jianmu
/
lammps
forked from lijiext/lammps
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

workaround CUDA View::reference_type 

it is not an lvalue reference in CUDA.
also, the previous code assumed contiguous
entries for one atom; now it should be
robust in the case of LayoutLeft.
This commit is contained in:
Dan Ibanez 2016-12-21 10:21:55 -07:00
parent f3327ca214
commit 7559bc9c5f
1 changed files with 7 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
src/KOKKOS/fix_momentum_kokkos.cpp
View File

@ -149,12 +149,14 @@ void FixMomentumKokkos<DeviceType>::end_of_step()
auto triclinic = domain->triclinic;
Kokkos::parallel_for(nlocal, LAMMPS_LAMBDA(int i) {
if (mask[i] & groupbit2) {
double dx,dy,dz;
auto x_i = Few<double,3>(&x(i,0));
Few<double,3> x_i;
x_i[0] = x(i,0);
x_i[1] = x(i,1);
x_i[2] = x(i,2);
auto unwrap = DomainKokkos::unmap(prd,h,triclinic,x_i,image(i));
dx = unwrap[0] - xcm[0];
dy = unwrap[1] - xcm[1];
dz = unwrap[2] - xcm[2];
auto dx = unwrap[0] - xcm[0];
auto dy = unwrap[1] - xcm[1];
auto dz = unwrap[2] - xcm[2];
v(i,0) -= omega[1]*dz - omega[2]*dy;
v(i,1) -= omega[2]*dx - omega[0]*dz;
v(i,2) -= omega[0]*dy - omega[1]*dx;