Memory consumption info strange

[kostrzewa]

Memory consumption:
        OperatorFactory:        0.05 Gb
        RandomVector:   0.01 Gb
        Perambulator:   85.08 Gb
        DiagramParts:
                Q0:     0.08 Gb
                Q1:     0.10 Gb
                Q2L:    0.08 Gb
                Q2V:    0.08 Gb
        trQ1Q1: 0.00 Gb
        trQ0Q2: 0.00 Gb
        trQ1:   0.00 Gb
        Diagrams:
        Perambulators initialised
        Random vectors initialised
        Meson operators initialised

85 Gb for 6 perambulators, each of which is 2.4 Gb?

[kostrzewa]

Are you sure that the factor of peram_construct.nb_entities in global_data.cpp is necessary? The loops should account for all random vectors of all flavours, so peram_matrix_size_sum should already be correct. Note that there is potential for an integer overflow here, however. (for large lattices with many eigenvectors). peram_matrix_size_sum should probably be size_t.

[maowerner]

I did not sufficiently check the memory consumption output. I am fairly convinced, the whole peram_construct is useless. But I currently don't have enough time to simplify it. There are dependencies left and right in the global data...

[kostrzewa]

This is fixed for now. I'm working through global_data to understand it all.