# set LD_LIBRARY_PATH
export CC  = gcc
export CXX = g++
export NVCC =nvcc
export CFLAGS = -Wall -O3 -g -msse3 -Wno-unknown-pragmas -funroll-loops -I../
export LDFLAGS= -g -lm -lcublas -lcudart
export NVCCFLAGS = -O3 --use_fast_math -ccbin $(CXX)

# specify tensor path
BIN = test_tblob
OBJ =
CUOBJ =
CUBIN = test
.PHONY: clean all

all: $(CUBIN) $(BIN)

test: test.cu

test_tblob: test_tblob.cc

$(BIN) :
	$(CXX) $(CFLAGS) -std=c++0x -o $@ $(filter %.cpp %.o %.c %.cc, $^)  $(LDFLAGS)

$(OBJ) :
	$(CXX) -c $(CFLAGS) -o $@ $(firstword $(filter %.cpp %.c, $^) )

$(CUOBJ) :
	$(NVCC) -c -o $@ $(NVCCFLAGS) -Xcompiler "$(CFLAGS)" $(filter %.cu, $^)

$(CUBIN) :
	$(NVCC) -o $@ $(NVCCFLAGS) -Xcompiler "$(CFLAGS)" -Xlinker "$(LDFLAGS)" $(filter %.cu %.cpp %.o, $^)

clean:
	$(RM) $(OBJ) $(BIN) $(CUBIN) $(CUOBJ) *~
