node_pool_sample.cpp

Demonstrates basic allocation and deallocation, concurrent multi-threaded usage, performance comparison against standard new/delete, and memory efficiency including fragmentation behaviour.

See also

node_pool

// BSD 3-Clause License
// Copyright (c) 2024, 🍀☀🌕🌥 🌊
// See the LICENSE file in the project root for full license information.
 
#include "thread_base/lockfree/memory/node_pool.h"
#include "logger/core/logger.h"
#include <thread>
#include <vector>
#include <chrono>
#include <random>
#include <atomic>
#include <memory>
#include <iomanip>
 
using namespace kcenon::thread;
 
struct TestData {
    int value{0};
    double data{0.0};
    char padding[48] = {0}; // Make it a decent size for testing
    
    TestData() = default;
    TestData(int v, double d) : value(v), data(d) {}
};
 
void demonstrate_basic_usage() {
    log_module::write_information("\n=== Basic Node Pool Usage Demo ===");
    
    // Create a node pool with 2 initial chunks, 512 nodes per chunk
    node_pool<TestData> pool(2, 512);
    
    // Show initial statistics
    auto stats = pool.get_statistics();
    log_module::write_information("Initial pool statistics:");
    log_module::write_information("  Total chunks: {}", stats.total_chunks);
    log_module::write_information("  Total nodes: {}", stats.total_nodes);
    log_module::write_information("  Allocated nodes: {}", stats.allocated_nodes);
    log_module::write_information("  Free list size: {}", stats.free_list_size);
    
    // Allocate some nodes
    std::vector<TestData*> allocated_nodes;
    const int NUM_ALLOCATIONS = 100;
    
    log_module::write_information("\nAllocating {} nodes...", NUM_ALLOCATIONS);
    for (int i = 0; i < NUM_ALLOCATIONS; ++i) {
        auto* node = pool.allocate();
        node->value = i;
        node->data = i * 3.14;
        allocated_nodes.push_back(node);
    }
    
    // Show statistics after allocation
    stats = pool.get_statistics();
    log_module::write_information("After allocation:");
    log_module::write_information("  Total chunks: {}", stats.total_chunks);
    log_module::write_information("  Total nodes: {}", stats.total_nodes);
    log_module::write_information("  Allocated nodes: {}", stats.allocated_nodes);
    log_module::write_information("  Free list size: {}", stats.free_list_size);
    
    // Verify data integrity
    log_module::write_information("\nVerifying data integrity...");
    bool integrity_ok = true;
    for (int i = 0; i < NUM_ALLOCATIONS; ++i) {
        if (allocated_nodes[i]->value != i || 
            std::abs(allocated_nodes[i]->data - i * 3.14) > 0.001) {
            integrity_ok = false;
            break;
        }
    }
    log_module::write_information("Data integrity: {}", integrity_ok ? "OK" : "FAILED");
    
    // Deallocate half the nodes
    log_module::write_information("\nDeallocating half the nodes...");
    for (int i = 0; i < NUM_ALLOCATIONS / 2; ++i) {
        pool.deallocate(allocated_nodes[i]);
        allocated_nodes[i] = nullptr;
    }
    
    // Show statistics after partial deallocation
    stats = pool.get_statistics();
    log_module::write_information("After partial deallocation:");
    log_module::write_information("  Total chunks: {}", stats.total_chunks);
    log_module::write_information("  Total nodes: {}", stats.total_nodes);
    log_module::write_information("  Allocated nodes: {}", stats.allocated_nodes);
    log_module::write_information("  Free list size: {}", stats.free_list_size);
    
    // Deallocate remaining nodes
    for (int i = NUM_ALLOCATIONS / 2; i < NUM_ALLOCATIONS; ++i) {
        pool.deallocate(allocated_nodes[i]);
    }
    
    // Final statistics
    stats = pool.get_statistics();
    log_module::write_information("After full deallocation:");
    log_module::write_information("  Total chunks: {}", stats.total_chunks);
    log_module::write_information("  Total nodes: {}", stats.total_nodes);
    log_module::write_information("  Allocated nodes: {}", stats.allocated_nodes);
    log_module::write_information("  Free list size: {}", stats.free_list_size);
}
 
void demonstrate_concurrent_usage() {
    log_module::write_information("\n=== Concurrent Usage Demo ===");
    
    constexpr int NUM_THREADS = 4;
    constexpr int OPERATIONS_PER_THREAD = 1000;
    constexpr int INITIAL_CHUNKS = 2;
    constexpr int CHUNK_SIZE = 256;
    
    node_pool<TestData> pool(INITIAL_CHUNKS, CHUNK_SIZE);
    
    std::atomic<int> total_allocations{0};
    std::atomic<int> total_deallocations{0};
    std::atomic<int> allocation_failures{0};
    
    auto start_time = std::chrono::high_resolution_clock::now();
    
    std::vector<std::thread> threads;
    threads.reserve(NUM_THREADS);
    
    // Create worker threads
    for (int thread_id = 0; thread_id < NUM_THREADS; ++thread_id) {
        threads.emplace_back([&, thread_id]() {
            std::random_device rd;
            std::mt19937 gen(rd());
            std::uniform_int_distribution<> dis(0, 100);
            
            std::vector<TestData*> local_nodes;
            local_nodes.reserve(OPERATIONS_PER_THREAD / 2);
            
            for (int op = 0; op < OPERATIONS_PER_THREAD; ++op) {
                if (dis(gen) < 70 || local_nodes.empty()) { // 70% chance to allocate
                    try {
                        auto* node = pool.allocate();
                        node->value = thread_id * 10000 + op;
                        node->data = thread_id + op * 0.001;
                        local_nodes.push_back(node);
                        total_allocations.fetch_add(1, std::memory_order_relaxed);
                    } catch (const std::bad_alloc&) {
                        allocation_failures.fetch_add(1, std::memory_order_relaxed);
                    }
                } else { // Deallocate
                    if (!local_nodes.empty()) {
                        auto idx = std::uniform_int_distribution<size_t>(0, local_nodes.size() - 1)(gen);
                        pool.deallocate(local_nodes[idx]);
                        local_nodes.erase(local_nodes.begin() + idx);
                        total_deallocations.fetch_add(1, std::memory_order_relaxed);
                    }
                }
            }
            
            // Clean up remaining nodes
            for (auto* node : local_nodes) {
                if (node) {
                    pool.deallocate(node);
                    total_deallocations.fetch_add(1, std::memory_order_relaxed);
                }
            }
        });
    }
    
    // Wait for all threads to complete
    for (auto& thread : threads) {
        thread.join();
    }
    
    auto end_time = std::chrono::high_resolution_clock::now();
    auto duration = std::chrono::duration_cast<std::chrono::milliseconds>(end_time - start_time);
    
    log_module::write_information("Concurrent operations completed in {} ms", duration.count());
    log_module::write_information("Total allocations: {}", total_allocations.load());
    log_module::write_information("Total deallocations: {}", total_deallocations.load());
    log_module::write_information("Allocation failures: {}", allocation_failures.load());
    
    // Final pool statistics
    auto stats = pool.get_statistics();
    log_module::write_information("Final pool statistics:");
    log_module::write_information("  Total chunks: {}", stats.total_chunks);
    log_module::write_information("  Total nodes: {}", stats.total_nodes);
    log_module::write_information("  Allocated nodes: {}", stats.allocated_nodes);
    log_module::write_information("  Free list size: {}", stats.free_list_size);
    
    // Calculate performance
    int total_ops = total_allocations.load() + total_deallocations.load();
    double ops_per_second = (double)total_ops / (duration.count() / 1000.0);
    log_module::write_information("Performance: {} ops/second", static_cast<int>(ops_per_second));
}
 
void demonstrate_performance_comparison() {
    log_module::write_information("\n=== Performance Comparison Demo ===");
    
    constexpr int NUM_OPERATIONS = 100000;
    constexpr int WARMUP_OPERATIONS = 10000;
    
    // Test with node pool
    log_module::write_information("Testing node pool performance...");
    node_pool<TestData> pool(4, 1024);
    
    // Warmup
    std::vector<TestData*> warmup_nodes;
    for (int i = 0; i < WARMUP_OPERATIONS; ++i) {
        warmup_nodes.push_back(pool.allocate());
    }
    for (auto* node : warmup_nodes) {
        pool.deallocate(node);
    }
    
    auto start_time = std::chrono::high_resolution_clock::now();
    
    std::vector<TestData*> pool_nodes;
    pool_nodes.reserve(NUM_OPERATIONS);
    
    // Allocate
    for (int i = 0; i < NUM_OPERATIONS; ++i) {
        pool_nodes.push_back(pool.allocate());
    }
    
    // Deallocate
    for (auto* node : pool_nodes) {
        pool.deallocate(node);
    }
    
    auto end_time = std::chrono::high_resolution_clock::now();
    auto pool_duration = std::chrono::duration_cast<std::chrono::microseconds>(end_time - start_time);
    
    // Test with standard allocation
    log_module::write_information("Testing standard allocation performance...");
    
    start_time = std::chrono::high_resolution_clock::now();
    
    std::vector<std::unique_ptr<TestData>> std_nodes;
    std_nodes.reserve(NUM_OPERATIONS);
    
    // Allocate
    for (int i = 0; i < NUM_OPERATIONS; ++i) {
        std_nodes.push_back(std::make_unique<TestData>());
    }
    
    // Deallocate (automatic with unique_ptr)
    std_nodes.clear();
    
    end_time = std::chrono::high_resolution_clock::now();
    auto std_duration = std::chrono::duration_cast<std::chrono::microseconds>(end_time - start_time);
    
    log_module::write_information("Results:");
    log_module::write_information("  Node pool: {} μs", pool_duration.count());
    log_module::write_information("  Standard allocation: {} μs", std_duration.count());
    
    if (std_duration.count() > 0) {
        double speedup = (double)std_duration.count() / pool_duration.count();
        log_module::write_information("  Speedup: {:.2f}x", speedup);
    }
    
    // Calculate operations per second
    double pool_ops_per_sec = (2.0 * NUM_OPERATIONS) / (pool_duration.count() / 1000000.0);
    double std_ops_per_sec = (2.0 * NUM_OPERATIONS) / (std_duration.count() / 1000000.0);
    
    log_module::write_information("  Node pool ops/sec: {}", static_cast<int>(pool_ops_per_sec));
    log_module::write_information("  Standard ops/sec: {}", static_cast<int>(std_ops_per_sec));
}
 
void demonstrate_memory_efficiency() {
    log_module::write_information("\n=== Memory Efficiency Demo ===");
    
    // Small pool for testing
    node_pool<TestData> small_pool(1, 256);
    node_pool<TestData> medium_pool(2, 512);
    node_pool<TestData> large_pool(4, 1024);
    
    auto show_pool_info = [](const auto& pool, const char* name) {
        auto stats = pool.get_statistics();
        size_t memory_usage = stats.total_nodes * sizeof(TestData);
        log_module::write_information("{}:", name);
        log_module::write_information("  Total chunks: {}", stats.total_chunks);
        log_module::write_information("  Total nodes: {}", stats.total_nodes);
        log_module::write_information("  Memory usage: {} bytes ({:.1f} KB)", memory_usage, memory_usage / 1024.0);
        log_module::write_information("  Node size: {} bytes\n", sizeof(TestData));
    };
    
    show_pool_info(small_pool, "Small pool (1x256)");
    show_pool_info(medium_pool, "Medium pool (2x512)");
    show_pool_info(large_pool, "Large pool (4x1024)");
    
    // Test fragmentation
    log_module::write_information("Testing fragmentation scenario...");
    std::vector<TestData*> nodes;
    
    // Allocate many nodes
    for (int i = 0; i < 100; ++i) {
        nodes.push_back(medium_pool.allocate());
    }
    
    // Deallocate every other node (create fragmentation)
    for (size_t i = 0; i < nodes.size(); i += 2) {
        medium_pool.deallocate(nodes[i]);
        nodes[i] = nullptr;
    }
    
    auto stats = medium_pool.get_statistics();
    log_module::write_information("After fragmentation:");
    log_module::write_information("  Allocated nodes: {}", stats.allocated_nodes);
    log_module::write_information("  Free list size: {}", stats.free_list_size);
    
    // Allocate new nodes (should reuse freed nodes)
    int reused_count = 0;
    for (size_t i = 0; i < nodes.size() && reused_count < 25; i += 2) {
        if (!nodes[i]) {
            nodes[i] = medium_pool.allocate();
            reused_count++;
        }
    }
    
    stats = medium_pool.get_statistics();
    log_module::write_information("After reuse ({} nodes):", reused_count);
    log_module::write_information("  Allocated nodes: {}", stats.allocated_nodes);
    log_module::write_information("  Free list size: {}", stats.free_list_size);
    
    // Clean up
    for (auto* node : nodes) {
        if (node) {
            medium_pool.deallocate(node);
        }
    }
}
 
int main() {
    // Initialize logger
    log_module::start();
    log_module::console_target(log_module::log_types::Information);
    
    log_module::write_information("Node Pool Sample");
    log_module::write_information("================");
    
    try {
        demonstrate_basic_usage();
        demonstrate_concurrent_usage();
        demonstrate_performance_comparison();
        demonstrate_memory_efficiency();
        
        log_module::write_information("\n=== All demos completed successfully! ===");
        
    } catch (const std::exception& e) {
        log_module::write_error("Error: {}", e.what());
        log_module::stop();
        return 1;
    }
    
    // Cleanup logger
    log_module::stop();
    return 0;
}