#  Makefile for fltk applications
#

TARGET=mm


CUDA_INSTALL=/usr/local/cuda
CUDA_BIN=/usr/local/cuda/bin
CUDA_INC=/usr/local/cuda/include
CUDA_LIB=/usr/local/cuda/lib
CUDASDK_DIR=/Developer/"GPU Computing"

LOCAL = /usr/local

CC = c++

NVCC =$(CUDA_BIN)/nvcc 
#NVCC =nvcc -deviceemu
NVCC_INC+=-I $(CUDA_INC)
NVCC_INC+=-I $(CUDASDK_DIR)/C/common/inc
#LIBS  = $(LDLIBS) $(GLDLIBS) -lm

CFLAGS = -m32
LD_FLAGS = -m32 -L$(CUDA_LIB) -lcudart $(CUDASDK_DIR)/C/lib/libcutil_i386.a 
#LD_FLAGS =-m32 -L$(CUDA_LIB) -lcudartemu $(CUDASDK_DIR)/C/lib/libcutil_i386.a 


#.SUFFIXES: .o .cpp .cu .chpl 
#.SUFFIX: .o .cpp .cu .chpl 

%.o: %.c
	$(CC) $(CFLAGS) $(INCLUDE) $(NVCC_INC) -c -o $*.o $<

%.o: %.cu 
	$(NVCC) $(CFLAGS) $(NVCC_INC) -c -o $*.o $<

%.o: %.cpp
	$(CC) $(CFLAGS) $(INCLUDE) $(NVCC_INC) -c -o $*.o $<

OBJS = matrixmul.o matrixmul_gold.o
C_HEADERS = matrixmul_kernel.cu

all: $(TARGET)


$(TARGET): $(OBJS) $(C_HEADERS) 
	$(CC) $(LD_FLAGS) -o $@ $(OBJS) 


clean:
	rm -f $(OBJS) $(TARGET) 

clean_all:
	rm -f $(OBJS) $(TARGET)