a kernel need to be an instance of a functor

- remove `PMACC_TYPEKERNEL`
- use `PMACC_KERNEL`
- create the instance of a functor inside the user code

Author: psychocoderHPC

examples/SingleParticleCurrent/include/particles/ParticlesInitOneParticle.hpp

@@ -130,7 +130,7 @@ class ParticlesInitOneParticle
         std::cout << "localSuperCell: " << localSuperCell << std::endl;
         std::cout << "cellInSuperCell: " << cellInSuperCell << std::endl;
 
-        PMACC_TYPEKERNEL(kernelAddOneParticle)
+        PMACC_KERNEL(kernelAddOneParticle{})
             (1, 1)
             (parClass.getDeviceParticlesBox(),
              localSuperCell, cellInSuperCell);

examples/SingleParticleRadiationWithLaser/include/particles/ParticlesInitOneParticle.hpp

@@ -131,7 +131,7 @@ class ParticlesInitOneParticle
         localSuperCell = localSuperCell + cellDescription.getGuardingSuperCells();
 
 
-        PMACC_TYPEKERNEL(kernelAddOneParticle)
+        PMACC_KERNEL(kernelAddOneParticle{})
             (1, 1)
             (parClass.getDeviceParticlesBox(),
              localSuperCell, cellInSuperCell);

examples/SingleParticleTest/include/particles/ParticlesInitOneParticle.hpp

@@ -127,7 +127,7 @@ class ParticlesInitOneParticle
         localSuperCell = localSuperCell + cellDescription.getGuardingSuperCells();
 
 
-        PMACC_TYPEKERNEL(kernelAddOneParticle)
+        PMACC_KERNEL(kernelAddOneParticle{})
             (1, 1)
             (parClass.getDeviceParticlesBox(),
              localSuperCell, cellInSuperCell);

src/libPMacc/examples/gameOfLife2D/include/Evolution.hpp

@@ -83,7 +83,7 @@ namespace gol
                 buffWrite(blockCell + threadIndex) = isLife;
             }
         };
-        
+
         struct randomInit
         {
             template<class BoxWriteOnly, class Mapping>
@@ -136,7 +136,7 @@ namespace gol
             GridController<DIM2>& gc = Environment<DIM2>::get().GridController();
             uint32_t seed = gc.getGlobalSize() + gc.getGlobalRank();
 
-            PMACC_TYPEKERNEL(kernel::randomInit)
+            PMACC_KERNEL(kernel::randomInit{})
                     (mapper.getGridDim(), MappingDesc::SuperCellSize::toRT())
                     (
                      writeBox,
@@ -149,7 +149,7 @@ namespace gol
         void run(const DBox& readBox, const DBox & writeBox)
         {
             AreaMapping < Area, MappingDesc > mapper(mapping);
-            PMACC_TYPEKERNEL(kernel::evolution)
+            PMACC_KERNEL(kernel::evolution{})
                     (mapper.getGridDim(), MappingDesc::SuperCellSize::toRT())
                     (readBox,
                      writeBox,

src/libPMacc/include/cuSTL/algorithm/kernel/Foreach.hpp

@@ -66,7 +66,7 @@ namespace kernel
         auto blockDim = BlockDim::toRT();                                                                   \
         detail::SphericMapper<Zone::dim, BlockDim> mapper;                                                  \
         using namespace PMacc;                                                                              \
-        PMACC_TYPEKERNEL(detail::kernelForeach)(mapper.cudaGridDim(p_zone.size), blockDim)                  \
+        PMACC_KERNEL(detail::kernelForeach{})(mapper.cudaGridDim(p_zone.size), blockDim)                    \
                   /* c0_shifted, c1_shifted, ... */                                                         \
             (mapper, BOOST_PP_ENUM(N, SHIFTED_CURSOR, _), lambda::make_Functor(functor));                   \
     }

src/libPMacc/include/cuSTL/algorithm/kernel/ForeachBlock.hpp

@@ -94,7 +94,7 @@ BOOST_PP_REPEAT_FROM_TO(1, BOOST_PP_INC(FOREACH_KERNEL_MAX_PARAMS), KERNEL_FOREA
         auto blockDim = ThreadBlock::toRT();                                                                \
         detail::SphericMapper<Zone::dim, BlockDim> mapper;                                                  \
         using namespace PMacc;                                                                              \
-        PMACC_TYPEKERNEL(detail::kernelForeachBlock)(mapper.cudaGridDim(p_zone.size), blockDim)             \
+        PMACC_KERNEL(detail::kernelForeachBlock{})(mapper.cudaGridDim(p_zone.size), blockDim)               \
                     /* c0_shifted, c1_shifted, ... */                                                       \
             (mapper, BOOST_PP_ENUM(N, SHIFTED_CURSOR, _), lambda::make_Functor(functor));                   \
     }

src/libPMacc/include/cuSTL/algorithm/kernel/run-time/Foreach.hpp

@@ -142,7 +142,7 @@ math::Size_t<DIM3> getBestCudaBlockDim(const math::Size_t<dim> gridDim)
         );                                                                                                          \
         kernel::detail::SphericMapper<Zone::dim> mapper;                                                            \
         using namespace PMacc;                                                                                      \
-        PMACC_TYPEKERNEL(kernel::detail::kernelForeach)(mapper.cudaGridDim(p_zone.size, this->_blockDim), blockDim) \
+        PMACC_KERNEL(kernel::detail::kernelForeach{})(mapper.cudaGridDim(p_zone.size, this->_blockDim), blockDim)   \
                 /*   c0_shifted, ..., cN_shifted    */                                                              \
             (mapper, BOOST_PP_ENUM(N, SHIFTED_CURSOR, _), lambda::make_Functor(functor));                           \
     }

src/libPMacc/include/eventSystem/events/kernelEvents.hpp

@@ -279,12 +279,5 @@ namespace exec
  */
 #define PMACC_KERNEL( ... ) PMacc::exec::kernel( __VA_ARGS__, __FILE__,  static_cast< size_t >( __LINE__ ) )
 
-/** create a kernel object out of a functor type name
- *
- * this macro add the current filename and line number to the kernel object
- *
- * @param ... type of the kernel functor
- */
-#define PMACC_TYPEKERNEL( ... ) PMacc::exec::kernel( __VA_ARGS__{}, __FILE__,  static_cast< size_t >( __LINE__ ) )
 
 #include "eventSystem/events/kernelEvents.tpp"

src/libPMacc/include/nvidia/reduce/Reduce.hpp

@@ -62,7 +62,7 @@ namespace kernel
             Type* s_mem=(Type*)s_mem_extern;
 
 
-            
+
             bool isActive = (tid < src_count);
 
             if(isActive)
@@ -91,7 +91,7 @@ namespace kernel
                 /* New chunks is half number of chunks rounded up for uneven counts
                  * --> local_tid=0 will reduce the single element for an odd number of values at the end */
                 chunk_count = (chunk_count + 1) / 2;
-                
+
                 isActive = (tid < src_count) && !(localId != 0 && localId >= active_threads);
                 if(isActive)
                     func(s_mem[localId], s_mem[localId + chunk_count]);
@@ -154,7 +154,7 @@ namespace kernel
 
             uint32_t blocks = threads / 2 / blockcount;
             if (blocks == 0) blocks = 1;
-            PMACC_TYPEKERNEL(kernel::reduce < Type >)(blocks, blockcount, blockcount * sizeof (Type))(src, n, dest, func,
+            PMACC_KERNEL(kernel::reduce < Type >{})(blocks, blockcount, blockcount * sizeof (Type))(src, n, dest, func,
                                                                                                     PMacc::nvidia::functors::Assign());
             n = blocks;
             blockcount = optimalThreadsPerBlock(n, sizeof (Type));
@@ -171,13 +171,13 @@ namespace kernel
                     uint32_t problemSize = n - (blockOffset * blockcount);
                     Type* srcPtr = dest + (blockOffset * blockcount);
 
-                    PMACC_TYPEKERNEL(kernel::reduce < Type >)(useBlocks, blockcount, blockcount * sizeof (Type))(srcPtr, problemSize, dest, func, func);
+                    PMACC_KERNEL(kernel::reduce < Type >{})(useBlocks, blockcount, blockcount * sizeof (Type))(srcPtr, problemSize, dest, func, func);
                     blocks = blockOffset*blockcount;
                 }
                 else
                 {
 
-                    PMACC_TYPEKERNEL(kernel::reduce < Type >)(blocks, blockcount, blockcount * sizeof (Type))(dest, n, dest, func,
+                    PMACC_KERNEL(kernel::reduce < Type >{})(blocks, blockcount, blockcount * sizeof (Type))(dest, n, dest, func,
                                                                                                             PMacc::nvidia::functors::Assign());
                 }
 

src/libPMacc/include/particles/IdProvider.hpp

@@ -102,7 +102,7 @@ namespace PMacc {
     IdProvider<T_dim>::State IdProvider<T_dim>::getState()
     {
         HostDeviceBuffer<uint64_cu, 1> nextIdBuf(DataSpace<1>(1));
-        PMACC_TYPEKERNEL(idDetail::getNextId)(1, 1)(nextIdBuf.getDeviceBuffer().getDataBox());
+        PMACC_KERNEL(idDetail::getNextId{})(1, 1)(nextIdBuf.getDeviceBuffer().getDataBox());
         nextIdBuf.deviceToHost();
         State state;
         state.nextId = static_cast<uint64_t>(nextIdBuf.getHostBuffer().getDataBox()(0));
@@ -174,14 +174,14 @@ namespace PMacc {
     template<unsigned T_dim>
     void IdProvider<T_dim>::setNextId(uint64_t nextId)
     {
-        PMACC_TYPEKERNEL(idDetail::setNextId)(1, 1)(nextId);
+        PMACC_KERNEL(idDetail::setNextId{})(1, 1)(nextId);
     }
 
     template<unsigned T_dim>
     uint64_t IdProvider<T_dim>::getNewIdHost()
     {
         HostDeviceBuffer<uint64_cu, 1> newIdBuf(DataSpace<1>(1));
-        PMACC_TYPEKERNEL(idDetail::getNewId)(1, 1)(newIdBuf.getDeviceBuffer().getDataBox(), GetNewId());
+        PMACC_KERNEL(idDetail::getNewId{})(1, 1)(newIdBuf.getDeviceBuffer().getDataBox(), GetNewId());
         newIdBuf.deviceToHost();
         return static_cast<uint64_t>(newIdBuf.getHostBuffer().getDataBox()(0));
     }