jianmu
/
lammps
forked from lijiext/lammps
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
lammps/lib/gpu/lal_sw.cpp

266 lines
8.8 KiB
C++
Raw Blame History

/***************************************************************************
sw.cpp
-------------------
W. Michael Brown (ORNL)
Class for acceleration of the sw pair style.
__________________________________________________________________________
This file is part of the LAMMPS Accelerator Library (LAMMPS_AL)
__________________________________________________________________________
begin : Tue March 26, 2013
email : brownw@ornl.gov
***************************************************************************/
#if defined(USE_OPENCL)
#include "sw_cl.h"
#elif defined(USE_CUDART)
const char *sw=0;
#else
#include "sw_cubin.h"
#endif
#include "lal_sw.h"
#include <cassert>
using namespace LAMMPS_AL;
#define SWT SW<numtyp, acctyp>
extern Device<PRECISION,ACC_PRECISION> device;
template <class numtyp, class acctyp>
SWT::SW() : BaseThree<numtyp,acctyp>(), _allocated(false) {
}
template <class numtyp, class acctyp>
SWT::~SW() {
clear();
}
template <class numtyp, class acctyp>
int SWT::bytes_per_atom(const int max_nbors) const {
return this->bytes_per_atom_atomic(max_nbors);
}
template <class numtyp, class acctyp>
int SWT::init(const int ntypes, const int nlocal, const int nall, const int max_nbors,
const double cell_size, const double gpu_split, FILE *_screen,
int* host_map, const int nelements, int*** host_elem2param, const int nparams,
const double* epsilon, const double* sigma,
const double* lambda, const double* gamma,
const double* costheta, const double* biga,
const double* bigb, const double* powerp,
const double* powerq, const double* cut, const double* cutsq)
{
int success;
success=this->init_three(nlocal,nall,max_nbors,0,cell_size,gpu_split,
_screen,sw,"k_sw","k_sw_three_center",
"k_sw_three_end","k_sw_short_nbor");
if (success!=0)
return success;
// If atom type constants fit in shared memory use fast kernel
int lj_types=ntypes;
shared_types=false;
int max_shared_types=this->device->max_shared_types();
if (lj_types<=max_shared_types && this->_block_size>=max_shared_types) {
lj_types=max_shared_types;
shared_types=true;
}
_lj_types=lj_types;
_nparams = nparams;
_nelements = nelements;
UCL_H_Vec<numtyp4> dview(nparams,*(this->ucl_device),
UCL_WRITE_ONLY);
for (int i=0; i<nparams; i++) {
dview[i].x=(numtyp)0;
dview[i].y=(numtyp)0;
dview[i].z=(numtyp)0;
dview[i].w=(numtyp)0;
}
// pack coefficients into arrays
sw1.alloc(nparams,*(this->ucl_device),UCL_READ_ONLY);
for (int i=0; i<nparams; i++) {
dview[i].x=static_cast<numtyp>(epsilon[i]);
dview[i].y=static_cast<numtyp>(sigma[i]);
dview[i].z=static_cast<numtyp>(lambda[i]);
dview[i].w=static_cast<numtyp>(gamma[i]);
}
ucl_copy(sw1,dview,false);
sw1_tex.get_texture(*(this->pair_program),"sw1_tex");
sw1_tex.bind_float(sw1,4);
sw2.alloc(nparams,*(this->ucl_device),UCL_READ_ONLY);
for (int i=0; i<nparams; i++) {
dview[i].x=static_cast<numtyp>(biga[i]);
dview[i].y=static_cast<numtyp>(bigb[i]);
dview[i].z=static_cast<numtyp>(powerp[i]);
dview[i].w=static_cast<numtyp>(powerq[i]);
}
ucl_copy(sw2,dview,false);
sw2_tex.get_texture(*(this->pair_program),"sw2_tex");
sw2_tex.bind_float(sw2,4);
sw3.alloc(nparams,*(this->ucl_device),UCL_READ_ONLY);
for (int i=0; i<nparams; i++) {
double sw_cut = cut[i];
double sw_cutsq = cutsq[i];
if (sw_cutsq>=sw_cut*sw_cut)
sw_cutsq=sw_cut*sw_cut-1e-4;
dview[i].x=static_cast<numtyp>(sw_cut);
dview[i].y=static_cast<numtyp>(sw_cutsq);
dview[i].z=static_cast<numtyp>(costheta[i]);
dview[i].w=(numtyp)0;
}
ucl_copy(sw3,dview,false);
sw3_tex.get_texture(*(this->pair_program),"sw3_tex");
sw3_tex.bind_float(sw3,4);
UCL_H_Vec<int> dview_elem2param(nelements*nelements*nelements,
*(this->ucl_device), UCL_WRITE_ONLY);
elem2param.alloc(nelements*nelements*nelements,*(this->ucl_device),
UCL_READ_ONLY);
for (int i = 0; i < nelements; i++)
for (int j = 0; j < nelements; j++)
for (int k = 0; k < nelements; k++) {
int idx = i*nelements*nelements+j*nelements+k;
dview_elem2param[idx] = host_elem2param[i][j][k];
}
ucl_copy(elem2param,dview_elem2param,false);
UCL_H_Vec<int> dview_map(lj_types, *(this->ucl_device), UCL_WRITE_ONLY);
for (int i = 0; i < ntypes; i++)
dview_map[i] = host_map[i];
map.alloc(lj_types,*(this->ucl_device), UCL_READ_ONLY);
ucl_copy(map,dview_map,false);
_allocated=true;
this->_max_bytes=sw1.row_bytes()+sw2.row_bytes()+sw3.row_bytes()+
map.row_bytes()+elem2param.row_bytes();
return 0;
}
template <class numtyp, class acctyp>
void SWT::clear() {
if (!_allocated)
return;
_allocated=false;
sw1.clear();
sw2.clear();
sw3.clear();
map.clear();
elem2param.clear();
this->clear_atomic();
}
template <class numtyp, class acctyp>
double SWT::host_memory_usage() const {
return this->host_memory_usage_atomic()+sizeof(SW<numtyp,acctyp>);
}
#define KTHREADS this->_threads_per_atom
#define JTHREADS this->_threads_per_atom
// ---------------------------------------------------------------------------
// Calculate energies, forces, and torques
// ---------------------------------------------------------------------------
template <class numtyp, class acctyp>
void SWT::loop(const bool _eflag, const bool _vflag, const int evatom) {
// Compute the block size and grid size to keep all cores busy
int BX=this->block_pair();
int eflag, vflag;
if (_eflag)
eflag=1;
else
eflag=0;
if (_vflag)
vflag=1;
else
vflag=0;
// build the short neighbor list
int ainum=this->_ainum;
int nbor_pitch=this->nbor->nbor_pitch();
int GX=static_cast<int>(ceil(static_cast<double>(ainum)/
(BX/this->_threads_per_atom)));
this->k_short_nbor.set_size(GX,BX);
this->k_short_nbor.run(&this->atom->x, &sw3, &map, &elem2param, &_nelements,
&this->nbor->dev_nbor, &this->_nbor_data->begin(),
&this->dev_short_nbor, &ainum,
&nbor_pitch, &this->_threads_per_atom);
// this->_nbor_data == nbor->dev_packed for gpu_nbor == 0 and tpa > 1
// this->_nbor_data == nbor->dev_nbor for gpu_nbor == 1 or tpa == 1
ainum=this->ans->inum();
nbor_pitch=this->nbor->nbor_pitch();
GX=static_cast<int>(ceil(static_cast<double>(this->ans->inum())/
(BX/this->_threads_per_atom)));
this->time_pair.start();
this->k_pair.set_size(GX,BX);
this->k_pair.run(&this->atom->x, &sw1, &sw2, &sw3,
&map, &elem2param, &_nelements,
&this->nbor->dev_nbor, &this->_nbor_data->begin(),
&this->dev_short_nbor,
&this->ans->force, &this->ans->engv,
&eflag, &vflag, &ainum, &nbor_pitch,
&this->_threads_per_atom);
BX=this->block_size();
GX=static_cast<int>(ceil(static_cast<double>(this->ans->inum())/
(BX/(KTHREADS*JTHREADS))));
this->k_three_center.set_size(GX,BX);
this->k_three_center.run(&this->atom->x, &sw1, &sw2, &sw3,
&map, &elem2param, &_nelements,
&this->nbor->dev_nbor, &this->_nbor_data->begin(),
&this->dev_short_nbor,
&this->ans->force, &this->ans->engv, &eflag, &vflag, &ainum,
&nbor_pitch, &this->_threads_per_atom, &evatom);
Answer<numtyp,acctyp> *end_ans;
#ifdef THREE_CONCURRENT
end_ans=this->ans2;
#else
end_ans=this->ans;
#endif
if (evatom!=0) {
this->k_three_end_vatom.set_size(GX,BX);
this->k_three_end_vatom.run(&this->atom->x, &sw1, &sw2, &sw3,
&map, &elem2param, &_nelements,
&this->nbor->dev_nbor, &this->_nbor_data->begin(),
&this->nbor->dev_ilist, &this->dev_short_nbor,
&end_ans->force, &end_ans->engv, &eflag, &vflag, &ainum,
&nbor_pitch, &this->_threads_per_atom, &this->_gpu_nbor);
} else {
this->k_three_end.set_size(GX,BX);
this->k_three_end.run(&this->atom->x, &sw1, &sw2, &sw3,
&map, &elem2param, &_nelements,
&this->nbor->dev_nbor, &this->_nbor_data->begin(),
&this->nbor->dev_ilist, &this->dev_short_nbor,
&end_ans->force, &end_ans->engv, &eflag, &vflag, &ainum,
&nbor_pitch, &this->_threads_per_atom, &this->_gpu_nbor);
}
this->time_pair.stop();
}
template class SW<PRECISION,ACC_PRECISION>;
Reference in New Issue View Git Blame Copy Permalink