nbnxn_cuda_types.h File Reference

#include "gromacs/gpu_utils/cuda_arch_utils.cuh"
#include "gromacs/gpu_utils/cudautils.cuh"
#include "gromacs/mdlib/nbnxn_consts.h"
#include "gromacs/mdlib/nbnxn_gpu_types_common.h"
#include "gromacs/mdlib/nbnxn_pairlist.h"
#include "gromacs/mdtypes/interaction_const.h"
#include "gromacs/timing/gpu_timing.h"

Include dependency graph for nbnxn_cuda_types.h:

This graph shows which files directly or indirectly include this file:

Description

Data types used internally in the nbnxn_cuda module.

Author

Szilárd Páll pall..nosp@m.szil.nosp@m.ard@g.nosp@m.mail.nosp@m..com

Classes
struct	nb_staging
	Staging area for temporary data downloaded from the GPU. More...
struct	cu_atomdata
	Nonbonded atom data - both inputs and outputs. More...
struct	cu_nbparam
	Parameters required for the CUDA nonbonded calculations. More...
struct	cu_plist
	Pair list data. More...
struct	gmx_nbnxn_cuda_t
	Main data structure for CUDA nonbonded force calculations. More...

Macros
#define	GMX_NBNXN_PRUNE_KERNEL_J4_CONCURRENCY   4
	Macro definining default for the prune kernel's j4 processing concurrency. More...

Typedefs
typedef struct nbnxn_gpu_timers_t	cu_timers_t
	Typedef of actual timer type. More...

Enumerations
enum	eelCu {   eelCuCUT, eelCuRF, eelCuEWALD_TAB, eelCuEWALD_TAB_TWIN,   eelCuEWALD_ANA, eelCuEWALD_ANA_TWIN, eelCuNR }
	Electrostatic CUDA kernel flavors. More...
enum	evdwCu {   evdwCuCUT, evdwCuCUTCOMBGEOM, evdwCuCUTCOMBLB, evdwCuFSWITCH,   evdwCuPSWITCH, evdwCuEWALDGEOM, evdwCuEWALDLB, evdwCuNR }
	VdW CUDA kernel flavors. More...

Variables
const int	c_cudaPruneKernelJ4Concurrency = 4
	Default for the prune kernel's j4 processing concurrency. More...
static const int	c_numClPerSupercl = c_nbnxnGpuNumClusterPerSupercluster
	number of clusters per supercluster.
static const int	c_clSize = c_nbnxnGpuClusterSize
	cluster size = number of atoms per cluster.

Macro Definition Documentation

#define GMX_NBNXN_PRUNE_KERNEL_J4_CONCURRENCY   4

Macro definining default for the prune kernel's j4 processing concurrency.

The GMX_NBNXN_PRUNE_KERNEL_J4_CONCURRENCY macro allows compile-time override.

Typedef Documentation

typedef struct nbnxn_gpu_timers_t cu_timers_t

Typedef of actual timer type.

Enumeration Type Documentation

enum eelCu

Electrostatic CUDA kernel flavors.

Types of electrostatics implementations available in the CUDA non-bonded force kernels. These represent both the electrostatics types implemented by the kernels (cut-off, RF, and Ewald - a subset of what's defined in enums.h) as well as encode implementation details analytical/tabulated and single or twin cut-off (for Ewald kernels). Note that the cut-off and RF kernels have only analytical flavor and unlike in the CPU kernels, the tabulated kernels are ATM Ewald-only.

The row-order of pointers to different electrostatic kernels defined in nbnxn_cuda.cu by the nb_*_kfunc_ptr function pointer table should match the order of enumerated types below.

enum evdwCu

VdW CUDA kernel flavors.

The enumerates values correspond to the LJ implementations in the CUDA non-bonded kernels.

The column-order of pointers to different electrostatic kernels defined in nbnxn_cuda.cu by the nb_*_kfunc_ptr function pointer table should match the order of enumerated types below.

Variable Documentation

const int c_cudaPruneKernelJ4Concurrency = 4

Default for the prune kernel's j4 processing concurrency.

Initialized using the GMX_NBNXN_PRUNE_KERNEL_J4_CONCURRENCY macro which allows compile-time override.