Skip to content

Improve EMParticleContainer::InitParticles #166

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Nov 6, 2025
Merged

Improve EMParticleContainer::InitParticles #166

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
1cceded
Improve EMParticleContainer::InitParticles
AlexanderSinn Nov 6, 2025
03b0b0b
fix ;
AlexanderSinn Nov 6, 2025
ac6b10f
Update ExampleCodes/Particles/ElectromagneticPIC/Source/EMParticleCon…
atmyers Nov 6, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
89 changes: 26 additions & 63 deletions ExampleCodes/Particles/ElectromagneticPIC/Source/EMParticleContainerInit.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -67,14 +67,13 @@ InitParticles(const IntVect& a_num_particles_per_cell,
{
const Box& tile_box = mfi.tilebox();

const auto lo = amrex::lbound(tile_box);
const auto hi = amrex::ubound(tile_box);

Gpu::DeviceVector<unsigned int> counts(tile_box.numPts(), 0);
unsigned int* pcount = counts.dataPtr();
Array4<unsigned int> counts_arr {counts.dataPtr(), amrex::begin(tile_box),
amrex::end(tile_box), 1};

Gpu::DeviceVector<unsigned int> offsets(tile_box.numPts());
unsigned int* poffset = offsets.dataPtr();
Array4<unsigned int> offsets_arr {offsets.dataPtr(), amrex::begin(tile_box),
amrex::end(tile_box), 1};

amrex::ParallelFor(tile_box,
[=] AMREX_GPU_DEVICE (int i, int j, int k) noexcept
Expand All @@ -93,63 +92,28 @@ InitParticles(const IntVect& a_num_particles_per_cell,
y >= a_bounds.hi(1) || y < a_bounds.lo(1) ||
z >= a_bounds.hi(2) || z < a_bounds.lo(2) ) continue;

int ix = i - lo.x;
int iy = j - lo.y;
int iz = k - lo.z;
int nx = hi.x-lo.x+1;
int ny = hi.y-lo.y+1;
int nz = hi.z-lo.z+1;
unsigned int uix = amrex::min(nx-1,amrex::max(0,ix));
unsigned int uiy = amrex::min(ny-1,amrex::max(0,iy));
unsigned int uiz = amrex::min(nz-1,amrex::max(0,iz));
unsigned int cellid = (uix * ny + uiy) * nz + uiz;
pcount[cellid] += 1;
counts_arr(i, j, k) += 1;
}
});

Gpu::exclusive_scan(counts.begin(), counts.end(), offsets.begin());

unsigned int last_offset;
unsigned int last_count;
#ifdef AMREX_USE_GPU
Gpu::dtoh_memcpy(&last_offset,offsets.dataPtr()+tile_box.numPts()-1,sizeof(unsigned int));
Gpu::dtoh_memcpy(&last_count,counts.dataPtr()+tile_box.numPts()-1,sizeof(unsigned int));
#else
std::memcpy(&last_offset,offsets.dataPtr()+tile_box.numPts()-1,sizeof(unsigned int)
std::memcpy(&last_count,counts.dataPtr()+tile_box.numPts()-1,sizeof(unsigned int)
#endif
int num_to_add = last_offset + last_count;
int num_to_add = Scan::ExclusiveSum(counts.size(), counts.data(), offsets.data());

auto& particles = GetParticles(lev);
auto& particle_tile = particles[std::make_pair(mfi.index(), mfi.LocalTileIndex())];
auto& particle_tile = DefineAndReturnParticleTile(lev, mfi.index(), mfi.LocalTileIndex());

auto old_size = particle_tile.GetArrayOfStructs().size();
auto new_size = old_size + num_to_add;
particle_tile.resize(new_size);

if (num_to_add == 0) continue;

ParticleType* pstruct = particle_tile.GetArrayOfStructs()().data();

auto arrdata = particle_tile.GetStructOfArrays().realarray();
auto ptd = particle_tile.getParticleTileData();

int procID = ParallelDescriptor::MyProc();

amrex::ParallelForRNG(tile_box,
[=] AMREX_GPU_DEVICE (int i, int j, int k, amrex::RandomEngine const& engine) noexcept
{
int ix = i - lo.x;
int iy = j - lo.y;
int iz = k - lo.z;
int nx = hi.x-lo.x+1;
int ny = hi.y-lo.y+1;
int nz = hi.z-lo.z+1;
unsigned int uix = amrex::min(nx-1,amrex::max(0,ix));
unsigned int uiy = amrex::min(ny-1,amrex::max(0,iy));
unsigned int uiz = amrex::min(nz-1,amrex::max(0,iz));
unsigned int cellid = (uix * ny + uiy) * nz + uiz;

int pidx = poffset[cellid] + old_size;
int pidx = offsets_arr(i, j, k) + old_size;

for (int i_part=0; i_part<num_ppc;i_part++)
{
Expand Down Expand Up @@ -179,24 +143,23 @@ InitParticles(const IntVect& a_num_particles_per_cell,
y >= a_bounds.hi(1) || y < a_bounds.lo(1) ||
z >= a_bounds.hi(2) || z < a_bounds.lo(2) ) continue;

ParticleType& p = pstruct[pidx];
p.id() = pidx + 1;
p.cpu() = procID;
p.pos(0) = x;
p.pos(1) = y;
p.pos(2) = z;

arrdata[PIdx::ux ][pidx] = u[0] * PhysConst::c;
arrdata[PIdx::uy ][pidx] = u[1] * PhysConst::c;
arrdata[PIdx::uz ][pidx] = u[2] * PhysConst::c;
arrdata[PIdx::w ][pidx] = a_density * scale_fac;
arrdata[PIdx::Ex ][pidx] = 0.0;
arrdata[PIdx::Ey ][pidx] = 0.0;
arrdata[PIdx::Ez ][pidx] = 0.0;
arrdata[PIdx::Bx ][pidx] = 0.0;
arrdata[PIdx::By ][pidx] = 0.0;
arrdata[PIdx::Bz ][pidx] = 0.0;
arrdata[PIdx::ginv][pidx] = 0.0;
ptd.id(pidx) = pidx + 1;
ptd.cpu(pidx) = procID;
ptd.pos(0, pidx) = x;
ptd.pos(1, pidx) = y;
ptd.pos(2, pidx) = z;

ptd.rdata(PIdx::ux )[pidx] = u[0] * PhysConst::c;
ptd.rdata(PIdx::uy )[pidx] = u[1] * PhysConst::c;
ptd.rdata(PIdx::uz )[pidx] = u[2] * PhysConst::c;
ptd.rdata(PIdx::w )[pidx] = a_density * scale_fac;
ptd.rdata(PIdx::Ex )[pidx] = 0.0;
ptd.rdata(PIdx::Ey )[pidx] = 0.0;
ptd.rdata(PIdx::Ez )[pidx] = 0.0;
ptd.rdata(PIdx::Bx )[pidx] = 0.0;
ptd.rdata(PIdx::By )[pidx] = 0.0;
ptd.rdata(PIdx::Bz )[pidx] = 0.0;
ptd.rdata(PIdx::ginv)[pidx] = 0.0;

++pidx;
}
Expand Down
Loading