1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
|
# Headers for Geryon
UCL_H = $(wildcard ./geryon/ucl*.h)
NVD_H = $(wildcard ./geryon/nvd*.h) $(UCL_H) lal_preprocessor.h
ALL_H = $(NVD_H) $(wildcard ./lal_*.h)
# Source files
SRCS := $(wildcard ./lal_*.cpp)
OBJS := $(subst ./,$(OBJ_DIR)/,$(SRCS:%.cpp=%.o))
CUS := $(wildcard lal_*.cu)
CUHS := $(filter-out pppm_cubin.h, $(CUS:lal_%.cu=%_cubin.h)) pppm_f_cubin.h pppm_d_cubin.h
CUHS := $(addprefix $(OBJ_DIR)/, $(CUHS))
ifdef CUDPP_OPT
CUDPP = $(OBJ_DIR)/cudpp.o $(OBJ_DIR)/cudpp_plan.o \
$(OBJ_DIR)/cudpp_maximal_launch.o $(OBJ_DIR)/cudpp_plan_manager.o \
$(OBJ_DIR)/radixsort_app.cu_o $(OBJ_DIR)/scan_app.cu_o
endif
# targets
GPU_LIB = $(LIB_DIR)/libgpu.a
EXECS = $(BIN_DIR)/nvc_get_devices
all: $(OBJ_DIR) $(CUHS) $(GPU_LIB) $(EXECS)
$(OBJ_DIR):
mkdir -p $@
# Compilers and linkers
CUDA = $(NVCC) $(CUDA_INCLUDE) $(CUDA_OPTS) $(CUDA_ARCH) \
$(CUDA_PRECISION)
CUDR = $(CUDR_CPP) $(CUDR_OPTS) $(CUDA_PRECISION) $(CUDA_INCLUDE) \
$(CUDPP_OPT)
CUDA_LINK = $(CUDA_LIB) -lcudart
BIN2C = $(CUDA_HOME)/bin/bin2c
# device code compilation
$(OBJ_DIR)/pppm_f.cubin: lal_pppm.cu lal_precision.h lal_preprocessor.h
$(CUDA) --fatbin -DNV_KERNEL -Dgrdtyp=float -Dgrdtyp4=float4 -o $@ lal_pppm.cu
$(OBJ_DIR)/pppm_f_cubin.h: $(OBJ_DIR)/pppm_f.cubin
$(BIN2C) -c -n pppm_f $(OBJ_DIR)/pppm_f.cubin > $(OBJ_DIR)/pppm_f_cubin.h
$(OBJ_DIR)/pppm_d.cubin: lal_pppm.cu lal_precision.h lal_preprocessor.h
$(CUDA) --fatbin -DNV_KERNEL -Dgrdtyp=double -Dgrdtyp4=double4 -o $@ lal_pppm.cu
$(OBJ_DIR)/pppm_d_cubin.h: $(OBJ_DIR)/pppm_d.cubin
$(BIN2C) -c -n pppm_d $(OBJ_DIR)/pppm_d.cubin > $(OBJ_DIR)/pppm_d_cubin.h
$(OBJ_DIR)/%_cubin.h: lal_%.cu $(ALL_H)
$(CUDA) --fatbin -DNV_KERNEL -o $(OBJ_DIR)/$*.cubin $(OBJ_DIR)/lal_$*.cu
$(BIN2C) -c -n $* $(OBJ_DIR)/$*.cubin > $@
# host code compilation
$(OBJ_DIR)/lal_%.o: lal_%.cpp $(CUHS) $(ALL_H)
$(CUDR) -o $@ -c $< -I$(OBJ_DIR)
#ifdef CUDPP_OPT
$(OBJ_DIR)/cudpp.o: cudpp_mini/cudpp.cpp
$(CUDR) -o $@ -c cudpp_mini/cudpp.cpp -Icudpp_mini
$(OBJ_DIR)/cudpp_plan.o: cudpp_mini/cudpp_plan.cpp
$(CUDR) -o $@ -c cudpp_mini/cudpp_plan.cpp -Icudpp_mini
$(OBJ_DIR)/cudpp_maximal_launch.o: cudpp_mini/cudpp_maximal_launch.cpp
$(CUDR) -o $@ -c cudpp_mini/cudpp_maximal_launch.cpp -Icudpp_mini
$(OBJ_DIR)/cudpp_plan_manager.o: cudpp_mini/cudpp_plan_manager.cpp
$(CUDR) -o $@ -c cudpp_mini/cudpp_plan_manager.cpp -Icudpp_mini
$(OBJ_DIR)/radixsort_app.cu_o: cudpp_mini/radixsort_app.cu
$(CUDA) -o $@ -c cudpp_mini/radixsort_app.cu
$(OBJ_DIR)/scan_app.cu_o: cudpp_mini/scan_app.cu
$(CUDA) -o $@ -c cudpp_mini/scan_app.cu
#endif
# build libgpu.a
$(GPU_LIB): $(OBJS) $(CUDPP)
$(AR) -crusv $(GPU_LIB) $(OBJS) $(CUDPP)
@cp $(EXTRAMAKE) Makefile.lammps
# test app for querying device info
$(BIN_DIR)/nvc_get_devices: ./geryon/ucl_get_devices.cpp $(NVD_H)
$(CUDR) -o $@ ./geryon/ucl_get_devices.cpp -DUCL_CUDADR $(CUDA_LIB) -lcuda
clean:
-rm -f $(EXECS) $(GPU_LIB) $(OBJS) $(CUDPP) $(CUHS) *.linkinfo
veryclean: clean
-rm -rf *~ *.linkinfo
cleanlib:
-rm -f $(EXECS) $(GPU_LIB) $(OBJS) $(CUHS) *.linkinfo
|
