include makefile.def

NAME:=triplets
NAME2:=trackdiff
SRC_C:=src/track.cu src/data.cu src/test.cu src/global-ctx.cu src/bunch-ctx.cu \
  src/options.cu src/sector-layer.cu
SRC_C2:=src/track.cu src/trackdiff.cu
SRC_H:=src/*.h
SRC_H2:=src/data.h src/track.h src/util.h src/vec.h
SRC:=$(SRC_C) $(SRC_H)
SRC2:=$(SRC_C2) $(SRC_H2)
#OBJ:=bin/$(NAME).o
OBJS:=src/track.o src/data.o src/test.o src/global-ctx.o src/bunch-ctx.o \
  src/bunch-ctx-host.o src/options.o src/sector-layer.o
TGT2:=bin/$(NAME2)
TMP:= $(OBJS) *~ src/*~ 

NVFLAGS := -w -Xptxas -dlcm=ca -O3 -Xcompiler -fopenmp
#NVFLAGS := -O3 -Xcompiler -fopenmp
#NVFLAGS := -g -G -O3 -Xcompiler -fopenmp
NVLINK_FLAGS :=
NVCC_FLAGS :=

ifeq ($(WITH_DYN_PAR),yes)
TGT:=bin/$(NAME)
NVFLAGS += -arch=sm_35 -DWITH_DYN_PAR=1
#NVCC_FLAGS := $(NVFLAGS) -rdc=true -lcudadevrt
NVLINK_FLAGS += -lcudadevrt
else
TGT:=bin/$(NAME).nodyn
NVFLAGS += -gencode arch=compute_30,code=sm_30 \
	-gencode arch=compute_35,code=sm_35 -DWITH_DYN_PAR=0
#NVCC_FLAGS := $(NVFLAGS) -rdc=true
endif 

#NVCC_FLAGS += -lnvToolsExt -lpthread
NVLINK_FLAGS += -lnvToolsExt -lpthread
NVCC_FLAGS += $(NVFLAGS)
NVLINK_FLAGS += $(NVFLAGS)
#NVFLAGS := -arch=sm_35 -Xptxas -dlcm=ca -O3 -Xcompiler -fopenmp
#NVCC_FLAGS := $(NVFLAGS) -rdc=true -lcudadevrt -lnvToolsExt
NVCC_FLAGS2 := $(NVFLAGS) 

build: $(TGT)

#$(TGT): $(SRC) makefile makefile.def
#	nvcc $(NVCC_FLAGS) $(SRC_C) -o $(TGT)

$(TGT): $(OBJS) makefile makefile.def
	nvcc $(NVLINK_FLAGS) -o $(TGT) $(OBJS)

%.o: %.cu $(SRC_H) makefile makefile.def
	nvcc $(NVCC_FLAGS) -dc -o $@ $<

#$(TGT): $(OBJ) makefile
#	nvcc $(NVLINK_FLAGS) -o $(TGT) $(OBJ)

$(TGT2):	$(SRC2) makefile makefile.def
	nvcc $(NVCC_FLAGS2) -o $(TGT2) $(SRC_C2)

all:	$(TGT) $(TGT2)

run: $(TGT)
	./$(TGT)

run-dyn: $(TGT)
	./$(TGT) -B dyn

run-tblock: $(TGT)
	./$(TGT) -B tblock

run-hstream: $(TGT)
	CUDA_DEVICE_MAX_CONNECTIONS=32 ./$(TGT) -B hstream

run-hthread: $(TGT)
	CUDA_DEVICE_MAX_CONNECTIONS=32 ./$(TGT) -B hthread

clean:
	rm -f $(TMP) $(TGT) $(TGT2)