28#include <boost/mpi/collectives/gather.hpp>
29#include <boost/mpi/communicator.hpp>
31namespace Utils::Mpi::detail {
34int size_and_offset(std::vector<int> &
sizes, std::vector<int> &
displ,
35 int n_elem,
const boost::mpi::communicator &comm,
37 auto const world_size =
static_cast<unsigned int>(comm.size());
38 sizes.resize(world_size);
39 displ.resize(world_size);
47 for (std::size_t i = 0; i <
sizes.size(); i++) {
55inline void size_and_offset(
int n_elem,
const boost::mpi::communicator &comm,
cudaStream_t stream[1]
CUDA streams for parallel computing on CPU and GPU.