ICODE-ADC/lammps
4
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Feb2021 GPU Package Update - GPU Package Files

Browse Source
This commit is contained in:
Michael Brown
2021-02-15 08:20:50 -08:00
parent 16004e8f45
commit e7e2d2323b
345 changed files with 13424 additions and 7708 deletions
Download Patch File Download Diff File Expand all files Collapse all files

53
lib/gpu/lal_born_coul_wolf.cu
View File

@ -51,6 +51,9 @@ __kernel void k_born_coul_wolf(const __global numtyp4 *restrict x_,
atom_info(t_per_atom,ii,tid,offset);
__local numtyp sp_lj[8];
int n_stride;
local_allocate_store_charge();
sp_lj[0]=sp_lj_in[0];
sp_lj[1]=sp_lj_in[1];
sp_lj[2]=sp_lj_in[2];
@ -60,18 +63,18 @@ __kernel void k_born_coul_wolf(const __global numtyp4 *restrict x_,
sp_lj[6]=sp_lj_in[6];
sp_lj[7]=sp_lj_in[7];
acctyp energy=(acctyp)0;
acctyp e_coul=(acctyp)0;
acctyp4 f;
f.x=(acctyp)0; f.y=(acctyp)0; f.z=(acctyp)0;
acctyp virial[6];
for (int i=0; i<6; i++)
virial[i]=(acctyp)0;
acctyp energy, e_coul, virial[6];
if (EVFLAG) {
energy=(acctyp)0;
e_coul=(acctyp)0;
for (int i=0; i<6; i++) virial[i]=(acctyp)0;
}
if (ii<inum) {
int nbor, nbor_end;
int i, numj;
__local int n_stride;
nbor_info(dev_nbor,dev_packed,nbor_pitch,t_per_atom,ii,offset,i,numj,
n_stride,nbor_end,nbor);
@ -79,7 +82,7 @@ __kernel void k_born_coul_wolf(const __global numtyp4 *restrict x_,
numtyp qtmp; fetch(qtmp,i,q_tex);
int itype=ix.w;
if (eflag>0) {
if (EVFLAG && eflag) {
acctyp e_self = -((acctyp)0.5*e_shift + alf/MY_PIS) *
qtmp*qtmp*qqrd2e/(acctyp)t_per_atom;
e_coul += (acctyp)2.0*e_self;
@ -137,7 +140,7 @@ __kernel void k_born_coul_wolf(const __global numtyp4 *restrict x_,
f.y+=dely*force;
f.z+=delz*force;
if (eflag>0) {
if (EVFLAG && eflag) {
if (rsq < cut_coulsq) {
numtyp e=v_sh;
if (factor_coul < (numtyp)1.0) e -= ((numtyp)1.0-factor_coul)*prefactor;
@ -149,7 +152,7 @@ __kernel void k_born_coul_wolf(const __global numtyp4 *restrict x_,
energy+=factor_lj*(e-coeff2[mtype].w);
}
}
if (vflag>0) {
if (EVFLAG && vflag) {
virial[0] += delx*delx*force;
virial[1] += dely*dely*force;
virial[2] += delz*delz*force;
@ -160,9 +163,9 @@ __kernel void k_born_coul_wolf(const __global numtyp4 *restrict x_,
}
} // for nbor
store_answers_q(f,energy,e_coul,virial,ii,inum,tid,t_per_atom,offset,eflag,
vflag,ans,engv);
} // if ii
store_answers_q(f,energy,e_coul,virial,ii,inum,tid,t_per_atom,offset,eflag,
vflag,ans,engv);
}
__kernel void k_born_coul_wolf_fast(const __global numtyp4 *restrict x_,
@ -186,28 +189,31 @@ __kernel void k_born_coul_wolf_fast(const __global numtyp4 *restrict x_,
__local numtyp4 coeff1[MAX_SHARED_TYPES*MAX_SHARED_TYPES];
__local numtyp4 coeff2[MAX_SHARED_TYPES*MAX_SHARED_TYPES];
__local numtyp sp_lj[8];
int n_stride;
local_allocate_store_charge();
if (tid<8)
sp_lj[tid]=sp_lj_in[tid];
if (tid<MAX_SHARED_TYPES*MAX_SHARED_TYPES) {
coeff1[tid]=coeff1_in[tid];
if (eflag>0)
if (EVFLAG && eflag)
coeff2[tid]=coeff2_in[tid];
}
acctyp energy=(acctyp)0;
acctyp e_coul=(acctyp)0;
acctyp4 f;
f.x=(acctyp)0; f.y=(acctyp)0; f.z=(acctyp)0;
acctyp virial[6];
for (int i=0; i<6; i++)
virial[i]=(acctyp)0;
acctyp energy, e_coul, virial[6];
if (EVFLAG) {
energy=(acctyp)0;
e_coul=(acctyp)0;
for (int i=0; i<6; i++) virial[i]=(acctyp)0;
}
__syncthreads();
if (ii<inum) {
int nbor, nbor_end;
int i, numj;
__local int n_stride;
nbor_info(dev_nbor,dev_packed,nbor_pitch,t_per_atom,ii,offset,i,numj,
n_stride,nbor_end,nbor);
@ -216,7 +222,7 @@ __kernel void k_born_coul_wolf_fast(const __global numtyp4 *restrict x_,
int iw=ix.w;
int itype=fast_mul((int)MAX_SHARED_TYPES,iw);
if (eflag>0) {
if (EVFLAG && eflag) {
acctyp e_self = -((acctyp)0.5*e_shift + alf/MY_PIS) *
qtmp*qtmp*qqrd2e/(acctyp)t_per_atom;
e_coul += (acctyp)2.0*e_self;
@ -273,7 +279,7 @@ __kernel void k_born_coul_wolf_fast(const __global numtyp4 *restrict x_,
f.y+=dely*force;
f.z+=delz*force;
if (eflag>0) {
if (EVFLAG && eflag) {
if (rsq < cut_coulsq) {
numtyp e=v_sh;
if (factor_coul < (numtyp)1.0) e -= ((numtyp)1.0-factor_coul)*prefactor;
@ -285,7 +291,7 @@ __kernel void k_born_coul_wolf_fast(const __global numtyp4 *restrict x_,
energy+=factor_lj*(e-coeff2[mtype].w);
}
}
if (vflag>0) {
if (EVFLAG && vflag) {
virial[0] += delx*delx*force;
virial[1] += dely*dely*force;
virial[2] += delz*delz*force;
@ -296,8 +302,7 @@ __kernel void k_born_coul_wolf_fast(const __global numtyp4 *restrict x_,
}
} // for nbor
store_answers_q(f,energy,e_coul,virial,ii,inum,tid,t_per_atom,offset,eflag,
vflag,ans,engv);
} // if ii
store_answers_q(f,energy,e_coul,virial,ii,inum,tid,t_per_atom,offset,eflag,
vflag,ans,engv);
}