Switched Implementation to OpenCL

Author: Ryan

calc.cpp

@@ -1,24 +1,19 @@
-CONFIG += c++11
-QT += widgets printsupport 3dcore 3drender 3dextras
-win32:QT += winextras
-
-win32:QMAKE_CXXFLAGS +=  /openmp
-unix: QMAKE_CXXFLAGS += -fopenmp
-unix: QMAKE_CFLAGS   += -fopenmp -std=c99
-
-RESOURCES += shaders.qrc
-HEADERS += *.h
-unix:SOURCES += shaders/calc.c
-win32:SOURCES += calc.cpp
-SOURCES += coloredlabel.cpp dialog.cpp dialogbutton.cpp directioncalculator.cpp main.cpp param.cpp paramgroup.cpp qcustomplot.cpp spacingcalculator.cpp spectrogram.cpp spectromanager.cpp thread.cpp tinyexpr.c vector3d.cpp window.cpp window3d.cpp
-
-#magic needed to make CUDA work
-DISTFILES += configure kernel.cu README.md LICENSE.txt debian/*
-LIBS += $$OUT_PWD/kernel$$QMAKE_EXT_OBJ -lcuda -lcudart_static
-win32:LIBS += -L$$(CUDA_PATH)/lib/x64
-unix:LIBS += -fopenmp -lrt -ldl
-
-kernel.target += $$OUT_PWD/kernel$$QMAKE_EXT_OBJ
-kernel.commands += nvcc -c -o $$OUT_PWD/kernel$$QMAKE_EXT_OBJ $$PWD/kernel.cu
-QMAKE_EXTRA_TARGETS += kernel
-PRE_TARGETDEPS += $$OUT_PWD/kernel$$QMAKE_EXT_OBJ
+CONFIG += c++11
+QT += widgets printsupport 3dcore 3drender 3dextras
+win32:QT += winextras
+
+win32:QMAKE_CXXFLAGS +=  /openmp
+unix: QMAKE_CXXFLAGS += -fopenmp
+unix: QMAKE_CFLAGS   += -fopenmp
+
+RESOURCES += shaders.qrc
+HEADERS += *.h
+win32:SOURCES += calc.cpp
+unix:SOURCES += shaders/calc.c
+SOURCES += coloredlabel.cpp dialog.cpp dialogbutton.cpp directioncalculator.cpp main.cpp param.cpp paramgroup.cpp qcustomplot.cpp spacingcalculator.cpp spectrogram.cpp spectromanager.cpp thread.cpp tinyexpr.c vector3d.cpp window.cpp window3d.cpp
+
+DISTFILES += configure README.md LICENSE.txt debian/*
+win32:LIBS += -L$$(CUDA_PATH)/lib/x64 -L$$(AMDAPPSDKROOT)/lib/x86_64 -L$$(INTELOCLSDKROOT)/lib64 -lOpenCL
+unix:LIBS += -fopenmp -lrt -ldl
+
+win32:INCLUDEPATH += $$(CUDA_PATH)/include $$(AMDAPPSDKROOT)/include $$(INTELOCLSDKROOT)/include

difsim.pro

@@ -61,15 +61,7 @@ extern int g_max_gpus;//global maximum number of GPUs to use in calculation
 int get_num_procs_omp();
 size_t get_num_iterations();
 
-int getDeviceCountCuda();
-int getMaxThreadsPerBlockCuda();
-void setDeviceCuda(int);
-void*mallocCuda(size_t);
-void freeCuda(void*);
-void memsetCuda(void*,int,size_t);
-void kernelCuda(size_t,size_t,vector*,void*,double,double,vector,double,double);
-void memcpyHostToDeviceCuda(void*,const void*,size_t);
-void memcpyDeviceToHostCuda(void*,const void*,size_t);
+size_t getDeviceCountOpenCL();
 
 vector vec_unit(vector);
 vector vec_num(double,vector);