|
core
Shared MPI block decomposition + asynchronous ghost-layer exchange (header-only C++20)
|
Classes | |
| struct | NonDeduced |
Functions | |
| bool | gpuAwareMpi () |
| Whether to hand DEVICE pointers straight to MPI (GPU-aware MPI) instead of host-staging. | |
|
inline |
Whether to hand DEVICE pointers straight to MPI (GPU-aware MPI) instead of host-staging.
Gated on an env var (read once) rather than MPIX_Query_cuda_support(), which under-reports on some stacks.
Definition at line 29 of file grid_halo.hpp.
Referenced by peclet::core::halo::GridHalo< T >::exchange(), peclet::core::halo::ParticleHalo< Dim >::forward(), peclet::core::amr::DistributedGatherHalo< Dim, Bits >::gather(), and peclet::core::halo::ParticleHalo< Dim >::reverse().