nicspalla/q-e-old:devel-compute_mmn into develop

Inside the routine compute_mmn of the code pw2wannier90 there was a call to the mp_sum routine for every single element of the matrix Mkb. This led to a loss of performance for big matrices. Now the mp_sum routine is moved out the double loop of the indexes and it is performed by column. This increased the performances.

Merge request reports