134 lines
5.1 KiB
C++
134 lines
5.1 KiB
C++
/***************************************************************************
|
|
mdpd_ext.cpp
|
|
-------------------
|
|
Trung Dac Nguyen (U Chicago)
|
|
|
|
Functions for LAMMPS access to mdpd acceleration routines.
|
|
|
|
__________________________________________________________________________
|
|
This file is part of the LAMMPS Accelerator Library (LAMMPS_AL)
|
|
__________________________________________________________________________
|
|
|
|
begin : December 2023
|
|
email : ndactrung@gmail.com
|
|
***************************************************************************/
|
|
|
|
#include <iostream>
|
|
#include <cassert>
|
|
#include <cmath>
|
|
|
|
#include "lal_mdpd.h"
|
|
|
|
using namespace std;
|
|
using namespace LAMMPS_AL;
|
|
|
|
static MDPD<PRECISION,ACC_PRECISION> MDPDMF;
|
|
|
|
// ---------------------------------------------------------------------------
|
|
// Allocate memory on host and device and copy constants to device
|
|
// ---------------------------------------------------------------------------
|
|
int mdpd_gpu_init(const int ntypes, double **cutsq,
|
|
double **host_A_att, double **host_B_rep,
|
|
double **host_gamma, double **host_sigma,
|
|
double **host_cut, double **host_cut_r,
|
|
double *special_lj, const int inum,
|
|
const int nall, const int max_nbors, const int maxspecial,
|
|
const double cell_size, int &gpu_mode, FILE *screen) {
|
|
MDPDMF.clear();
|
|
gpu_mode=MDPDMF.device->gpu_mode();
|
|
double gpu_split=MDPDMF.device->particle_split();
|
|
int first_gpu=MDPDMF.device->first_device();
|
|
int last_gpu=MDPDMF.device->last_device();
|
|
int world_me=MDPDMF.device->world_me();
|
|
int gpu_rank=MDPDMF.device->gpu_rank();
|
|
int procs_per_gpu=MDPDMF.device->procs_per_gpu();
|
|
|
|
MDPDMF.device->init_message(screen,"mdpd",first_gpu,last_gpu);
|
|
|
|
bool message=false;
|
|
if (MDPDMF.device->replica_me()==0 && screen)
|
|
message=true;
|
|
|
|
if (message) {
|
|
fprintf(screen,"Initializing Device and compiling on process 0...");
|
|
fflush(screen);
|
|
}
|
|
|
|
int init_ok=0;
|
|
if (world_me==0)
|
|
init_ok=MDPDMF.init(ntypes, cutsq, host_A_att, host_B_rep, host_gamma, host_sigma,
|
|
host_cut, host_cut_r, special_lj, inum, nall, max_nbors,
|
|
maxspecial, cell_size, gpu_split, screen);
|
|
|
|
MDPDMF.device->world_barrier();
|
|
if (message)
|
|
fprintf(screen,"Done.\n");
|
|
|
|
for (int i=0; i<procs_per_gpu; i++) {
|
|
if (message) {
|
|
if (last_gpu-first_gpu==0)
|
|
fprintf(screen,"Initializing Device %d on core %d...",first_gpu,i);
|
|
else
|
|
fprintf(screen,"Initializing Devices %d-%d on core %d...",first_gpu,
|
|
last_gpu,i);
|
|
fflush(screen);
|
|
}
|
|
if (gpu_rank==i && world_me!=0)
|
|
init_ok=MDPDMF.init(ntypes, cutsq, host_A_att, host_B_rep, host_gamma, host_sigma,
|
|
host_cut, host_cut_r, special_lj, inum, nall, max_nbors,
|
|
maxspecial, cell_size, gpu_split, screen);
|
|
|
|
MDPDMF.device->serialize_init();
|
|
if (message)
|
|
fprintf(screen,"Done.\n");
|
|
}
|
|
if (message)
|
|
fprintf(screen,"\n");
|
|
|
|
if (init_ok==0)
|
|
MDPDMF.estimate_gpu_overhead();
|
|
return init_ok;
|
|
}
|
|
|
|
void mdpd_gpu_clear() {
|
|
MDPDMF.clear();
|
|
}
|
|
|
|
int ** mdpd_gpu_compute_n(const int ago, const int inum_full, const int nall,
|
|
double **host_x, int *host_type, double *sublo,
|
|
double *subhi, tagint *tag, int **nspecial,
|
|
tagint **special, const bool eflag, const bool vflag,
|
|
const bool eatom, const bool vatom, int &host_start,
|
|
int **ilist, int **jnum, const double cpu_time, bool &success,
|
|
double **host_v, const double dtinvsqrt,
|
|
const int seed, const int timestep,
|
|
double *boxlo, double *prd) {
|
|
return MDPDMF.compute(ago, inum_full, nall, host_x, host_type, sublo,
|
|
subhi, tag, nspecial, special, eflag, vflag, eatom,
|
|
vatom, host_start, ilist, jnum, cpu_time, success,
|
|
host_v, dtinvsqrt, seed, timestep, boxlo, prd);
|
|
}
|
|
|
|
void mdpd_gpu_compute(const int ago, const int inum_full, const int nall,
|
|
double **host_x, int *host_type, int *ilist, int *numj,
|
|
int **firstneigh, const bool eflag, const bool vflag,
|
|
const bool eatom, const bool vatom, int &host_start,
|
|
const double cpu_time, bool &success, tagint *tag,
|
|
double **host_v, const double dtinvsqrt,
|
|
const int seed, const int timestep,
|
|
const int nlocal, double *boxlo, double *prd) {
|
|
MDPDMF.compute(ago, inum_full, nall, host_x, host_type, ilist, numj,
|
|
firstneigh, eflag, vflag, eatom, vatom, host_start, cpu_time, success,
|
|
tag, host_v, dtinvsqrt, seed, timestep, nlocal, boxlo, prd);
|
|
}
|
|
|
|
void mdpd_gpu_get_extra_data(double *host_rho) {
|
|
MDPDMF.get_extra_data(host_rho);
|
|
}
|
|
|
|
double mdpd_gpu_bytes() {
|
|
return MDPDMF.host_memory_usage();
|
|
}
|
|
|
|
|