diff --git a/.gitignore b/.gitignore index 62d5619..23c6353 100644 --- a/.gitignore +++ b/.gitignore @@ -8,4 +8,5 @@ compile_commands.json *.swp # Temporary files *~ -.cache/ \ No newline at end of file +.cache/ +*.bak \ No newline at end of file diff --git a/lab3/nbody/lab3_nbody.sh b/lab3/nbody/lab3_nbody.sh new file mode 100755 index 0000000..bb1febb --- /dev/null +++ b/lab3/nbody/lab3_nbody.sh @@ -0,0 +1,26 @@ +#!/bin/bash + +# N体问题实验脚本 + +echo "==========================================" +echo "N体问题串行模拟实验" +echo "==========================================" +echo "" + +# 默认天体数量 +N=${1:-4} + +echo "运行参数：" +echo " 天体数量: $N" +echo " 时间步长: 0.01 s" +echo " 总步数: 100" +echo "" +# 编译程序 +xmake build nbody_ser +# 运行程序 +./build/linux/x86_64/release/nbody_ser $N + +echo "" +echo "==========================================" +echo "实验完成" +echo "==========================================" diff --git a/lab3/nbody/nbody_par.cpp b/lab3/nbody/nbody_par.cpp new file mode 100644 index 0000000..c4e5780 --- /dev/null +++ b/lab3/nbody/nbody_par.cpp @@ -0,0 +1,301 @@ +#include +#include +#include +#include +#include +#include +#include +using namespace std; + +// 常量定义 +const double G = 6.67430e-11; // 引力常数 (m^3 kg^-1 s^-2) +const double DT = 0.01; // 时间步长 +const int TMAX = 100; // 总时间步数 +const double mass_scale = 1e24; // 质量缩放因子 +const double dist_scale = 1e8; // 距离缩放因子 +const double vel_scale = 1e3; // 速度缩放因子 + +// 三维向量结构体 +struct Vec3 { + double x, y, z; + + Vec3() : x(0), y(0), z(0) {} + Vec3(double x, double y, double z) : x(x), y(y), z(z) {} + + Vec3 operator+(const Vec3 &other) const { + return Vec3(x + other.x, y + other.y, z + other.z); + } + + Vec3 operator-(const Vec3 &other) const { + return Vec3(x - other.x, y - other.y, z - other.z); + } + + Vec3 operator*(double scalar) const { + return Vec3(x * scalar, y * scalar, z * scalar); + } + + Vec3 operator/(double scalar) const { + return Vec3(x / scalar, y / scalar, z / scalar); + } + + double magnitude() const { return sqrt(x * x + y * y + z * z); } +}; + +// 天体结构体 +struct Body { + double mass; // 质量 + Vec3 position; // 位置 + Vec3 velocity; // 速度 + Vec3 force; // 受力 +}; + +// 初始化天体系统 +void init_bodies(vector &bodies, int n, bool verbose=false) { + // 中心天体（类似太阳） + bodies[0].mass = 1000 * mass_scale; + bodies[0].position = Vec3(0, 0, 0); + bodies[0].velocity = Vec3(0, 0, 0); + + // 其他天体（类似行星） + for (int i = 1; i < n; i++) { + bodies[i].mass = (1.0 + i * 0.5) * mass_scale; + double angle = 2.0 * M_PI * i / n; + double radius = (1.0 + i * 0.5) * dist_scale; + + bodies[i].position = Vec3(radius * cos(angle), radius * sin(angle), 0.0); + + // 给予切向速度以形成轨道 + double orbital_speed = sqrt(G * bodies[0].mass / radius); + bodies[i].velocity = + Vec3(-orbital_speed * sin(angle), orbital_speed * cos(angle), 0.0); + } + + // 输出初始状态 + if(verbose){ + + cout << fixed << setprecision(6); + cout << "\n初始状态:" << endl; + for (int i = 0; i < n; i++) { + cout << "天体 " << i << ": 质量=" << bodies[i].mass / mass_scale + << "e24 kg, " + << "位置=(" << bodies[i].position.x / dist_scale << ", " + << bodies[i].position.y / dist_scale << ", " + << bodies[i].position.z / dist_scale << ")e8 m" << endl; + } + } +} + +// 计算local_particles中每个物体受到all_particles中所有物体的作用力 +// 并更新local_particles中物体的速度和位置 +void compute_local_forces(vector& local_particles, + const vector& all_particles, + int local_start) { + for (size_t i = 0; i < local_particles.size(); i++) { + Vec3 total_force(0, 0, 0); + int global_idx = local_start + i; + + // 计算all_particles中所有物体对local_particles[i]的作用力 + for (size_t j = 0; j < all_particles.size(); j++) { + // 跳过自己 + if (global_idx == static_cast(j)) continue; + + // 计算从物体i指向物体j的向量 + Vec3 r_vec = all_particles[j].position - local_particles[i].position; + double distance = r_vec.magnitude(); + + // 避免除以零 + if (distance < 1e-10) continue; + + // 计算引力大小 + double force_magnitude = G * local_particles[i].mass * all_particles[j].mass + / (distance * distance); + + // 计算力的方向并累加 + Vec3 force_direction = r_vec / distance; + total_force = total_force + force_direction * force_magnitude; + } + + // 更新local_particles[i]的速度和位置 + Vec3 v_new = local_particles[i].velocity + total_force * DT / local_particles[i].mass; + Vec3 x_new = local_particles[i].position + v_new * DT; + + local_particles[i].velocity = v_new; + local_particles[i].position = x_new; + } +} + +void get_rank_info(int rank_id, + int bodies_count, // total number of bodies + int world_size, // total number of processes + int& send_size, // number of bodies to be sent from `rank_id` process + int& send_offset // offset of bodies to be sent from `rank_id` process +) { + int particles_per_proc = bodies_count / world_size; + int remainder = bodies_count % world_size; + + if (rank_id < remainder) { + send_size = particles_per_proc + 1; + send_offset = rank_id * (particles_per_proc + 1); + } else { + send_size = particles_per_proc; + send_offset = rank_id * particles_per_proc + remainder; + } + // for np = 2 and bodies_count = 5 + // rank_id=0: send_size=3, send_offset=0 + // rank_id=1: send_size=2, send_offset=3 +} + +int main(int argc, char **argv) { + MPI_Init(&argc, &argv); + + // 获取进程数量和当前进程rank + int world_size, world_rank; + bool verbose=false; + MPI_Comm_size(MPI_COMM_WORLD, &world_size); + MPI_Comm_rank(MPI_COMM_WORLD, &world_rank); + + // 从命令行参数获取天体数量 + int n = 4; // 默认4个天体 + if (argc > 1) { + n = atoi(argv[1]); + } + if (argc > 2) { + verbose = (strcmp(argv[2], "--verbose") == 0 || strcmp(argv[2], "-v") == 0); + } + // 只有rank 0打印初始信息 + if (verbose && world_rank == 0) { + cout << "N体问题并行模拟" << endl; + cout << "天体数量: " << n << endl; + cout << "进程数量: " << world_size << endl; + cout << "时间步长: " << DT << " s" << endl; + cout << "总步数: " << TMAX << endl; + cout << "----------------------------------------" << endl; + } + + // 定义Body的MPI数据类型 + // Body结构包含: mass(1) + position(3) + velocity(3) + force(3) = 10个double + MPI_Datatype MPI_BODY; + MPI_Type_contiguous(10, MPI_DOUBLE, &MPI_BODY); + MPI_Type_commit(&MPI_BODY); + + // ============================================ + // 步骤1: 获取分配给本进程的物体的初始信息local_particles + // 步骤2: 获取应用程序中所有物体的信息all_particles + // ============================================ + + vector all_particles(n); + vector local_particles; + + // 计算每个进程分配到的物体数量 + int particles_per_proc = n / world_size; + int remainder = n % world_size; + + int local_start, local_count; + if (world_rank < remainder) { + local_count = particles_per_proc + 1; + local_start = world_rank * local_count; + } else { + local_count = particles_per_proc; + local_start = world_rank * particles_per_proc + remainder; + } + + // Rank 0初始化所有物体 + if (world_rank == 0) { + init_bodies(all_particles, n, verbose); + } + + // 广播所有物体的初始信息到所有进程 + MPI_Bcast(all_particles.data(), n, MPI_BODY, 0, MPI_COMM_WORLD); + + // 每个进程提取自己负责的物体 + local_particles.resize(local_count); + for (int i = 0; i < local_count; i++) { + local_particles[i] = all_particles[local_start + i]; + } + + if (world_rank == 0) { + cout << "\n开始模拟..." << endl; + } + + // 创建发送和接收缓冲区信息 + vector all_send_size(world_size); + vector all_send_offset(world_size); + + for (int r = 0; r < world_size; r++) { + get_rank_info(r, n, world_size, all_send_size[r], all_send_offset[r]); + #ifdef DEBUG + if (world_rank == 0) { // 只让rank 0打印 + cout << "Process " << r << " will send " + << all_send_size[r] << " bodies starting from offset " + << all_send_offset[r] << endl; + } + #endif + } + + double start_time = MPI_Wtime(); + vector send_buf(local_count); // 使用local_count确定大小 + + #ifdef DEBUG + if (verbose || world_rank == 0) { + cout << fixed << setprecision(6); + cout << "\n进程 " << world_rank << " 负责天体 " << local_start + << " 到 " << (local_start + local_count - 1) << endl; + } + #endif + // ============================================ + // 主循环：N体模拟 + // ============================================ + for (int t = 0; t < TMAX; t++) { + // ------------------------------------------ + // 计算所有物体对分配给本进程的物体的作用力 + // 并据此更新local_particles的本进程的物体信息 + // ------------------------------------------ + + compute_local_forces(local_particles, all_particles, local_start); + + // ------------------------------------------ + // 将本进程信息local_particles保存到发送缓冲区send_buf + // 同时更新all_particles中的部分信息 + // ------------------------------------------ + send_buf = local_particles; + + // 更新all_particles中本进程负责的部分信息 + for (int i = 0; i < local_count; i++) { + all_particles[local_start + i] = local_particles[i]; + } + + // ------------------------------------------ + // 环形通信：对每个进程进行m-1次通信 + // ------------------------------------------ + MPI_Allgatherv(send_buf.data(), local_count, + MPI_BODY, all_particles.data(), + all_send_size.data(), all_send_offset.data(), + MPI_BODY, MPI_COMM_WORLD); + + + // 每10步输出一次状态（仅rank 0） + if (verbose && (t + 1) % 10 == 0 && world_rank == 0) { + cout << "时间步 " << t + 1 << ":" << endl; + for (int i = 0; i < n; i++) { + cout << " 天体 " << i << ": " + << "位置=(" << all_particles[i].position.x / dist_scale << ", " + << all_particles[i].position.y / dist_scale << ", " + << all_particles[i].position.z / dist_scale << ")e8 m, " + << "速度=(" << all_particles[i].velocity.x / vel_scale << ", " + << all_particles[i].velocity.y / vel_scale << ", " + << all_particles[i].velocity.z / vel_scale << ")e3 m/s" << endl; + } + } + } + + if (world_rank == 0) { + cout << "" << endl; + double end_time = MPI_Wtime(); + cout << "模拟用时: " << end_time - start_time << " 秒" << endl; + cout << "\n模拟完成!" << endl; + } + + MPI_Type_free(&MPI_BODY); + MPI_Finalize(); + return 0; +} diff --git a/lab3/nbody/nbody_ser.cpp b/lab3/nbody/nbody_ser.cpp new file mode 100644 index 0000000..72d0070 --- /dev/null +++ b/lab3/nbody/nbody_ser.cpp @@ -0,0 +1,190 @@ +#include +#include +#include +#include +#include +#include + +using namespace std; + +// 常量定义 +const double G = 6.67430e-11; // 引力常数 (m^3 kg^-1 s^-2) +const double DT = 0.01; // 时间步长 +const int TMAX = 100; // 总时间步数 + +// 三维向量结构体 +struct Vec3 { + double x, y, z; + + Vec3() : x(0), y(0), z(0) {} + Vec3(double x, double y, double z) : x(x), y(y), z(z) {} + + Vec3 operator+(const Vec3& other) const { + return Vec3(x + other.x, y + other.y, z + other.z); + } + + Vec3 operator-(const Vec3& other) const { + return Vec3(x - other.x, y - other.y, z - other.z); + } + + Vec3 operator*(double scalar) const { + return Vec3(x * scalar, y * scalar, z * scalar); + } + + Vec3 operator/(double scalar) const { + return Vec3(x / scalar, y / scalar, z / scalar); + } + + double magnitude() const { + return sqrt(x*x + y*y + z*z); + } +}; + +// 天体结构体 +struct Body { + double mass; // 质量 + Vec3 position; // 位置 + Vec3 velocity; // 速度 + Vec3 force; // 受力 +}; + +// 计算第i个物体所受的引力 +Vec3 compute_force(int i, const vector& bodies) { + Vec3 total_force(0, 0, 0); + + for (size_t j = 0; j < bodies.size(); j++) { + if (i == j) continue; // 跳过自己 + + // 计算从物体i指向物体j的向量 + Vec3 r_vec = bodies[j].position - bodies[i].position; + double distance = r_vec.magnitude(); + + // 避免除以零（物体重合的情况） + if (distance < 1e-10) continue; + + // 计算引力大小: F = G * m_i * m_j / r^2 + double force_magnitude = G * bodies[i].mass * bodies[j].mass / (distance * distance); + + // 计算力的方向（单位向量） + Vec3 force_direction = r_vec / distance; + + // 累加力（考虑方向） + total_force = total_force + force_direction * force_magnitude; + } + + return total_force; +} + +int main(int argc, char** argv) { + // 可以从命令行参数获取天体数量 + int n = 4; // 默认4个天体 + bool verbose = false; + if (argc > 1) { + n = atoi(argv[1]); + } + if (argc > 2) { + verbose = (strcmp(argv[2], "--verbose") == 0 || strcmp(argv[2], "-v") == 0); + } + + cout << "N体问题串行模拟" << endl; + cout << "天体数量: " << n << endl; + cout << "时间步长: " << DT << " s" << endl; + cout << "总步数: " << TMAX << endl; + cout << "----------------------------------------" << endl; + + // 初始化天体系统 + vector bodies(n); + vector bodies_new(n); + + // 初始化天体数据（简化版：模拟太阳系内行星） + // 使用简化的单位系统以便观察效果 + double mass_scale = 1e24; // 质量缩放因子 + double dist_scale = 1e8; // 距离缩放因子 + double vel_scale = 1e3; // 速度缩放因子 + + // 中心天体（类似太阳） + bodies[0].mass = 1000 * mass_scale; + bodies[0].position = Vec3(0, 0, 0); + bodies[0].velocity = Vec3(0, 0, 0); + + // 其他天体（类似行星） + for (int i = 1; i < n; i++) { + bodies[i].mass = (1.0 + i * 0.5) * mass_scale; + double angle = 2.0 * M_PI * i / n; + double radius = (1.0 + i * 0.5) * dist_scale; + + bodies[i].position = Vec3( + radius * cos(angle), + radius * sin(angle), + 0.0 + ); + + // 给予切向速度以形成轨道 + double orbital_speed = sqrt(G * bodies[0].mass / radius); + bodies[i].velocity = Vec3( + -orbital_speed * sin(angle), + orbital_speed * cos(angle), + 0.0 + ); + } + + // 输出初始状态 + cout << fixed << setprecision(6); + if(verbose){ + cout << "\n初始状态:" << endl; + for (int i = 0; i < n; i++) { + cout << "天体 " << i << ": 质量=" << bodies[i].mass/mass_scale << "e24 kg, " + << "位置=(" << bodies[i].position.x/dist_scale << ", " + << bodies[i].position.y/dist_scale << ", " + << bodies[i].position.z/dist_scale << ")e8 m" << endl; + } + } + + // 主循环：N体模拟 + cout << "\n开始模拟..." << endl; + time_t start_time = clock(); + for (int t = 0; t < TMAX; t++) { + // 第一步：计算所有物体新的速度和位置 + for (int i = 0; i < n; i++) { + // 计算第i个物体所受的力 + Vec3 F = compute_force(i, bodies); + + // 计算新速度: v^(t+1) = v^t + F * dt / m + Vec3 v_new = bodies[i].velocity + F * DT / bodies[i].mass; + + // 计算新位置: x^(t+1) = x^t + v^(t+1) * dt + Vec3 x_new = bodies[i].position + v_new * DT; + + // 保存到临时数组 + bodies_new[i].mass = bodies[i].mass; + bodies_new[i].position = x_new; + bodies_new[i].velocity = v_new; + } + + // 第二步：更新所有物体的速度和位置 + for (int i = 0; i < n; i++) { + bodies[i].position = bodies_new[i].position; + bodies[i].velocity = bodies_new[i].velocity; + } + + // 每10步输出一次状态 + if (verbose && (t + 1) % 10 == 0) { + cout << "时间步 " << t + 1 << ":" << endl; + for (int i = 0; i < n; i++) { + cout << " 天体 " << i << ": " + << "位置=(" << bodies[i].position.x/dist_scale << ", " + << bodies[i].position.y/dist_scale << ", " + << bodies[i].position.z/dist_scale << ")e8 m, " + << "速度=(" << bodies[i].velocity.x/vel_scale << ", " + << bodies[i].velocity.y/vel_scale << ", " + << bodies[i].velocity.z/vel_scale << ")e3 m/s" << endl; + } + } + } + time_t end_time = clock(); + double elapsed_secs = double(end_time - start_time) / CLOCKS_PER_SEC; + cout << "\n模拟用时: " << elapsed_secs << " 秒" << endl; + cout << "\n模拟完成!" << endl; + + return 0; +} diff --git a/lab3/nbody/xmake.lua b/lab3/nbody/xmake.lua new file mode 100644 index 0000000..6291d52 --- /dev/null +++ b/lab3/nbody/xmake.lua @@ -0,0 +1,15 @@ +add_rules("mode.debug", "mode.release") + +-- Find MPI package +add_requires("mpi", {system = true}) +add_requires("mpi_cxx", {system = true}) + +target("nbody_ser") + set_kind("binary") + add_files("nbody_ser.cpp") + +target("nbody_par") + set_kind("binary") + add_files("nbody_par.cpp") + add_packages("mpi") + add_packages("mpi_cxx") diff --git a/lab3/prime/lab3_prime.sh b/lab3/prime/lab3_prime.sh new file mode 100755 index 0000000..277c28b --- /dev/null +++ b/lab3/prime/lab3_prime.sh @@ -0,0 +1,65 @@ +#!/bin/bash + +# Lab 3: Prime Number Calculation Performance Test +# This script tests the parallel prime calculation program with different N and process counts + +echo "==========================================" +echo "Lab 3: Prime Number Calculation Performance Test" +echo "==========================================" +echo "" + +# Array of N values +N_VALUES=(100000 200000 400000 800000) + +# Array of process counts +PROCESS_COUNTS=(1 2 4 6 8) + +# Output file for results +OUTPUT_FILE="prime_results.txt" + +# Clear previous results +> $OUTPUT_FILE + +# Print header +echo "N值 进程数 素数个数 执行时间(秒)" | tee -a $OUTPUT_FILE +echo "--------------------------------------------------------" | tee -a $OUTPUT_FILE + +# Loop through each N value +for N in "${N_VALUES[@]}"; do + echo "" + echo "Testing N = $N" + echo "------------------------" + + # Loop through each process count + for P in "${PROCESS_COUNTS[@]}"; do + echo -n "Running with $P process(es)... " + + # Run the program and capture output + OUTPUT=$(mpirun -n $P ./build/linux/x86_64/release/prime_par_naive $N 2>&1) + + # Extract prime count and time from output + PRIME_COUNT=$(echo "$OUTPUT" | grep "Between" | grep -oP '\d+(?= primes)') + TIME=$(echo "$OUTPUT" | grep "Time =" | grep -oP '[0-9.]+(?= seconds)') + + # Print result + if [ ! -z "$PRIME_COUNT" ] && [ ! -z "$TIME" ]; then + echo "$N $P $PRIME_COUNT $TIME" | tee -a $OUTPUT_FILE + echo "Done! (Primes: $PRIME_COUNT, Time: ${TIME}s)" + else + echo "Error running program!" + echo "$N $P ERROR ERROR" | tee -a $OUTPUT_FILE + fi + done +done + +echo "" +echo "==========================================" +echo "Test completed!" +echo "==========================================" +echo "" +echo "Results saved to: $OUTPUT_FILE" +echo "" +echo "Summary Table:" +echo "--------------------------------------------------------" +cat $OUTPUT_FILE +echo "--------------------------------------------------------" diff --git a/lab3/prime/src/prime_par.cpp b/lab3/prime/src/prime_par.cpp new file mode 100644 index 0000000..c05f76f --- /dev/null +++ b/lab3/prime/src/prime_par.cpp @@ -0,0 +1,181 @@ +#include +#include +#include +#include + +// Function to perform the Sieve of Eratosthenes on a local segment +void local_sieve(int low, int high, std::vector& is_prime, const std::vector& base_primes) { + // Initialize all numbers in the local segment as potentially prime + is_prime.assign(high - low + 1, true); + + // If the segment starts from 0 or 1, mark them as not prime + if (low == 0) { + is_prime[0] = false; + if (high >= 1) { + is_prime[1] = false; + } + } else if (low == 1) { + is_prime[0] = false; + } + + // Use the base primes to mark non-primes in the local segment + for (int p : base_primes) { + // Find the first multiple of p within the [low, high] range + int start_multiple = (low / p) * p; + if (start_multiple < low) { + start_multiple += p; + } + // Ensure we don't mark the prime number itself as non-prime + if (start_multiple == p) { + start_multiple += p; + } + + // Mark all multiples of p in the local segment as non-prime + for (int multiple = start_multiple; multiple <= high; multiple += p) { + is_prime[multiple - low] = false; + } + } +} + +int main(int argc, char* argv[]) { + MPI_Init(&argc, &argv); + + int rank, size; + double wtime; + MPI_Comm_rank(MPI_COMM_WORLD, &rank); + MPI_Comm_size(MPI_COMM_WORLD, &size); + + // Check for correct number of arguments + if (argc != 3) { + if (rank == 0) { + std::cerr << "Usage: " << argv[0] << " " << std::endl; + std::cerr << " N: Upper bound of the range [2, N]." << std::endl; + std::cerr << " B: Block size for distributing the range." << std::endl; + } + MPI_Finalize(); + return 1; + } + + int N = std::atoi(argv[1]); + int B = std::atoi(argv[2]); + + if (N < 2) { + if (rank == 0) { + std::cout << "The range [2, " << N << "] contains 0 prime numbers." << std::endl; + } + MPI_Finalize(); + return 0; + } + + // --- Step 1: Process 0 finds base primes up to sqrt(N) --- + std::vector base_primes; + int limit = static_cast(std::sqrt(N)); + if (rank == 0) { + wtime = MPI_Wtime ( ); + + std::vector is_prime_small(limit + 1, true); + is_prime_small[0] = is_prime_small[1] = false; + for (int p = 2; p * p <= limit; ++p) { + if (is_prime_small[p]) { + for (int i = p * p; i <= limit; i += p) { + is_prime_small[i] = false; + } + } + } + for (int i = 2; i <= limit; ++i) { + if (is_prime_small[i]) { + base_primes.push_back(i); + } + } + } + + // --- Step 2: Broadcast base primes to all processes --- + int num_base_primes = base_primes.size(); + MPI_Bcast(&num_base_primes, 1, MPI_INT, 0, MPI_COMM_WORLD); + if (rank != 0) { + base_primes.resize(num_base_primes); + } + MPI_Bcast(base_primes.data(), num_base_primes, MPI_INT, 0, MPI_COMM_WORLD); + + // --- Step 3: Distribute the range [sqrt(N)+1, N] among processes --- + int start_range = limit + 1; + if (start_range > N) { + // No range to distribute, all primes are base primes + int total_count = base_primes.size(); + if (rank == 0) { + std::cout << "Total prime count in [2, " << N << "] is " << total_count << "." << std::endl; + } + MPI_Finalize(); + return 0; + } + + int total_elements = N - start_range + 1; + int local_low, local_high; + std::vector is_prime_local; + + // Calculate local range for this process + int num_blocks = (total_elements + B - 1) / B; + for (int i = 0; i < num_blocks; ++i) { + if (i % size == rank) { + int block_start = start_range + i * B; + int block_end = std::min(block_start + B - 1, N); + + // Perform sieve on this block + std::vector is_prime_block; + local_sieve(block_start, block_end, is_prime_block, base_primes); + + // Count primes in this block + int block_count = 0; + for (bool prime : is_prime_block) { + if (prime) { + block_count++; + } + } + + // In a real implementation, you would aggregate these counts. + // For simplicity, we'll just print from rank 0 after gathering. + // This part of the logic is simplified for the example. + // A more robust solution would gather all local counts. + } + } + + // Simplified counting: each process calculates its total assigned range and counts. + // This is a more straightforward approach than iterating through blocks. + int elements_per_proc = total_elements / size; + int remainder = total_elements % size; + + if (rank < remainder) { + local_low = start_range + rank * (elements_per_proc + 1); + local_high = local_low + elements_per_proc; + } else { + local_low = start_range + rank * elements_per_proc + remainder; + local_high = local_low + elements_per_proc - 1; + } + local_high = std::min(local_high, N); + + // Perform sieve on the assigned local range + local_sieve(local_low, local_high, is_prime_local, base_primes); + + // Count primes in the local range + int local_prime_count = 0; + for (bool prime : is_prime_local) { + if (prime) { + local_prime_count++; + } + } + + // --- Step 4: Gather local prime counts --- + int global_prime_count = 0; + MPI_Reduce(&local_prime_count, &global_prime_count, 1, MPI_INT, MPI_SUM, 0, MPI_COMM_WORLD); + double end_wtime; + // --- Step 5: Process 0 prints the final result --- + if (rank == 0) { + end_wtime = MPI_Wtime ( ) - wtime; + int total_count = base_primes.size() + global_prime_count; + std::cout << "Total prime count in [2, " << N << "] is " << total_count << "." << std::endl; + std::cout << "Time = " << end_wtime << " seconds" << std::endl; + } + + MPI_Finalize(); + return 0; +} \ No newline at end of file diff --git a/lab3/prime/src/prime_par_naive.cpp b/lab3/prime/src/prime_par_naive.cpp new file mode 100644 index 0000000..b8a748d --- /dev/null +++ b/lab3/prime/src/prime_par_naive.cpp @@ -0,0 +1,99 @@ +#include +#include +#include +#include + +int main ( int argc, char *argv[] ); +int prime_part ( int id, int p, int n ); + +int main ( int argc, char *argv[] ) +{ + int id; + int n = 100000; + int p; + int total; + int total_part; + double wtime; + + // (1) 调用MPI头文件 - 已在顶部添加 #include + + // (2) 在并行处理之前调用MPI_Init(), MPI_Comm_size(), MPI_Comm_rank() + MPI_Init ( &argc, &argv ); + MPI_Comm_size ( MPI_COMM_WORLD, &p ); + MPI_Comm_rank ( MPI_COMM_WORLD, &id ); + + // (7) 利用MPI_Wtime()来统计时间 + if ( id == 0 ) + { + wtime = MPI_Wtime ( ); + } + + // (8) 将N改为用户输入参数 + // Check for correct number of arguments + if (argc == 2) { + n = std::atoi(argv[1]); + } else if (argc > 2) { + if ( id == 0 ) + { + printf("Usage: %s [n]\n", argv[0]); + } + MPI_Finalize(); + return 1; + } + + // 每个进程计算自己负责的部分 + total_part = prime_part ( id, p, n ); + + // (4) 调用MPI_Reduce()收集各进程的计算结果 + MPI_Reduce ( &total_part, &total, 1, MPI_INT, MPI_SUM, 0, MPI_COMM_WORLD ); + + // (5) 进程0打印最终的计算结果 + if ( id == 0 ) + { + wtime = MPI_Wtime ( ) - wtime; + printf ( "\n" ); + printf ( "Between 2 and %d, there are %d primes\n", n, total ); + printf ( "Time = %f seconds\n", wtime ); + } + + // (3) 在并行处理之后调用MPI_Finalize() + MPI_Finalize ( ); + + return 0; +} + +int prime_part ( int id, int p, int n ) +{ + int i; + int j; + int prime; + int total_part; + + total_part = 0; + + // 每个进程处理自己负责的子列表 + // 例如：P=4时 + // Part 0: 2, 6, 10, 14, ... + // Part 1: 3, 7, 11, 15, ... + // Part 2: 4, 8, 12, ... + // Part 3: 5, 9, 13, ... + for ( i = 2 + id; i <= n; i = i + p ) + { + prime = 1; + + for ( j = 2; j < i; j++ ) + { + if ( i % j == 0 ) + { + prime = 0; + break; + } + } + if ( prime ) + { + total_part = total_part + 1; + } + } + + return total_part; +} diff --git a/lab3/prime/src/prime_ser.cpp b/lab3/prime/src/prime_ser.cpp new file mode 100644 index 0000000..7436c8b --- /dev/null +++ b/lab3/prime/src/prime_ser.cpp @@ -0,0 +1,58 @@ +#include +#include +#include +int main ( int argc, char *argv[] ); +int prime_part ( int id, int p, int n ); + +int main ( int argc, char *argv[] ) +{ + int id; + int n = 100000; + int p; + int total; + int total_part; + p = 4; + // Check for correct number of arguments + if (argc == 2) { + n = std::atoi(argv[1]); + } else if (argc > 2) { + printf("Usage: %s [n]\n", argv[0]); + return 1; + } + + total = 0; + for ( id = 0; id < p; id++ ) + { + total_part = prime_part ( id, p, n ); + total = total + total_part; + } + printf ( "\n" ); + printf ( "Between 2 and %d, there are %d primes\n", n, total ); + return 0; +} +int prime_part ( int id, int p, int n ) +{ + int i; + int j; + int prime; + int total_part; + total_part = 0; +for ( i = 2 + id; i <= n; i = i + p ) + { + prime = 1; + + for ( j = 2; j < i; j++ ) + { + if ( i % j == 0 ) + { + prime = 0; + break; + } + } + if ( prime ) + { + total_part = total_part + 1; + } + } + return total_part; +} diff --git a/lab3/prime/xmake.lua b/lab3/prime/xmake.lua new file mode 100644 index 0000000..37e1a97 --- /dev/null +++ b/lab3/prime/xmake.lua @@ -0,0 +1,93 @@ +add_rules("mode.debug", "mode.release") + +-- Find MPI package +add_requires("mpi", {system = true}) +add_requires("mpi_cxx", {system = true}) + +target("prime_ser") + set_kind("binary") + add_files("src/prime_ser.cpp") + +target("prime_par") + set_kind("binary") + add_files("src/prime_par.cpp") + add_packages("mpi") + add_packages("mpi_cxx") + + +target("prime_par_naive") + set_kind("binary") + add_files("src/prime_par_naive.cpp") + add_packages("mpi") + add_packages("mpi_cxx") + -- Alternatively, if MPI is installed system-wide, you can use: + +-- +-- If you want to known more usage about xmake, please see https://xmake.io +-- +-- ## FAQ +-- +-- You can enter the project directory firstly before building project. +-- +-- $ cd projectdir +-- +-- 1. How to build project? +-- +-- $ xmake +-- +-- 2. How to configure project? +-- +-- $ xmake f -p [macosx|linux|iphoneos ..] -a [x86_64|i386|arm64 ..] -m [debug|release] +-- +-- 3. Where is the build output directory? +-- +-- The default output directory is `./build` and you can configure the output directory. +-- +-- $ xmake f -o outputdir +-- $ xmake +-- +-- 4. How to run and debug target after building project? +-- +-- $ xmake run [targetname] +-- $ xmake run -d [targetname] +-- +-- 5. How to install target to the system directory or other output directory? +-- +-- $ xmake install +-- $ xmake install -o installdir +-- +-- 6. Add some frequently-used compilation flags in xmake.lua +-- +-- @code +-- -- add debug and release modes +-- add_rules("mode.debug", "mode.release") +-- +-- -- add macro definition +-- add_defines("NDEBUG", "_GNU_SOURCE=1") +-- +-- -- set warning all as error +-- set_warnings("all", "error") +-- +-- -- set language: c99, c++11 +-- set_languages("c99", "c++11") +-- +-- -- set optimization: none, faster, fastest, smallest +-- set_optimize("fastest") +-- +-- -- add include search directories +-- add_includedirs("/usr/include", "/usr/local/include") +-- +-- -- add link libraries and search directories +-- add_links("tbox") +-- add_linkdirs("/usr/local/lib", "/usr/lib") +-- +-- -- add system link libraries +-- add_syslinks("z", "pthread") +-- +-- -- add compilation and link flags +-- add_cxflags("-stdnolib", "-fno-strict-aliasing") +-- add_ldflags("-L/usr/local/lib", "-lpthread", {force = true}) +-- +-- @endcode +-- +