According to Apple, to perform multiplication of matrices in a vector processing system, partial products are obtained by dot multiplication of vector registers containing multiple copies of elements ...

This is because currently the optimization in AOCL (which includes matrix reordering) isn't performed properly. Future versions of AOCL will hopefully resolve this issue. nproc could be set to 24, or ...

In every experiment, the number of nodes is constant; relatively small (usually lower than 100); the vehicle capacity is constant; the location of nodes is i.i.d uniformly-distributed within the unit ...