Sparse-Matrix-Vector-Multiplication/Makefile.cuda at main · ranimeshehata/Sparse-Matrix-Vector-Multiplication · GitHub

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
102
103
104
105
106
107
108
109
# Makefile for CUDA-enabled builds
# Requires: CUDA Toolkit installed

# Compilers
CXX = g++
NVCC = nvcc

# Compiler flags
CXXFLAGS = -std=c++17 -Wall -Wextra -O3 -fopenmp -Isrc -Isrc/serial -Isrc/openmp -Isrc/cuda
NVCCFLAGS = -std=c++17 -O3 -arch=sm_60 --compiler-options -fopenmp -Isrc -Isrc/serial -Isrc/openmp -Isrc/cuda
LDFLAGS = -fopenmp -lcudart

# CUDA paths (adjust if needed)
CUDA_PATH ?= /usr/local/cuda
CUDA_INC = -I$(CUDA_PATH)/include
CUDA_LIB = -L$(CUDA_PATH)/lib64

# Source directories
SRC_DIR = src
SERIAL_DIR = $(SRC_DIR)/serial
OPENMP_DIR = $(SRC_DIR)/openmp
CUDA_DIR = $(SRC_DIR)/cuda
TESTING_DIR = $(SRC_DIR)/testing
PLOTTING_DIR = $(SRC_DIR)/plotting

# Source files
SPARSE_MATRIX_SRC = $(SRC_DIR)/SparseMatrix.cpp
SERIAL_SRC = $(SERIAL_DIR)/serial_spmv.cpp
OPENMP_SRC = $(OPENMP_DIR)/parallel_spmv.cpp
CUDA_SRC = $(CUDA_DIR)/cuda_spmv.cu

CPP_SOURCES = $(SPARSE_MATRIX_SRC) $(SERIAL_SRC) $(OPENMP_SRC)
CU_SOURCES = $(CUDA_SRC)

CPP_OBJECTS = SparseMatrix.o serial_spmv.o parallel_spmv.o
CU_OBJECTS = cuda_spmv.o

# Headers
SPARSE_MATRIX_HDR = $(SRC_DIR)/SparseMatrix.h
SERIAL_HDR = $(SERIAL_DIR)/serial_spmv.h
OPENMP_HDR = $(OPENMP_DIR)/parallel_spmv.h
CUDA_HDR = $(CUDA_DIR)/cuda_spmv.cuh
HEADERS = $(SPARSE_MATRIX_HDR) $(SERIAL_HDR) $(OPENMP_HDR) $(CUDA_HDR)

# Executables
TARGETS = validate_cuda benchmark_cuda

# Default target
all: $(TARGETS)

# Check CUDA availability
check_cuda:
	@which nvcc > /dev/null || (echo "ERROR: nvcc not found. Please install CUDA Toolkit." && exit 1)
	@echo "CUDA compiler found: $$(which nvcc)"
	@echo "CUDA version: $$(nvcc --version | grep release | awk '{print $$5}')"

# Validate program with CUDA
validate_cuda: validate_cuda.o $(CPP_OBJECTS) $(CU_OBJECTS)
	$(NVCC) $(NVCCFLAGS) -o $@ $^ $(CUDA_LIB) $(LDFLAGS)

# Benchmark program with CUDA
benchmark_cuda: benchmark_cuda.o $(CPP_OBJECTS) $(CU_OBJECTS)
	$(NVCC) $(NVCCFLAGS) -o $@ $^ $(CUDA_LIB) $(LDFLAGS)

# C++ object files
SparseMatrix.o: $(SPARSE_MATRIX_SRC) $(SPARSE_MATRIX_HDR)
	$(CXX) $(CXXFLAGS) $(CUDA_INC) -c $< -o $@

serial_spmv.o: $(SERIAL_SRC) $(SERIAL_HDR) $(SPARSE_MATRIX_HDR)
	$(CXX) $(CXXFLAGS) $(CUDA_INC) -c $< -o $@

parallel_spmv.o: $(OPENMP_SRC) $(OPENMP_HDR) $(SPARSE_MATRIX_HDR)
	$(CXX) $(CXXFLAGS) $(CUDA_INC) -c $< -o $@

validate_cuda.o: $(TESTING_DIR)/validate_cuda.cpp $(HEADERS)
	$(CXX) $(CXXFLAGS) $(CUDA_INC) -c $< -o $@

benchmark_cuda.o: $(TESTING_DIR)/benchmark_cuda.cpp $(HEADERS)
	$(CXX) $(CXXFLAGS) $(CUDA_INC) -c $< -o $@

# CUDA object files
cuda_spmv.o: $(CUDA_SRC) $(CUDA_HDR)
	$(NVCC) $(NVCCFLAGS) -c $< -o $@

# Clean build artifacts
clean:
	rm -f $(CPP_OBJECTS) $(CU_OBJECTS) validate_cuda.o benchmark_cuda.o $(TARGETS)
	rm -f benchmark_cuda_results.csv
	rm -rf results_cuda/

# Clean and rebuild
rebuild: clean all

# Run validation tests
test: validate_cuda
	./validate_cuda

# Run benchmarks
bench: benchmark_cuda
	./benchmark_cuda

# Generate plots (requires Python with matplotlib and pandas)
plot: bench
	python3 $(PLOTTING_DIR)/plot_cuda_results.py

# Full workflow
run: check_cuda all test bench plot

.PHONY: all check_cuda clean rebuild test bench plot run