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

Fixed bugs with _tep and _fieldp to allow mixed-precision builds, being defensive with acctyp for these variables

Browse Source
This commit is contained in:
Trung Nguyen
2021-09-20 11:38:50 -05:00
parent 0228867d8e
commit 4e88cd158e
3 changed files with 72 additions and 72 deletions
Download Patch File Download Diff File Expand all files Collapse all files

114
lib/gpu/lal_amoeba.cu
View File

@ -102,7 +102,7 @@ _texture( q_tex,int2);
dufld[5]=red_acc[5][tid]; \
} \
if (offset==0 && ii<inum) { \
numtyp4 t; \
acctyp4 t; \
t.x = diz*ufld[1] - diy*ufld[2] + qixz*dufld[1] - qixy*dufld[3] + \
(numtyp)2.0*qiyz*(dufld[2]-dufld[5]) + (qizz-qiyy)*dufld[4]; \
t.y = dix*ufld[2] - diz*ufld[0] - qiyz*dufld[1] + qixy*dufld[4] + \
@ -136,7 +136,7 @@ _texture( q_tex,int2);
_fieldp[5]=red_acc[5][tid]; \
} \
if (offset==0 && ii<inum) { \
numtyp4 f, fp; \
acctyp4 f, fp; \
f.x = _fieldp[0]; \
f.y = _fieldp[1]; \
f.z = _fieldp[2]; \
@ -243,7 +243,7 @@ _texture( q_tex,int2);
} \
} \
if (offset==0 && ii<inum) { \
numtyp4 t; \
acctyp4 t; \
t.x = diz*ufld[1] - diy*ufld[2] + qixz*dufld[1] - qixy*dufld[3] + \
(numtyp)2.0*qiyz*(dufld[2]-dufld[5]) + (qizz-qiyy)*dufld[4]; \
t.y = dix*ufld[2] - diz*ufld[0] - qiyz*dufld[1] + qixy*dufld[4] + \
@ -266,7 +266,7 @@ _texture( q_tex,int2);
} \
} \
if (offset==0 && ii<inum) { \
numtyp4 f, fp; \
acctyp4 f, fp; \
f.x = _fieldp[0]; \
f.y = _fieldp[1]; \
f.z = _fieldp[2]; \
@ -591,7 +591,7 @@ __kernel void k_amoeba_multipole(const __global numtyp4 *restrict x_,
const __global int *dev_short_nbor,
__global acctyp4 *restrict ans,
__global acctyp *restrict engv,
__global numtyp4 *restrict tep,
__global acctyp4 *restrict tep,
const int eflag, const int vflag, const int inum,
const int nall, const int nbor_pitch,
const int t_per_atom, const numtyp aewald,
@ -883,7 +883,7 @@ __kernel void k_amoeba_udirect2b(const __global numtyp4 *restrict x_,
const __global int *dev_nbor,
const __global int *dev_packed,
const __global int *dev_short_nbor,
__global numtyp4 *restrict fieldp,
__global acctyp4 *restrict fieldp,
const int inum, const int nall,
const int nbor_pitch, const int t_per_atom,
const numtyp aewald, const numtyp off2,
@ -1097,7 +1097,7 @@ __kernel void k_amoeba_umutual2b(const __global numtyp4 *restrict x_,
const __global int *dev_nbor,
const __global int *dev_packed,
const __global int *dev_short_nbor,
__global numtyp4 *restrict fieldp,
__global acctyp4 *restrict fieldp,
const int inum, const int nall,
const int nbor_pitch, const int t_per_atom,
const numtyp aewald, const numtyp off2,
@ -1256,55 +1256,6 @@ __kernel void k_amoeba_umutual2b(const __global numtyp4 *restrict x_,
store_answers_fieldp(_fieldp,ii,inum,tid,t_per_atom,offset,i,fieldp);
}
/* ----------------------------------------------------------------------
scan standard neighbor list and make it compatible with 1-5 neighbors
if IJ entry is a 1-2,1-3,1-4 neighbor then adjust offset to SBBITS15
else scan special15 to see if a 1-5 neighbor and adjust offset to SBBITS15
else do nothing to IJ entry
------------------------------------------------------------------------- */
__kernel void k_special15(__global int * dev_nbor,
const __global int * dev_packed,
const __global tagint *restrict tag,
const __global int *restrict nspecial15,
const __global tagint *restrict special15,
const int inum, const int nall, const int nbor_pitch,
const int t_per_atom) {
int tid, ii, offset, n_stride, i;
atom_info(t_per_atom,ii,tid,offset);
if (ii<inum) {
int numj, nbor, nbor_end;
nbor_info(dev_nbor,dev_packed,nbor_pitch,t_per_atom,ii,offset,i,numj,
n_stride,nbor_end,nbor);
int n15 = nspecial15[ii];
for ( ; nbor<nbor_end; nbor+=n_stride) {
int sj=dev_packed[nbor];
int which = sj >> SBBITS & 3;
int j = sj & NEIGHMASK;
tagint jtag = tag[j];
if (!which) {
int offset=ii;
for (int k=0; k<n15; k++) {
if (special15[offset] == jtag) {
which = 4;
break;
}
offset += nall;
}
}
if (which) dev_nbor[nbor] = j ^ (which << SBBITS15);
} // for nbor
} // if ii
}
/* ----------------------------------------------------------------------
polar_real = real-space portion of induced dipole polarization
adapted from Tinker epreal1d() routine
@ -1319,7 +1270,7 @@ __kernel void k_amoeba_polar(const __global numtyp4 *restrict x_,
const __global int *dev_short_nbor,
__global acctyp4 *restrict ans,
__global acctyp *restrict engv,
__global numtyp4 *restrict tep,
__global acctyp4 *restrict tep,
const int eflag, const int vflag, const int inum,
const int nall, const int nbor_pitch, const int t_per_atom,
const numtyp aewald, const numtyp felec,
@ -1828,6 +1779,55 @@ __kernel void k_amoeba_polar(const __global numtyp4 *restrict x_,
offset,eflag,vflag,ans,engv,NUM_BLOCKS_X);
}
/* ----------------------------------------------------------------------
scan standard neighbor list and make it compatible with 1-5 neighbors
if IJ entry is a 1-2,1-3,1-4 neighbor then adjust offset to SBBITS15
else scan special15 to see if a 1-5 neighbor and adjust offset to SBBITS15
else do nothing to IJ entry
------------------------------------------------------------------------- */
__kernel void k_special15(__global int * dev_nbor,
const __global int * dev_packed,
const __global tagint *restrict tag,
const __global int *restrict nspecial15,
const __global tagint *restrict special15,
const int inum, const int nall, const int nbor_pitch,
const int t_per_atom) {
int tid, ii, offset, n_stride, i;
atom_info(t_per_atom,ii,tid,offset);
if (ii<inum) {
int numj, nbor, nbor_end;
nbor_info(dev_nbor,dev_packed,nbor_pitch,t_per_atom,ii,offset,i,numj,
n_stride,nbor_end,nbor);
int n15 = nspecial15[ii];
for ( ; nbor<nbor_end; nbor+=n_stride) {
int sj=dev_packed[nbor];
int which = sj >> SBBITS & 3;
int j = sj & NEIGHMASK;
tagint jtag = tag[j];
if (!which) {
int offset=ii;
for (int k=0; k<n15; k++) {
if (special15[offset] == jtag) {
which = 4;
break;
}
offset += nall;
}
}
if (which) dev_nbor[nbor] = j ^ (which << SBBITS15);
} // for nbor
} // if ii
}
__kernel void k_amoeba_short_nbor(const __global numtyp4 *restrict x_,
const __global int * dev_nbor,
const __global int * dev_packed,

2
lib/gpu/lal_amoeba_ext.cpp
View File

@ -52,7 +52,7 @@ int amoeba_gpu_init(const int ntypes, const int max_amtype, const int max_amclas
int gpu_rank=AMOEBAMF.device->gpu_rank();
int procs_per_gpu=AMOEBAMF.device->procs_per_gpu();
tep_size=sizeof(PRECISION);
tep_size=sizeof(ACC_PRECISION); // tep_size=sizeof(PRECISION);
AMOEBAMF.device->init_message(screen,"amoeba",first_gpu,last_gpu);

2
lib/gpu/lal_base_amoeba.h
View File

@ -235,7 +235,7 @@ class BaseAmoeba {
double** uind, double** uinp);
/// Per-atom arrays
UCL_Vector<numtyp,numtyp> _tep, _fieldp;
UCL_Vector<acctyp,acctyp> _tep, _fieldp;
int _nmax, _max_tep_size, _max_fieldp_size;
// ------------------------ FORCE/ENERGY DATA -----------------------