Homec4science
Diffusion lammps (master) Browse

lammps/src/USER-INTEL/pair_lj_charmm_coul_charmm_intel.cpp
382de5034188master

Last Change
Raw File

pair_lj_charmm_coul_charmm_intel.cpp
View Options

/* ----------------------------------------------------------------------
LAMMPS - Large-scale Atomic/Molecular Massively Parallel Simulator
http://lammps.sandia.gov, Sandia National Laboratories
Steve Plimpton, sjplimp@sandia.gov
This software is distributed under the GNU General Public License.
See the README file in the top-level LAMMPS directory.
------------------------------------------------------------------------- */
/* ----------------------------------------------------------------------
Contributing author: W. Michael Brown (Intel)
------------------------------------------------------------------------- */
#include <math.h>
#include "pair_lj_charmm_coul_charmm_intel.h"
#include "atom.h"
#include "comm.h"
#include "force.h"
#include "group.h"
#include "memory.h"
#include "modify.h"
#include "neighbor.h"
#include "neigh_list.h"
#include "neigh_request.h"
#include "memory.h"
#include "suffix.h"
using namespace LAMMPS_NS;
#define LJ_T typename IntelBuffers<flt_t,flt_t>::vec4_t
/* ---------------------------------------------------------------------- */
PairLJCharmmCoulCharmmIntel::PairLJCharmmCoulCharmmIntel(LAMMPS *lmp) :
PairLJCharmmCoulCharmm(lmp)
{
suffix_flag |= Suffix::INTEL;
}
/* ---------------------------------------------------------------------- */
PairLJCharmmCoulCharmmIntel::~PairLJCharmmCoulCharmmIntel()
{
}
/* ---------------------------------------------------------------------- */
void PairLJCharmmCoulCharmmIntel::compute(int eflag, int vflag)
{
if (fix->precision()==FixIntel::PREC_MODE_MIXED)
compute<float,double>(eflag, vflag, fix->get_mixed_buffers(),
force_const_single);
else if (fix->precision()==FixIntel::PREC_MODE_DOUBLE)
compute<double,double>(eflag, vflag, fix->get_double_buffers(),
force_const_double);
else
compute<float,float>(eflag, vflag, fix->get_single_buffers(),
force_const_single);
fix->balance_stamp();
vflag_fdotr = 0;
}
template <class flt_t, class acc_t>
void PairLJCharmmCoulCharmmIntel::compute(int eflag, int vflag,
IntelBuffers<flt_t,acc_t> *buffers,
const ForceConst<flt_t> &fc)
{
if (eflag || vflag) {
ev_setup(eflag,vflag);
} else evflag = vflag_fdotr = 0;
const int inum = list->inum;
const int nthreads = comm->nthreads;
const int host_start = fix->host_start_pair();
const int offload_end = fix->offload_end_pair();
const int ago = neighbor->ago;
if (ago != 0 && fix->separate_buffers() == 0) {
fix->start_watch(TIME_PACK);
int packthreads;
if (nthreads > INTEL_HTHREADS) packthreads = nthreads;
else packthreads = 1;
#if defined(_OPENMP)
#pragma omp parallel if(packthreads > 1)
#endif
{
int ifrom, ito, tid;
IP_PRE_omp_range_id_align(ifrom, ito, tid, atom->nlocal+atom->nghost,
packthreads, sizeof(ATOM_T));
buffers->thr_pack(ifrom,ito,ago);
}
fix->stop_watch(TIME_PACK);
}
// -------------------- Regular version
int ovflag = 0;
if (vflag_fdotr) ovflag = 2;
else if (vflag) ovflag = 1;
if (eflag) {
if (force->newton_pair) {
eval<1,1>(1, ovflag, buffers, fc, 0, offload_end);
eval<1,1>(0, ovflag, buffers, fc, host_start, inum);
} else {
eval<1,0>(1, ovflag, buffers, fc, 0, offload_end);
eval<1,0>(0, ovflag, buffers, fc, host_start, inum);
}
} else {
if (force->newton_pair) {
eval<0,1>(1, ovflag, buffers, fc, 0, offload_end);
eval<0,1>(0, ovflag, buffers, fc, host_start, inum);
} else {
eval<0,0>(1, ovflag, buffers, fc, 0, offload_end);
eval<0,0>(0, ovflag, buffers, fc, host_start, inum);
}
}
}
/* ---------------------------------------------------------------------- */
template <int EFLAG, int NEWTON_PAIR, class flt_t, class acc_t>
void PairLJCharmmCoulCharmmIntel::eval(const int offload, const int vflag,
IntelBuffers<flt_t,acc_t> *buffers,
const ForceConst<flt_t> &fc,
const int astart, const int aend)
{
const int inum = aend - astart;
if (inum == 0) return;
int nlocal, nall, minlocal;
fix->get_buffern(offload, nlocal, nall, minlocal);
const int ago = neighbor->ago;
IP_PRE_pack_separate_buffers(fix, buffers, ago, offload, nlocal, nall);
ATOM_T * _noalias const x = buffers->get_x(offload);
flt_t * _noalias const q = buffers->get_q(offload);
const int * _noalias const numneigh = list->numneigh;
const int * _noalias const cnumneigh = buffers->cnumneigh(list);
const int * _noalias const firstneigh = buffers->firstneigh(list);
const flt_t * _noalias const special_coul = fc.special_coul;
const flt_t * _noalias const special_lj = fc.special_lj;
const flt_t qqrd2e = force->qqrd2e;
const flt_t inv_denom_lj = (flt_t)1.0/denom_lj;
const flt_t inv_denom_coul = (flt_t)1.0/denom_coul;
const flt_t * _noalias const cutsq = fc.cutsq[0];
const LJ_T * _noalias const lj = fc.lj[0];
const flt_t cut_ljsq = fc.cut_ljsq;
const flt_t cut_lj_innersq = fc.cut_lj_innersq;
const flt_t cut_coul_innersq = fc.cut_coul_innersq;
const flt_t cut_coulsq = fc.cut_coulsq;
const int ntypes = atom->ntypes + 1;
const int eatom = this->eflag_atom;
flt_t * _noalias const ccachex = buffers->get_ccachex();
flt_t * _noalias const ccachey = buffers->get_ccachey();
flt_t * _noalias const ccachez = buffers->get_ccachez();
flt_t * _noalias const ccachew = buffers->get_ccachew();
int * _noalias const ccachei = buffers->get_ccachei();
int * _noalias const ccachej = buffers->get_ccachej();
const int ccache_stride = _ccache_stride;
// Determine how much data to transfer
int x_size, q_size, f_stride, ev_size, separate_flag;
IP_PRE_get_transfern(ago, NEWTON_PAIR, EFLAG, vflag,
buffers, offload, fix, separate_flag,
x_size, q_size, ev_size, f_stride);
int tc;
FORCE_T * _noalias f_start;
acc_t * _noalias ev_global;
IP_PRE_get_buffers(offload, buffers, fix, tc, f_start, ev_global);
const int nthreads = tc;
#ifdef _LMP_INTEL_OFFLOAD
int *overflow = fix->get_off_overflow_flag();
double *timer_compute = fix->off_watch_pair();
if (offload) fix->start_watch(TIME_OFFLOAD_LATENCY);
#pragma offload target(mic:_cop) if(offload) \
in(special_lj,special_coul:length(0) alloc_if(0) free_if(0)) \
in(cutsq,lj:length(0) alloc_if(0) free_if(0)) \
in(firstneigh:length(0) alloc_if(0) free_if(0)) \
in(cnumneigh:length(0) alloc_if(0) free_if(0)) \
in(numneigh:length(0) alloc_if(0) free_if(0)) \
in(x:length(x_size) alloc_if(0) free_if(0)) \
in(q:length(q_size) alloc_if(0) free_if(0)) \
in(overflow:length(0) alloc_if(0) free_if(0)) \
in(ccachex,ccachey,ccachez,ccachew:length(0) alloc_if(0) free_if(0)) \
in(ccachei,ccachej:length(0) alloc_if(0) free_if(0)) \
in(ccache_stride,nthreads,qqrd2e,inum,nall,ntypes,cut_coulsq) \
in(vflag,eatom,f_stride,separate_flag,offload) \
in(astart,cut_ljsq,cut_lj_innersq,nlocal,inv_denom_lj,minlocal) \
in(inv_denom_coul,cut_coul_innersq) \
out(f_start:length(f_stride) alloc_if(0) free_if(0)) \
out(ev_global:length(ev_size) alloc_if(0) free_if(0)) \
out(timer_compute:length(1) alloc_if(0) free_if(0)) \
signal(f_start)
#endif
{
#if defined(__MIC__) && defined(_LMP_INTEL_OFFLOAD)
*timer_compute = MIC_Wtime();
#endif
IP_PRE_repack_for_offload(NEWTON_PAIR, separate_flag, nlocal, nall,
f_stride, x, q);
acc_t oevdwl, oecoul, ov0, ov1, ov2, ov3, ov4, ov5;
if (EFLAG) oevdwl = oecoul = (acc_t)0;
if (vflag) ov0 = ov1 = ov2 = ov3 = ov4 = ov5 = (acc_t)0;
// loop over neighbors of my atoms
#if defined(_OPENMP)
#pragma omp parallel reduction(+:oevdwl,oecoul,ov0,ov1,ov2,ov3,ov4,ov5)
#endif
{
int iifrom, iip, iito, tid;
IP_PRE_omp_stride_id(iifrom, iip, iito, tid, inum, nthreads);
iifrom += astart;
iito += astart;
int foff;
if (NEWTON_PAIR) foff = tid * f_stride - minlocal;
else foff = -minlocal;
FORCE_T * _noalias const f = f_start + foff;
if (NEWTON_PAIR) memset(f + minlocal, 0, f_stride * sizeof(FORCE_T));
flt_t cutboth = cut_coulsq;
const int toffs = tid * ccache_stride;
flt_t * _noalias const tdelx = ccachex + toffs;
flt_t * _noalias const tdely = ccachey + toffs;
flt_t * _noalias const tdelz = ccachez + toffs;
flt_t * _noalias const trsq = ccachew + toffs;
int * _noalias const tj = ccachei + toffs;
int * _noalias const tjtype = ccachej + toffs;
for (int i = iifrom; i < iito; i += iip) {
// const int i = ilist[ii];
const int itype = x[i].w;
const int ptr_off = itype * ntypes;
const flt_t * _noalias const cutsqi = cutsq + ptr_off;
const LJ_T * _noalias const lji = lj + ptr_off;
const int * _noalias const jlist = firstneigh + cnumneigh[i];
const int jnum = numneigh[i];
acc_t fxtmp,fytmp,fztmp,fwtmp;
acc_t sevdwl, secoul, sv0, sv1, sv2, sv3, sv4, sv5;
const flt_t xtmp = x[i].x;
const flt_t ytmp = x[i].y;
const flt_t ztmp = x[i].z;
const flt_t qtmp = q[i];
fxtmp = fytmp = fztmp = (acc_t)0;
if (EFLAG) fwtmp = sevdwl = secoul = (acc_t)0;
if (NEWTON_PAIR == 0)
if (vflag==1) sv0 = sv1 = sv2 = sv3 = sv4 = sv5 = (acc_t)0;
int ej = 0;
#if defined(LMP_SIMD_COMPILER)
#pragma vector aligned
#pragma ivdep
#endif
for (int jj = 0; jj < jnum; jj++) {
const int j = jlist[jj] & NEIGHMASK;
const flt_t delx = xtmp - x[j].x;
const flt_t dely = ytmp - x[j].y;
const flt_t delz = ztmp - x[j].z;
const flt_t rsq = delx * delx + dely * dely + delz * delz;
if (rsq < cut_coulsq) {
trsq[ej]=rsq;
tdelx[ej]=delx;
tdely[ej]=dely;
tdelz[ej]=delz;
tjtype[ej]=x[j].w;
tj[ej]=jlist[jj];
ej++;
}
}
#if defined(LMP_SIMD_COMPILER)
#pragma vector aligned
#pragma simd reduction(+:fxtmp, fytmp, fztmp, fwtmp, sevdwl, secoul, \
sv0, sv1, sv2, sv3, sv4, sv5)
#endif
for (int jj = 0; jj < ej; jj++) {
flt_t forcecoul, forcelj, evdwl;
forcecoul = forcelj = evdwl = (flt_t)0.0;
const int j = tj[jj] & NEIGHMASK;
const int sbindex = tj[jj] >> SBBITS & 3;
const flt_t rsq = trsq[jj];
const flt_t r2inv = (flt_t)1.0 / rsq;
const flt_t r_inv = (flt_t)1.0 / sqrt(rsq);
forcecoul = qqrd2e * qtmp * q[j] * r_inv;
if (rsq > cut_coul_innersq) {
const flt_t ccr = cut_coulsq - rsq;
const flt_t switch1 = ccr * ccr * inv_denom_coul *
(cut_coulsq + (flt_t)2.0 * rsq - (flt_t)3.0 * cut_coul_innersq);
forcecoul *= switch1;
}
#ifdef INTEL_VMASK
if (rsq < cut_ljsq) {
#endif
const int jtype = tjtype[jj];
flt_t r6inv = r2inv * r2inv * r2inv;
forcelj = r6inv * (lji[jtype].x * r6inv - lji[jtype].y);
if (EFLAG) evdwl = r6inv*(lji[jtype].z * r6inv - lji[jtype].w);
#ifdef INTEL_VMASK
if (rsq > cut_lj_innersq) {
#endif
const flt_t drsq = cut_ljsq - rsq;
const flt_t cut2 = (rsq - cut_lj_innersq) * drsq;
const flt_t switch1 = drsq * (drsq * drsq + (flt_t)3.0 * cut2) *
inv_denom_lj;
const flt_t switch2 = (flt_t)12.0 * rsq * cut2 * inv_denom_lj;
if (EFLAG) {
#ifndef INTEL_VMASK
if (rsq > cut_lj_innersq) {
#endif
forcelj = forcelj * switch1 + evdwl * switch2;
evdwl *= switch1;
#ifndef INTEL_VMASK
}
#endif
} else {
const flt_t philj = r6inv * (lji[jtype].z*r6inv -
lji[jtype].w);
#ifndef INTEL_VMASK
if (rsq > cut_lj_innersq)
#endif
forcelj = forcelj * switch1 + philj * switch2;
}
#ifdef INTEL_VMASK
}
#endif
#ifdef INTEL_VMASK
}
#else
if (rsq > cut_ljsq) { forcelj = (flt_t)0.0; evdwl = (flt_t)0.0; }
#endif
if (sbindex) {
const flt_t factor_coul = special_coul[sbindex];
forcecoul *= factor_coul;
const flt_t factor_lj = special_lj[sbindex];
forcelj *= factor_lj;
if (EFLAG) evdwl *= factor_lj;
}
const flt_t fpair = (forcecoul + forcelj) * r2inv;
const flt_t fpx = fpair * tdelx[jj];
fxtmp += fpx;
if (NEWTON_PAIR) f[j].x -= fpx;
const flt_t fpy = fpair * tdely[jj];
fytmp += fpy;
if (NEWTON_PAIR) f[j].y -= fpy;
const flt_t fpz = fpair * tdelz[jj];
fztmp += fpz;
if (NEWTON_PAIR) f[j].z -= fpz;
if (EFLAG) {
sevdwl += evdwl;
secoul += forcecoul;
if (eatom) {
fwtmp += (flt_t)0.5 * evdwl + (flt_t)0.5 * forcecoul;
if (NEWTON_PAIR)
f[j].w += (flt_t)0.5 * evdwl + (flt_t)0.5 * forcecoul;
}
}
if (NEWTON_PAIR == 0)
IP_PRE_ev_tally_nborv(vflag, tdelx[jj], tdely[jj], tdelz[jj],
fpx, fpy, fpz);
} // for jj
if (NEWTON_PAIR) {
f[i].x += fxtmp;
f[i].y += fytmp;
f[i].z += fztmp;
} else {
f[i].x = fxtmp;
f[i].y = fytmp;
f[i].z = fztmp;
}
IP_PRE_ev_tally_atomq(NEWTON_PAIR, EFLAG, vflag, f, fwtmp);
} // for ii
IP_PRE_fdotr_reduce_omp(NEWTON_PAIR, nall, minlocal, nthreads, f_start,
f_stride, x, offload, vflag, ov0, ov1, ov2, ov3,
ov4, ov5);
} // end of omp parallel region
IP_PRE_fdotr_reduce(NEWTON_PAIR, nall, nthreads, f_stride, vflag,
ov0, ov1, ov2, ov3, ov4, ov5);
if (EFLAG) {
if (NEWTON_PAIR == 0) {
oevdwl *= (acc_t)0.5;
oecoul *= (acc_t)0.5;
}
ev_global[0] = oevdwl;
ev_global[1] = oecoul;
}
if (vflag) {
if (NEWTON_PAIR == 0) {
ov0 *= (acc_t)0.5;
ov1 *= (acc_t)0.5;
ov2 *= (acc_t)0.5;
ov3 *= (acc_t)0.5;
ov4 *= (acc_t)0.5;
ov5 *= (acc_t)0.5;
}
ev_global[2] = ov0;
ev_global[3] = ov1;
ev_global[4] = ov2;
ev_global[5] = ov3;
ev_global[6] = ov4;
ev_global[7] = ov5;
}
#if defined(__MIC__) && defined(_LMP_INTEL_OFFLOAD)
*timer_compute = MIC_Wtime() - *timer_compute;
#endif
} // end of offload region
if (offload)
fix->stop_watch(TIME_OFFLOAD_LATENCY);
else
fix->stop_watch(TIME_HOST_PAIR);
if (EFLAG || vflag)
fix->add_result_array(f_start, ev_global, offload, eatom, 0, vflag);
else
fix->add_result_array(f_start, 0, offload);
}
/* ---------------------------------------------------------------------- */
void PairLJCharmmCoulCharmmIntel::init_style()
{
PairLJCharmmCoulCharmm::init_style();
if (force->newton_pair == 0) {
neighbor->requests[neighbor->nrequest-1]->half = 0;
neighbor->requests[neighbor->nrequest-1]->full = 1;
}
neighbor->requests[neighbor->nrequest-1]->intel = 1;
int ifix = modify->find_fix("package_intel");
if (ifix < 0)
error->all(FLERR,
"The 'package intel' command is required for /intel styles");
fix = static_cast<FixIntel *>(modify->fix[ifix]);
fix->pair_init_check();
#ifdef _LMP_INTEL_OFFLOAD
_cop = fix->coprocessor_number();
#endif
if (fix->precision() == FixIntel::PREC_MODE_MIXED)
pack_force_const(force_const_single, fix->get_mixed_buffers());
else if (fix->precision() == FixIntel::PREC_MODE_DOUBLE)
pack_force_const(force_const_double, fix->get_double_buffers());
else
pack_force_const(force_const_single, fix->get_single_buffers());
}
template <class flt_t, class acc_t>
void PairLJCharmmCoulCharmmIntel::pack_force_const(ForceConst<flt_t> &fc,
IntelBuffers<flt_t,acc_t> *buffers)
{
int off_ccache = 0;
#ifdef _LMP_INTEL_OFFLOAD
if (_cop >= 0) off_ccache = 1;
#endif
buffers->grow_ccache(off_ccache, comm->nthreads, 1);
_ccache_stride = buffers->ccache_stride();
int tp1 = atom->ntypes + 1;
fc.set_ntypes(tp1, memory, _cop);
buffers->set_ntypes(tp1);
flt_t **cutneighsq = buffers->get_cutneighsq();
// Repeat cutsq calculation because done after call to init_style
double cut, cutneigh;
if (cut_lj > cut_coul)
error->all(FLERR,
"Intel varient of lj/charmm/coul/long expects lj cutoff<=coulombic");
for (int i = 1; i <= atom->ntypes; i++) {
for (int j = i; j <= atom->ntypes; j++) {
if (setflag[i][j] != 0 || (setflag[i][i] != 0 && setflag[j][j] != 0)) {
cut = init_one(i, j);
cutneigh = cut + neighbor->skin;
cutsq[i][j] = cutsq[j][i] = cut*cut;
cutneighsq[i][j] = cutneighsq[j][i] = cutneigh * cutneigh;
}
}
}
cut_coul_innersq = cut_coul_inner * cut_coul_inner;
cut_lj_innersq = cut_lj_inner * cut_lj_inner;
cut_ljsq = cut_lj * cut_lj;
cut_coulsq = cut_coul * cut_coul;
cut_bothsq = MAX(cut_ljsq, cut_coulsq);
fc.cut_coulsq = cut_coulsq;
fc.cut_ljsq = cut_ljsq;
fc.cut_coul_innersq = cut_coul_innersq;
fc.cut_lj_innersq = cut_lj_innersq;
for (int i = 0; i < 4; i++) {
fc.special_lj[i] = force->special_lj[i];
fc.special_coul[i] = force->special_coul[i];
fc.special_coul[0] = 1.0;
fc.special_lj[0] = 1.0;
}
for (int i = 1; i < tp1; i++) {
for (int j = 1; j < tp1; j++) {
fc.lj[i][j].x = lj1[i][j];
fc.lj[i][j].y = lj2[i][j];
fc.lj[i][j].z = lj3[i][j];
fc.lj[i][j].w = lj4[i][j];
fc.cutsq[i][j] = cutsq[i][j];
}
}
#ifdef _LMP_INTEL_OFFLOAD
if (_cop < 0) return;
flt_t * special_lj = fc.special_lj;
flt_t * special_coul = fc.special_coul;
flt_t * cutsq = fc.cutsq[0];
LJ_T * lj = fc.lj[0];
flt_t * ocutneighsq = cutneighsq[0];
int tp1sq = tp1 * tp1;
#pragma offload_transfer target(mic:_cop) \
in(special_lj, special_coul: length(4) alloc_if(0) free_if(0)) \
in(cutsq,lj: length(tp1sq) alloc_if(0) free_if(0)) \
in(ocutneighsq: length(tp1sq) alloc_if(0) free_if(0))
#endif
}
/* ---------------------------------------------------------------------- */
template <class flt_t>
void PairLJCharmmCoulCharmmIntel::ForceConst<flt_t>::set_ntypes(
const int ntypes, Memory *memory, const int cop) {
if (ntypes != _ntypes) {
if (_ntypes > 0) {
#ifdef _LMP_INTEL_OFFLOAD
flt_t * ospecial_lj = special_lj;
flt_t * ospecial_coul = special_coul;
flt_t * ocutsq = cutsq[0];
typename IntelBuffers<flt_t,flt_t>::vec4_t * olj = lj[0];
if (ospecial_lj != NULL && ocutsq != NULL && olj != NULL &&
ospecial_coul != NULL && cop >= 0) {
#pragma offload_transfer target(mic:cop) \
nocopy(ospecial_lj, ospecial_coul: alloc_if(0) free_if(1)) \
nocopy(ocutsq, olj: alloc_if(0) free_if(1))
}
#endif
_memory->destroy(cutsq);
_memory->destroy(lj);
}
if (ntypes > 0) {
_cop = cop;
memory->create(cutsq,ntypes,ntypes,"fc.cutsq");
memory->create(lj,ntypes,ntypes,"fc.lj");
#ifdef _LMP_INTEL_OFFLOAD
flt_t * ospecial_lj = special_lj;
flt_t * ospecial_coul = special_coul;
flt_t * ocutsq = cutsq[0];
typename IntelBuffers<flt_t,flt_t>::vec4_t * olj = lj[0];
int tp1sq = ntypes*ntypes;
if (ospecial_lj != NULL && ocutsq != NULL && olj != NULL &&
ospecial_coul != NULL && cop >= 0) {
#pragma offload_transfer target(mic:cop) \
nocopy(ospecial_lj: length(4) alloc_if(1) free_if(0)) \
nocopy(ospecial_coul: length(4) alloc_if(1) free_if(0)) \
nocopy(ocutsq,olj: length(tp1sq) alloc_if(1) free_if(0))
}
#endif
}
}
_ntypes=ntypes;
_memory=memory;
}
c4science · Help