Cmake option (#98)

* add cmake option for pulsar * remove unused var * add CI compilation * eol * explicit amrex space, use literals in namepsace
RevathiJambunathan · May 19, 2022 · 3b8b35b · 3b8b35b
1 parent 48d0e4a
commit 3b8b35b
Show file tree

Hide file tree

Showing 7 changed files with 79 additions and 28 deletions.
diff --git a/.github/workflows/cuda.yml b/.github/workflows/cuda.yml
@@ -76,6 +76,21 @@ jobs:
         PYWARPX_LIB_DIR=$PWD/build_sp/lib python3 -m pip wheel .
         python3 -m pip install *.whl
 
+        cmake -S . -B build_pulsar      \
+          -DCMAKE_VERBOSE_MAKEFILE=ON   \
+          -DWarpX_COMPUTE=CUDA          \
+          -DWarpX_EB=OFF                \
+          -DWarpX_LIB=OFF               \
+          -DAMReX_CUDA_ARCH=6.0         \
+          -DWarpX_OPENPMD=OFF           \
+          -DWarpX_openpmd_internal=OFF  \
+          -DWarpX_PRECISION=DOUBLE      \
+          -DWarpX_PSATD=OFF             \
+          -DPULSAR=ON                   \
+          -DAMReX_CUDA_ERROR_CROSS_EXECUTION_SPACE_CALL=ON \
+          -DAMReX_CUDA_ERROR_CAPTURE_THIS=ON
+        cmake --build build_sp -j 2
+
   # make sure legacy build system continues to build, i.e., that we don't forget
   # to add new .cpp files
   build_nvcc_gnumake:
@@ -108,6 +123,7 @@ jobs:
         git clone https://github.com/AMReX-Codes/amrex.git ../amrex
         cd amrex && git checkout --detach 843a7dff266273a0f5b7b9f6cc9233a278f41fe1 && cd -
         make COMP=gcc QED=FALSE USE_MPI=TRUE USE_GPU=TRUE USE_OMP=FALSE USE_PSATD=TRUE USE_CCACHE=TRUE -j 2
+        make COMP=gcc QED=FALSE USE_OPENPMD=FALSE USE_MPI=TRUE USE_GPU=TRUE USE_OMP=FALSE USE_PSATD=FALSE PULSAR=TRUE USE_CCACHE=TRUE -j 2
 
   build_nvhpc21-11-nvcc:
     name: [email protected] NVCC/NVC++ Release [tests]
@@ -165,6 +181,20 @@ jobs:
           -DAMReX_CUDA_ERROR_CAPTURE_THIS=ON
         cmake --build build -j 2
 
+        cmake -S . -B build_pulsar     \
+          -DCMAKE_VERBOSE_MAKEFILE=ON  \
+          -DWarpX_COMPUTE=CUDA         \
+          -DWarpX_EB=OFF               \
+          -DWarpX_LIB=ON               \
+          -DAMReX_CUDA_ARCH=8.0        \
+          -DWarpX_OPENPMD=OFF          \
+          -DWarpX_PSATD=OFF            \
+          -DPULSAR=ON                  \
+          -DAMReX_CUDA_ERROR_CROSS_EXECUTION_SPACE_CALL=ON \
+          -DAMReX_CUDA_ERROR_CAPTURE_THIS=ON
+        cmake --build build -j 2
+
+
         # work-around for mpi4py 3.1.1 build system issue with using
         # a GNU-built Python executable with non-GNU Python modules
         #   https://github.com/mpi4py/mpi4py/issues/114

diff --git a/.github/workflows/ubuntu.yml b/.github/workflows/ubuntu.yml
@@ -38,6 +38,13 @@ jobs:
           -DWarpX_MPI=OFF              \
           -DWarpX_QED=OFF
         cmake --build build_3D -j 2
+        cmake -S . -B build_3D_PULSAR  \
+          -DCMAKE_VERBOSE_MAKEFILE=ON  \
+          -DWarpX_EB=OFF               \
+          -DWarpX_MPI=OFF              \
+          -DWarpX_QED=OFF              \
+          -DPULSAR=ON
+        cmake --build build_3D_PULSAR -j 2
         cmake -S . -B build_RZ         \
           -DCMAKE_VERBOSE_MAKEFILE=ON  \
           -DWarpX_DIMS=RZ              \

diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -71,6 +71,7 @@ option(WarpX_PSATD         "spectral solver support"                    OFF)
 option(WarpX_SENSEI        "SENSEI in situ diagnostics"                 OFF)
 option(WarpX_QED           "QED support (requires PICSAR)"                    ON)
 option(WarpX_QED_TABLE_GEN "QED table generation (requires PICSAR and Boost)" OFF)
+option(PULSAR              "pulsar simulations support"                 OFF)
 
 set(WarpX_DIMS_VALUES 1 2 3 RZ)
 set(WarpX_DIMS 3 CACHE STRING "Simulation dimensionality (1/2/3/RZ)")
@@ -268,6 +269,10 @@ if(WarpX_QED)
     target_link_libraries(ablastr PUBLIC PXRMP_QED::PXRMP_QED)
 endif()
 
+if(PULSAR)
+    target_compile_definitions(WarpX PUBLIC PULSAR)
+endif()
+
 # AMReX helper function: propagate CUDA specific target & source properties
 if(WarpX_COMPUTE STREQUAL CUDA)
     foreach(warpx_tgt IN LISTS _ALL_TARGETS)
@@ -321,6 +326,10 @@ if(WarpX_QED)
     endif()
 endif()
 
+if(PULSAR)
+    target_compile_definitions(WarpX PUBLIC PULSAR)
+endif()
+
 if(WarpX_PSATD)
     target_compile_definitions(ablastr PUBLIC WARPX_USE_PSATD)
 endif()

diff --git a/Source/Particles/CMakeLists.txt b/Source/Particles/CMakeLists.txt
@@ -8,6 +8,7 @@ target_sources(WarpX
     WarpXParticleContainer.cpp
     LaserParticleContainer.cpp
     ParticleBoundaryBuffer.cpp
+    PulsarParameters.cpp
 )
 
 #add_subdirectory(Algorithms)

diff --git a/Source/Particles/PulsarParameters.H b/Source/Particles/PulsarParameters.H
@@ -12,7 +12,7 @@
 #include <AMReX_Parser.H>
 #include <list>
 
-using namespace amrex;
+using namespace amrex::literals;
 
 
 class Pulsar {

diff --git a/Source/Particles/PulsarParameters.cpp b/Source/Particles/PulsarParameters.cpp
@@ -408,7 +408,7 @@ Pulsar::InitializeConductorMultifabUsingParser(
     const amrex::GpuArray<amrex::Real, AMREX_SPACEDIM> dx_lev = warpx.Geom(lev).CellSizeArray();
     const amrex::RealBox& real_box = warpx.Geom(lev).ProbDomain();
     amrex::IntVect iv = mf->ixType().toIntVect();
-    for ( amrex::MFIter mfi(*mf, TilingIfNotGPU()); mfi.isValid(); ++mfi) {
+    for ( amrex::MFIter mfi(*mf, amrex::TilingIfNotGPU()); mfi.isValid(); ++mfi) {
         //InitializeGhost Cells also
         const amrex::Box& tb = mfi.tilebox(iv, mf->nGrowVect());
         amrex::Array4<amrex::Real> const& conductor_fab = mf->array(mfi);
@@ -448,10 +448,10 @@ Pulsar::InitializeExternalPulsarFieldsOnGrid ( amrex::MultiFab *mfx, amrex::Mult
     amrex::IntVect x_nodal_flag = mfx->ixType().toIntVect();
     amrex::IntVect y_nodal_flag = mfy->ixType().toIntVect();
     amrex::IntVect z_nodal_flag = mfz->ixType().toIntVect();
-    GpuArray<int, 3> x_IndexType;
-    GpuArray<int, 3> y_IndexType;
-    GpuArray<int, 3> z_IndexType;
-    GpuArray<amrex::Real, AMREX_SPACEDIM> center_star_arr;
+    amrex::GpuArray<int, 3> x_IndexType;
+    amrex::GpuArray<int, 3> y_IndexType;
+    amrex::GpuArray<int, 3> z_IndexType;
+    amrex::GpuArray<amrex::Real, AMREX_SPACEDIM> center_star_arr;
     for (int idim = 0; idim < 3; ++idim) {
         x_IndexType[idim] = x_nodal_flag[idim];
         y_IndexType[idim] = y_nodal_flag[idim];
@@ -471,7 +471,7 @@ Pulsar::InitializeExternalPulsarFieldsOnGrid ( amrex::MultiFab *mfx, amrex::Mult
 #ifdef AMREX_USE_OMP
 #pragma omp parallel if (amrex::Gpu::notInLaunchRegion())
 #endif
-    for (MFIter mfi(*mfx, TilingIfNotGPU()); mfi.isValid(); ++mfi)
+    for (amrex::MFIter mfi(*mfx, amrex::TilingIfNotGPU()); mfi.isValid(); ++mfi)
     {
         const amrex::Box& tbx = mfi.tilebox(x_nodal_flag, mfx->nGrowVect() );
         const amrex::Box& tby = mfi.tilebox(y_nodal_flag, mfy->nGrowVect() );
@@ -658,10 +658,10 @@ Pulsar::ApplyCorotatingEfield_BC ( std::array< std::unique_ptr<amrex::MultiFab>,
     amrex::IntVect x_nodal_flag = Efield[0]->ixType().toIntVect();
     amrex::IntVect y_nodal_flag = Efield[1]->ixType().toIntVect();
     amrex::IntVect z_nodal_flag = Efield[2]->ixType().toIntVect();
-    GpuArray<int, 3> x_IndexType;
-    GpuArray<int, 3> y_IndexType;
-    GpuArray<int, 3> z_IndexType;
-    GpuArray<amrex::Real, AMREX_SPACEDIM> center_star_arr;
+    amrex::GpuArray<int, 3> x_IndexType;
+    amrex::GpuArray<int, 3> y_IndexType;
+    amrex::GpuArray<int, 3> z_IndexType;
+    amrex::GpuArray<amrex::Real, AMREX_SPACEDIM> center_star_arr;
     for (int idim = 0; idim < 3; ++idim) {
         x_IndexType[idim] = x_nodal_flag[idim];
         y_IndexType[idim] = y_nodal_flag[idim];
@@ -679,7 +679,7 @@ Pulsar::ApplyCorotatingEfield_BC ( std::array< std::unique_ptr<amrex::MultiFab>,
 #ifdef AMREX_USE_OMP
 #pragma omp parallel if (amrex::Gpu::notInLaunchRegion())
 #endif
-    for (MFIter mfi(*Efield[0], TilingIfNotGPU()); mfi.isValid(); ++mfi)
+    for (amrex::MFIter mfi(*Efield[0], amrex::TilingIfNotGPU()); mfi.isValid(); ++mfi)
     {
         const amrex::Box& tex = mfi.tilebox(x_nodal_flag);
         const amrex::Box& tey = mfi.tilebox(y_nodal_flag);
@@ -811,10 +811,10 @@ Pulsar::ApplyDipoleBfield_BC ( std::array< std::unique_ptr<amrex::MultiFab>, 3>
     amrex::IntVect x_nodal_flag = Bfield[0]->ixType().toIntVect();
     amrex::IntVect y_nodal_flag = Bfield[1]->ixType().toIntVect();
     amrex::IntVect z_nodal_flag = Bfield[2]->ixType().toIntVect();
-    GpuArray<int, 3> x_IndexType;
-    GpuArray<int, 3> y_IndexType;
-    GpuArray<int, 3> z_IndexType;
-    GpuArray<amrex::Real, AMREX_SPACEDIM> center_star_arr;
+    amrex::GpuArray<int, 3> x_IndexType;
+    amrex::GpuArray<int, 3> y_IndexType;
+    amrex::GpuArray<int, 3> z_IndexType;
+    amrex::GpuArray<amrex::Real, AMREX_SPACEDIM> center_star_arr;
     for (int idim = 0; idim < 3; ++idim) {
         x_IndexType[idim] = x_nodal_flag[idim];
         y_IndexType[idim] = y_nodal_flag[idim];
@@ -832,7 +832,7 @@ Pulsar::ApplyDipoleBfield_BC ( std::array< std::unique_ptr<amrex::MultiFab>, 3>
 #ifdef AMREX_USE_OMP
 #pragma omp parallel if (amrex::Gpu::notInLaunchRegion())
 #endif
-    for (MFIter mfi(*Bfield[0], TilingIfNotGPU()); mfi.isValid(); ++mfi)
+    for (amrex::MFIter mfi(*Bfield[0], amrex::TilingIfNotGPU()); mfi.isValid(); ++mfi)
     {
         const amrex::Box& tbx = mfi.tilebox(x_nodal_flag);
         const amrex::Box& tby = mfi.tilebox(y_nodal_flag);
@@ -1013,7 +1013,7 @@ Pulsar::SetTangentialEforInternalConductor( std::array <std::unique_ptr<amrex::M
 #ifdef AMREX_USE_OMP
 #pragma omp parallel if (amrex::Gpu::notInLaunchRegion())
 #endif
-    for (MFIter mfi(*Efield[0], TilingIfNotGPU()); mfi.isValid(); ++mfi)
+    for (amrex::MFIter mfi(*Efield[0], amrex::TilingIfNotGPU()); mfi.isValid(); ++mfi)
     {
         const amrex::Box& tex = mfi.tilebox(x_nodal_flag);
         const amrex::Box& tey = mfi.tilebox(y_nodal_flag);
@@ -1093,7 +1093,7 @@ Pulsar::ComputePlasmaNumberDensity ()
         }
 
 
-        const Geometry& geom = warpx.Geom(lev);
+        const amrex::Geometry& geom = warpx.Geom(lev);
         const auto dx = geom.CellSizeArray();
 #if defined WARPX_DIM_3D
         amrex::Real inv_vol = 1._rt/(dx[0] * dx[1] * dx[2]);
@@ -1104,7 +1104,7 @@ Pulsar::ComputePlasmaNumberDensity ()
 #ifdef AMREX_USE_OMP
 #pragma omp parallel if (amrex::Gpu::notInLaunchRegion())
 #endif
-        for ( MFIter mfi(*m_plasma_number_density[lev], TilingIfNotGPU()); mfi.isValid(); ++mfi ) {
+        for ( amrex::MFIter mfi(*m_plasma_number_density[lev], amrex::TilingIfNotGPU()); mfi.isValid(); ++mfi ) {
             amrex::Array4<amrex::Real> const& density = m_plasma_number_density[0]->array(mfi);
             amrex::Box const& tbx = mfi.tilebox();
             const int ncomps = m_plasma_number_density[lev]->nComp();
@@ -1132,7 +1132,7 @@ Pulsar::ComputePlasmaMagnetization ()
         const amrex::MultiFab& Bx_mf = warpx.getBfield(lev, 0);
         const amrex::MultiFab& By_mf = warpx.getBfield(lev, 1);
         const amrex::MultiFab& Bz_mf = warpx.getBfield(lev, 2);
-        for (MFIter mfi(*m_magnetization[lev], TilingIfNotGPU()); mfi.isValid(); ++mfi)
+        for (amrex::MFIter mfi(*m_magnetization[lev], amrex::TilingIfNotGPU()); mfi.isValid(); ++mfi)
         {
             const amrex::Box& bx = mfi.tilebox();
             amrex::Array4<amrex::Real> const& mag = m_magnetization[lev]->array(mfi);
@@ -1191,7 +1191,7 @@ Pulsar::TuneSigma0Threshold (const int step)
     for (int isp = 0; isp < nspecies; ++isp) {
         amrex::Real ws_total = 0._rt;
         auto& pc = warpx.GetPartContainer().GetParticleContainer(isp);
-        amrex::ReduceOps<ReduceOpSum> reduce_ops;
+        amrex::ReduceOps<amrex::ReduceOpSum> reduce_ops;
         amrex::Real cur_time = warpx.gett_new(0);
         auto ws_r = amrex::ParticleReduce<
                         amrex::ReduceData < amrex::ParticleReal> >
@@ -1292,7 +1292,7 @@ Pulsar::TotalParticles ()
     for (int isp = 0; isp < nspecies; ++isp) {
         auto& pc = warpx.GetPartContainer().GetParticleContainer(isp);
         amrex::Long np_total = pc.TotalNumberOfParticles();
-        amrex::ParallelDescriptor::ReduceLongSum(np_total, ParallelDescriptor::IOProcessorNumber());
+        amrex::ParallelDescriptor::ReduceLongSum(np_total);
         total_particles += np_total;
     }
 }
@@ -1308,7 +1308,6 @@ Pulsar::PrintInjectedCellValues ()
 {
     auto& warpx = WarpX::GetInstance();
     std::vector species_names = warpx.GetPartContainer().GetSpeciesNames();
-    const int nspecies = species_names.size();
     int total_injected_cells = static_cast<int>(SumInjectionFlag());
     // x, y, z, r, theta, phi, injection_flag, magnetization, ndens_p, ndens_e, Bx, By, Bz, Bmag, rho
     int total_diags = 15;
@@ -1323,7 +1322,7 @@ Pulsar::PrintInjectedCellValues ()
     const amrex::MultiFab& injectionflag_mf = *m_injection_flag[lev];
     const amrex::MultiFab& magnetization_mf = *m_magnetization[lev];
     const amrex::MultiFab& ndens_mf = *m_plasma_number_density[lev];
-    GpuArray<amrex::Real, AMREX_SPACEDIM> center_star_arr;
+    amrex::GpuArray<amrex::Real, AMREX_SPACEDIM> center_star_arr;
     for (int idim = 0; idim < 3; ++idim) {
         center_star_arr[idim] = m_center_star[idim];
     }
@@ -1332,9 +1331,9 @@ Pulsar::PrintInjectedCellValues ()
     rho = mypc.GetChargeDensity(lev, true);
     amrex::MultiFab & rho_mf = *rho;
 
-    Gpu::DeviceScalar<int> cell_counter(0);
+    amrex::Gpu::DeviceScalar<int> cell_counter(0);
     int* cell_counter_d = cell_counter.dataPtr();
-    for (MFIter mfi(injectionflag_mf, TilingIfNotGPU()); mfi.isValid(); ++mfi)
+    for (amrex::MFIter mfi(injectionflag_mf, amrex::TilingIfNotGPU()); mfi.isValid(); ++mfi)
     {
         const amrex::Box & bx = mfi.tilebox();
         amrex::Array4<const amrex::Real> const& Bx = Bx_mf[mfi].array();
@@ -1386,7 +1385,7 @@ Pulsar::PrintInjectedCellValues ()
     }
     amrex::Print() << " counter : " << cell_counter.dataValue() << " total cells injected " << total_injected_cells << "\n";
     std::stringstream ss;
-    ss << Concatenate("InjectionCellData", warpx.getistep(0), 5);
+    ss << amrex::Concatenate("InjectionCellData", warpx.getistep(0), 5);
     amrex::AllPrintToFile(ss.str()) << " cell_index x y z r theta phi injection magnetization ndens_p ndens_e Bx By Bz Bmag rho \n" ;
     for (int icell = 0; icell < total_injected_cells; ++icell ) {
         if (InjectedCellDiagData[icell*total_diags + 6] == 1) {

diff --git a/cmake/WarpXFunctions.cmake b/cmake/WarpXFunctions.cmake
@@ -220,6 +220,10 @@ function(set_warpx_binary_name)
             set_property(TARGET ${tgt} APPEND_STRING PROPERTY OUTPUT_NAME ".PSATD")
         endif()
 
+        if(PULSAR)
+            set_property(TARGET ${tgt} APPEND_STRING PROPERTY OUTPUT_NAME ".PULSAR")
+        endif()
+
         if(WarpX_EB)
             set_property(TARGET ${tgt} APPEND_STRING PROPERTY OUTPUT_NAME ".EB")
         endif()
@@ -373,6 +377,7 @@ function(warpx_print_summary)
         message("    MPI (thread multiple): ${WarpX_MPI_THREAD_MULTIPLE}")
     endif()
     message("    PSATD: ${WarpX_PSATD}")
+    message("    PULSAR: ${PULSAR}")
     message("    PRECISION: ${WarpX_PRECISION}")
     message("    OPENPMD: ${WarpX_OPENPMD}")
     message("    QED: ${WarpX_QED}")