|
| communications.h |
| Communications to send and receive sites from other nodes using MPI with GPU.
|
|
| communications_reduced.h |
|
| cpu_geometry.h |
| This file contains macros to load elements of single sites of a field.
|
|
| geometry_check.h |
| Validity checks on the geometry of spinor fields that are passed as function parameters. Use these macros to make the code more compact. These are compiled to empty lines, if CHECK_SPINOR_MATCHING is inactive.
|
|
| geometry_descriptor.h |
| Geometry descriptor struct, that contains all necessary information for multi-node/ multi-GPU calculations.
|
|
| geometry_fuse.h |
| FIXME: Add docs.
|
|
| geometry_gpu_init.h |
| Initialization functions, that determine all important parameters of the geometry, so that communications and operations can be completed correctly.
|
|
| geometry_indexing.h |
| Finding indices from coordinates and the other way around.
|
|
| geometry_init.h |
| Initialization functions, that determine all important parameters of the geometry, so that communications and operations can be completed correctly.
|
|
| geometry_maskstate.h |
| Mask structs that can give information on whether neighbors of a given site are located in the buffer and require communications first or not.
|
|
| geometry_omp.h |
| This file contains useful macros that perform OpenMP reduction operations and are necessary to define iterations over sites on the lattice.
|
|
| geometry_type.h |
| This file contains a geometry type for the new geometry.
|
|
| gpu_affinity.h |
|
| gpu_geometry.h |
|
| hr_sendrecv.h |
| Header for custom communications, configurable over compilation flags.
|
|
| new_geometry.h |
| Functions needed for the new geometry implementation that will replace the current geometry in the future.
|
|
| read_clover.h |
|
| setup.h |
| Setup and finalize to run at the beginning and the end of every program.
|
|
| strided_reads_gpu.h |
|