From bf26b12668a46206df3104dc98937538721057c6 Mon Sep 17 00:00:00 2001
From: athomps <athomps@f3b2605a-c512-4ea7-a41b-209d697bcdaa>
Date: Wed, 21 Oct 2015 03:57:35 +0000
Subject: [PATCH] Added vashishta to USER-OMP from Axel
git-svn-id: svn://svn.icms.temple.edu/lammps-ro/trunk@14125 f3b2605a-c512-4ea7-a41b-209d697bcdaa
---
doc/pair_vashishta.txt | 24 ++++
src/MANYBODY/pair_vashishta.cpp | 2 +-
src/USER-OMP/pair_vashishta_omp.cpp | 213 ++++++++++++++++++++++++++++
src/USER-OMP/pair_vashishta_omp.h | 48 +++++++
4 files changed, 286 insertions(+), 1 deletion(-)
create mode 100644 src/USER-OMP/pair_vashishta_omp.cpp
create mode 100644 src/USER-OMP/pair_vashishta_omp.h
diff --git a/doc/pair_vashishta.txt b/doc/pair_vashishta.txt
index 053db937a1..6effdf7944 100644
--- a/doc/pair_vashishta.txt
+++ b/doc/pair_vashishta.txt
@@ -7,6 +7,7 @@
:line
pair_style vashishta command :h3
+pair_style vashishta/omp command :h3
[Syntax:]
@@ -150,6 +151,29 @@ two-body parameters from the CCC and CSiSi entries.
:line
+Styles with a {cuda}, {gpu}, {intel}, {kk}, {omp}, or {opt} suffix are
+functionally the same as the corresponding style without the suffix.
+They have been optimized to run faster, depending on your available
+hardware, as discussed in "Section_accelerate"_Section_accelerate.html
+of the manual. The accelerated styles take the same arguments and
+should produce the same results, except for round-off and precision
+issues.
+
+These accelerated styles are part of the USER-CUDA, GPU, USER-INTEL,
+KOKKOS, USER-OMP and OPT packages, respectively. They are only
+enabled if LAMMPS was built with those packages. See the "Making
+LAMMPS"_Section_start.html#start_3 section for more info.
+
+You can specify the accelerated styles explicitly in your input script
+by including their suffix, or you can use the "-suffix command-line
+switch"_Section_start.html#start_7 when you invoke LAMMPS, or you can
+use the "suffix"_suffix.html command in your input script.
+
+See "Section_accelerate"_Section_accelerate.html of the manual for
+more instructions on how to use the accelerated styles effectively.
+
+:line
+
[Mixing, shift, table, tail correction, restart, rRESPA info]:
For atom type pairs I,J and I != J, where types I and J correspond to
diff --git a/src/MANYBODY/pair_vashishta.cpp b/src/MANYBODY/pair_vashishta.cpp
index ad1df5034f..c8e6d65dc0 100755
--- a/src/MANYBODY/pair_vashishta.cpp
+++ b/src/MANYBODY/pair_vashishta.cpp
@@ -33,7 +33,6 @@
#include "error.h"
using namespace LAMMPS_NS;
-//using namespace MathConst;
#define MAXLINE 1024
#define DELTA 4
@@ -52,6 +51,7 @@ PairVashishta::PairVashishta(LAMMPS *lmp) : Pair(lmp)
nparams = maxparam = 0;
params = NULL;
elem2param = NULL;
+ map = NULL;
}
/* ----------------------------------------------------------------------
diff --git a/src/USER-OMP/pair_vashishta_omp.cpp b/src/USER-OMP/pair_vashishta_omp.cpp
new file mode 100644
index 0000000000..f89b2ad184
--- /dev/null
+++ b/src/USER-OMP/pair_vashishta_omp.cpp
@@ -0,0 +1,213 @@
+/* ----------------------------------------------------------------------
+ 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: Axel Kohlmeyer (Temple U)
+------------------------------------------------------------------------- */
+
+#include "math.h"
+#include "pair_vashishta_omp.h"
+#include "atom.h"
+#include "comm.h"
+#include "force.h"
+#include "neighbor.h"
+#include "neigh_list.h"
+
+#include "suffix.h"
+using namespace LAMMPS_NS;
+
+/* ---------------------------------------------------------------------- */
+
+PairVashishtaOMP::PairVashishtaOMP(LAMMPS *lmp) :
+ PairVashishta(lmp), ThrOMP(lmp, THR_PAIR)
+{
+ suffix_flag |= Suffix::OMP;
+ respa_enable = 0;
+}
+
+/* ---------------------------------------------------------------------- */
+
+void PairVashishtaOMP::compute(int eflag, int vflag)
+{
+ if (eflag || vflag) {
+ ev_setup(eflag,vflag);
+ } else evflag = vflag_fdotr = 0;
+
+ const int nall = atom->nlocal + atom->nghost;
+ const int nthreads = comm->nthreads;
+ const int inum = list->inum;
+
+#if defined(_OPENMP)
+#pragma omp parallel default(none) shared(eflag,vflag)
+#endif
+ {
+ int ifrom, ito, tid;
+
+ loop_setup_thr(ifrom, ito, tid, inum, nthreads);
+ ThrData *thr = fix->get_thr(tid);
+ thr->timer(Timer::START);
+ ev_setup_thr(eflag, vflag, nall, eatom, vatom, thr);
+
+ if (evflag) {
+ if (eflag) {
+ eval<1,1>(ifrom, ito, thr);
+ } else {
+ eval<1,0>(ifrom, ito, thr);
+ }
+ } else eval<0,0>(ifrom, ito, thr);
+
+ thr->timer(Timer::PAIR);
+ reduce_thr(this, eflag, vflag, thr);
+ } // end of omp parallel region
+}
+
+template <int EVFLAG, int EFLAG>
+void PairVashishtaOMP::eval(int iifrom, int iito, ThrData * const thr)
+{
+ int i,j,k,ii,jj,kk,jnum,jnumm1;
+ tagint itag,jtag;
+ int itype,jtype,ktype,ijparam,ikparam,ijkparam;
+ double xtmp,ytmp,ztmp,delx,dely,delz,evdwl,fpair;
+ double rsq,rsq1,rsq2;
+ double delr1[3],delr2[3],fj[3],fk[3];
+ int *ilist,*jlist,*numneigh,**firstneigh;
+
+ evdwl = 0.0;
+
+ const dbl3_t * _noalias const x = (dbl3_t *) atom->x[0];
+ dbl3_t * _noalias const f = (dbl3_t *) thr->get_f()[0];
+ const tagint * _noalias const tag = atom->tag;
+ const int * _noalias const type = atom->type;
+ const int nlocal = atom->nlocal;
+
+ ilist = list->ilist;
+ numneigh = list->numneigh;
+ firstneigh = list->firstneigh;
+
+ double fxtmp,fytmp,fztmp;
+
+ // loop over full neighbor list of my atoms
+
+ for (ii = iifrom; ii < iito; ++ii) {
+
+ i = ilist[ii];
+ itag = tag[i];
+ itype = map[type[i]];
+ xtmp = x[i].x;
+ ytmp = x[i].y;
+ ztmp = x[i].z;
+ fxtmp = fytmp = fztmp = 0.0;
+
+ // two-body interactions, skip half of them
+
+ jlist = firstneigh[i];
+ jnum = numneigh[i];
+
+ for (jj = 0; jj < jnum; jj++) {
+ j = jlist[jj];
+ j &= NEIGHMASK;
+ jtag = tag[j];
+
+ if (itag > jtag) {
+ if ((itag+jtag) % 2 == 0) continue;
+ } else if (itag < jtag) {
+ if ((itag+jtag) % 2 == 1) continue;
+ } else {
+ if (x[j].z < ztmp) continue;
+ if (x[j].z == ztmp && x[j].y < ytmp) continue;
+ if (x[j].z == ztmp && x[j].y == ytmp && x[j].x < xtmp) continue;
+ }
+
+ jtype = map[type[j]];
+
+ delx = xtmp - x[j].x;
+ dely = ytmp - x[j].y;
+ delz = ztmp - x[j].z;
+ rsq = delx*delx + dely*dely + delz*delz;
+
+ ijparam = elem2param[itype][jtype][jtype];
+ if (rsq >= params[ijparam].cutsq) continue;
+
+ twobody(¶ms[ijparam],rsq,fpair,EFLAG,evdwl);
+
+ fxtmp += delx*fpair;
+ fytmp += dely*fpair;
+ fztmp += delz*fpair;
+ f[j].x -= delx*fpair;
+ f[j].y -= dely*fpair;
+ f[j].z -= delz*fpair;
+
+ if (EVFLAG) ev_tally_thr(this,i,j,nlocal,/* newton_pair */ 1,
+ evdwl,0.0,fpair,delx,dely,delz,thr);
+ }
+
+ jnumm1 = jnum - 1;
+
+ for (jj = 0; jj < jnumm1; jj++) {
+ j = jlist[jj];
+ j &= NEIGHMASK;
+ jtype = map[type[j]];
+ ijparam = elem2param[itype][jtype][jtype];
+ delr1[0] = x[j].x - xtmp;
+ delr1[1] = x[j].y - ytmp;
+ delr1[2] = x[j].z - ztmp;
+ rsq1 = delr1[0]*delr1[0] + delr1[1]*delr1[1] + delr1[2]*delr1[2];
+ if (rsq1 >= params[ijparam].cutsq2) continue;
+
+ double fjxtmp,fjytmp,fjztmp;
+ fjxtmp = fjytmp = fjztmp = 0.0;
+
+ for (kk = jj+1; kk < jnum; kk++) {
+ k = jlist[kk];
+ k &= NEIGHMASK;
+ ktype = map[type[k]];
+ ikparam = elem2param[itype][ktype][ktype];
+ ijkparam = elem2param[itype][jtype][ktype];
+
+ delr2[0] = x[k].x - xtmp;
+ delr2[1] = x[k].y - ytmp;
+ delr2[2] = x[k].z - ztmp;
+ rsq2 = delr2[0]*delr2[0] + delr2[1]*delr2[1] + delr2[2]*delr2[2];
+ if (rsq2 >= params[ikparam].cutsq2) continue;
+
+ threebody(¶ms[ijparam],¶ms[ikparam],¶ms[ijkparam],
+ rsq1,rsq2,delr1,delr2,fj,fk,EFLAG,evdwl);
+
+ fxtmp -= fj[0] + fk[0];
+ fytmp -= fj[1] + fk[1];
+ fztmp -= fj[2] + fk[2];
+ fjxtmp += fj[0];
+ fjytmp += fj[1];
+ fjztmp += fj[2];
+ f[k].x += fk[0];
+ f[k].y += fk[1];
+ f[k].z += fk[2];
+
+ if (EVFLAG) ev_tally3_thr(this,i,j,k,evdwl,0.0,fj,fk,delr1,delr2,thr);
+ }
+ f[j].x += fjxtmp;
+ f[j].y += fjytmp;
+ f[j].z += fjztmp;
+ }
+ f[i].x += fxtmp;
+ f[i].y += fytmp;
+ f[i].z += fztmp;
+ }
+}
+
+/* ---------------------------------------------------------------------- */
+
+double PairVashishtaOMP::memory_usage()
+{
+ double bytes = memory_usage_thr();
+ bytes += PairVashishta::memory_usage();
+
+ return bytes;
+}
diff --git a/src/USER-OMP/pair_vashishta_omp.h b/src/USER-OMP/pair_vashishta_omp.h
new file mode 100644
index 0000000000..4c8efddb86
--- /dev/null
+++ b/src/USER-OMP/pair_vashishta_omp.h
@@ -0,0 +1,48 @@
+/* -*- c++ -*- ----------------------------------------------------------
+ LAMMPS - Large-scale Atomic/Molecular Massively Parallel Simulator
+ http://lammps.sandia.gov, Sandia National Laboratories
+ Steve Plimpton, sjplimp@sandia.gov
+
+ Copyright (2003) Sandia Corporation. Under the terms of Contract
+ DE-AC04-94AL85000 with Sandia Corporation, the U.S. Government retains
+ certain rights in this software. This software is distributed under
+ the GNU General Public License.
+
+ See the README file in the top-level LAMMPS directory.
+------------------------------------------------------------------------- */
+
+/* ----------------------------------------------------------------------
+ Contributing author: Axel Kohlmeyer (Temple U)
+------------------------------------------------------------------------- */
+
+#ifdef PAIR_CLASS
+
+PairStyle(vashishta/omp,PairVashishtaOMP)
+
+#else
+
+#ifndef LMP_PAIR_VASHISHTA_OMP_H
+#define LMP_PAIR_VASHISHTA_OMP_H
+
+#include "pair_vashishta.h"
+#include "thr_omp.h"
+
+namespace LAMMPS_NS {
+
+class PairVashishtaOMP : public PairVashishta, public ThrOMP {
+
+ public:
+ PairVashishtaOMP(class LAMMPS *);
+
+ virtual void compute(int, int);
+ virtual double memory_usage();
+
+ private:
+ template <int EVFLAG, int EFLAG>
+ void eval(int ifrom, int ito, ThrData * const thr);
+};
+
+}
+
+#endif
+#endif
--
GitLab