added some cuda stuff

studouglas · Nov 16, 2015 · bea3b45 · bea3b45
1 parent 497e0e7
commit bea3b45
Show file tree

Hide file tree

Showing 12 changed files with 30 additions and 29 deletions.
diff --git a/.gitignore b/.gitignore
@@ -26,6 +26,7 @@ G4STORK/SourceFiles
 
 geant4.10.00.p02/tmp
 geant4.10.00.p02-build/*
+geant4.10.00.p02-build2
 geant4.10.00.p02-install/*
 geant4.10.00.p02/examples/*
 !geant4.10.00.p02/examples/basic/B1

diff --git a/CUDA-test/CMakeLists.txt → CUDA-test/src/CMakeLists.txt b/CUDA-test/CMakeLists.txt → CUDA-test/src/CMakeLists.txt
@@ -1,15 +1,16 @@
 cmake_minimum_required (VERSION 2.8)
-project(TestingCudaSquareArray)
+project(MainExe)
 
 find_package (CUDA REQUIRED)
 
 include_directories(/usr/local/cuda/include)
 
-cuda_add_executable (TestingCudaSquareArray DeviceMain.h DeviceMain.cu HostMain.cpp)
+cuda_add_library (TestingCudaSquareArrayLib DeviceMain.h DeviceMain.cu)
+add_executable (MainExe HostMain.cpp)
 
 LIST(APPEND CMAKE_CXX_FLAGS "-std=c++0x -O3 -ffast-math -Wall")
 
 LIST(APPEND CUDA_NVCC_FLAGS --compiler-options -fno-strict-aliasing -lineinfo -use_fast_math -Xptxas -dlcm=cg)
 LIST(APPEND CUDA_NVCC_FLAGS -gencode arch=compute_30,code=sm_30)
 
-target_link_libraries(TestingCudaSquareArray /usr/local/cuda/lib/libcudart_static.a)
+target_link_libraries(MainExe TestingCudaSquareArrayLib /usr/local/cuda/lib/libcudart_static.a)
diff --git a/CUDA-test/DeviceMain.cu → CUDA-test/src/DeviceMain.cu b/CUDA-test/DeviceMain.cu → CUDA-test/src/DeviceMain.cu
diff --git a/CUDA-test/DeviceMain.h → CUDA-test/src/DeviceMain.h b/CUDA-test/DeviceMain.h → CUDA-test/src/DeviceMain.h
diff --git a/CUDA-test/HostMain.cpp → CUDA-test/src/HostMain.cpp b/CUDA-test/HostMain.cpp → CUDA-test/src/HostMain.cpp
diff --git a/geant4.10.00.p02/CMakeLists.txt b/geant4.10.00.p02/CMakeLists.txt
@@ -34,22 +34,6 @@ set(CMAKE_USER_MAKE_RULES_OVERRIDE_CXX
 #
 project(Geant4)
 
-
-#----------------------------------------------------------------------------
-if (GEANT4_CUDA_ENABLED)
-    find_package(CUDA)
-    set(CUDA_NVCC_FLAGS ${CUDA_NVCC_FLAGS};-gencode arch=compute_20,code=sm_20)
-    cuda_compile(VECTOR_O source/cuda/CUDA_G4NeutronHPVector.cu)
-    cuda_add_executable(
-	cuda_g4neutronhpvector
-	source/cuda/CUDA_G4NeutronHPVector.cu
-	source/cuda/CUDA_G4NeutronHPVector.h
-	source/processes/hadronic/models/neutron_hp/src/G4NeutronHPVector.cc
-	source/processes/hadronic/models/neutron_hp/include/G4NeutronHPVector.hh)
-endif()
-#----------------------------------------------------------------------------
-
-
 #----------------------------------------------------------------------------
 # - Prepend our own CMake Modules to the search path
 # This allows us to customize, add, and factor out functionality

diff --git a/geant4.10.00.p02/source/cuda/CUDA_TestSquareArray.h b/geant4.10.00.p02/source/cuda/CUDA_TestSquareArray.h
diff --git a/geant4.10.00.p02/source/processes/hadronic/models/neutron_hp/CMakeLists.txt b/geant4.10.00.p02/source/processes/hadronic/models/neutron_hp/CMakeLists.txt
@@ -11,6 +11,8 @@
 #
 #------------------------------------------------------------------------------
 
+add_subdirectory(cuda)
+
 if(GEANT4_BUILD_GRANULAR_LIBS)
     include(Geant4MacroLibraryTargets)
     GEANT4_GRANULAR_LIBRARY_TARGET(COMPONENT sources.cmake)

diff --git a/geant4.10.00.p02/source/processes/hadronic/models/neutron_hp/cuda/CMakeLists.txt b/geant4.10.00.p02/source/processes/hadronic/models/neutron_hp/cuda/CMakeLists.txt
@@ -0,0 +1,13 @@
+cmake_minimum_required (VERSION 2.8)
+
+find_package (CUDA REQUIRED)
+
+include_directories(/usr/local/cuda/include)
+
+cuda_add_library (CUDA_G4NeutronHPVector SHARED CUDA_G4NeutronHPVector.h CUDA_G4NeutronHPVector.cu)
+
+#LIST(APPEND CMAKE_CXX_FLAGS "-std=c++0x -O3 -ffast-math -Wall")
+LIST(APPEND CUDA_NVCC_FLAGS --compiler-options -fno-strict-aliasing -lineinfo -use_fast_math -Xptxas -dlcm=cg)
+LIST(APPEND CUDA_NVCC_FLAGS -gencode arch=compute_30,code=sm_30)
+
+#target_link_libraries(Geant4 CUDA_G4NeutronHPVector /usr/local/cuda/lib/libcudart_static.a)
diff --git a/....p02/source/cuda/CUDA_Test_SquareArray.cu → ...neutron_hp/cuda/CUDA_G4NeutronHPVector.cu b/....p02/source/cuda/CUDA_Test_SquareArray.cu → ...neutron_hp/cuda/CUDA_G4NeutronHPVector.cu
@@ -38,4 +38,4 @@ float squareArray(int N) {
   	cudaFree(a_d);
 
     return a_h[N-1];
-}
+}
diff --git a/geant4.10.00.p02/source/processes/hadronic/models/neutron_hp/cuda/CUDA_G4NeutronHPVector.h b/geant4.10.00.p02/source/processes/hadronic/models/neutron_hp/cuda/CUDA_G4NeutronHPVector.h
@@ -0,0 +1 @@
+float squareArray(int N);
diff --git a/geant4.10.00.p02/source/processes/hadronic/models/neutron_hp/src/G4NeutronHPVector.cc b/geant4.10.00.p02/source/processes/hadronic/models/neutron_hp/src/G4NeutronHPVector.cc
@@ -33,10 +33,10 @@
 #include "G4NeutronHPVector.hh"
 #include "G4SystemOfUnits.hh"
 
-#define CUDA_ENABLED 1
-#if CUDA_ENABLED
-#include "CUDA_G4NeutronHPVector.h"
-#endif
+//#define CUDA_ENABLED 1
+//#if CUDA_ENABLED
+//#include "CUDA_G4NeutronHPVector.h"
+//#endif
 
   // if the ranges do not match, constant extrapolation is used.
   G4NeutronHPVector & operator + (G4NeutronHPVector & left, G4NeutronHPVector & right)
@@ -291,10 +291,10 @@
   void G4NeutronHPVector::ThinOut(G4double precision)
   {
 
-#if CUDA_ENABLED
-      float result = squareArray(100);
-      G4cout <<"G4NeutronHPVector::ThinOut test on GPU result: "<< result << G4endl;
-#endif
+//#if CUDA_ENABLED
+//      float result = squareArray(100);
+//      G4cout <<"G4NeutronHPVector::ThinOut test on GPU result: "<< result << G4endl;
+//#endif
 
     // anything in there?
     if(GetVectorLength()==0) return;
-Original file line number
+Diff line change
@@ Expand Up / @@ -38,4 +38,4 @@ float squareArray(int N) { @@
       	cudaFree(a_d);
         return a_h[N-1];
-    }
+    }