Files
hm.h/benchmark.cpp
2026-04-10 17:57:39 +02:00

244 lines
6.3 KiB
C++

// Initially generated by AI (Google Gemini)
#include <iostream>
#include <vector>
#include <string>
#include <chrono>
#include <unordered_map>
#include <iomanip>
#define HM_IMPLEMENTATION
#include "hm.h"
using namespace std::chrono;
typedef time_point<high_resolution_clock> time_p;
// Configuration
const int NUM_ELEMENTS = 1000000;
time_p now()
{
return high_resolution_clock::now();
}
double time_diff_nano(time_p end, time_p start)
{
return duration_cast<nanoseconds>(end - start).count() / 1e6;
}
void print_div()
{
std::cout << "---------------------------------------------------------" << std::endl;
}
void print_head()
{
std::cout << std::left << std::setw(10) << "Operation"
<< std::right << std::setw(22) << "std::unordered_map"
<< std::right << std::setw(13) << "hm.h"
<< std::right << std::setw(11) << "Ratio" << std::endl;
}
void print_row(const std::string& label, double hm_ms, double std_ms)
{
std::cout << std::left << std::setw(10) << label
<< std::right << std::setw(19) << std::fixed << std::setprecision(2) << std_ms << " ms"
<< std::right << std::setw(10) << std::fixed << std::setprecision(2) << hm_ms << " ms"
<< std::right << std::setw(10) << (hm_ms / std_ms) << "x" << std::endl;
}
// Test with string key type
uint64_t hash_str(const void *buffer, size_t size)
{
(void)size;
const char *str = reinterpret_cast<const char*>(buffer);
uint64_t hash = (uint64_t)str;
return hash;
}
void test_str()
{
std::vector<std::string> key_pool;
std::vector<std::string> value_pool;
// Generate random strings to ensure we aren't just hashing "1", "2", "3"
for (int i = 0; i < NUM_ELEMENTS; ++i) {
key_pool.push_back("key_prefix_" + std::to_string(i) + "_suffix_" + std::to_string(rand() % 1000));
value_pool.push_back("value_" + std::to_string(i));
}
// --- Custom implementation ---
// Using 0 for sizes tells your library to treat them as NULL-terminated strings
HashMap hm = hm_create_ex(sizeof(const char *), 0, hash_str);
// Insertion
auto start = now();
for (int i = 0; i < NUM_ELEMENTS; ++i) {
hm_put(&hm, key_pool[i].c_str(), value_pool[i].c_str());
}
auto end = now();
double custom_put = time_diff_nano(end, start);
// Lookup
start = now();
for (int i = 0; i < NUM_ELEMENTS; ++i) {
char* val = (char*)hm_get(&hm, key_pool[i].c_str());
if (!val) {
std::cerr << "Lookup error!" << std::endl;
return;
}
}
end = now();
double custom_get = time_diff_nano(end, start);
// Deletion
start = now();
for (int i = 0; i < NUM_ELEMENTS; ++i) {
hm_remove(&hm, key_pool[i].c_str());
}
end = now();
double custom_del = time_diff_nano(end, start);
hm_free(&hm);
// --- Benchmark std::unordered_map ---
std::unordered_map<const char*, const char*> std_map;
// Insertion
start = now();
for (int i = 0; i < NUM_ELEMENTS; ++i) {
std_map[key_pool[i].c_str()] = value_pool[i].c_str();
}
end = now();
double std_put = time_diff_nano(end, start);
// Lookup
start = now();
for (int i = 0; i < NUM_ELEMENTS; ++i) {
auto it = std_map.find(key_pool[i].c_str());
if (it == std_map.end()) std::cerr << "Lookup error!" << std::endl;
}
end = now();
double std_get = time_diff_nano(end, start);
// Deletion
start = now();
for (int i = 0; i < NUM_ELEMENTS; ++i) {
std_map.erase(key_pool[i].c_str());
}
end = now();
double std_del = time_diff_nano(end, start);
// Output Results
print_head();
print_div();
print_row("Insert", custom_put, std_put);
print_row("Lookup", custom_get, std_get);
print_row("Delete", custom_del, std_del);
print_div();
}
// Test with integer key type
uint64_t hash_int(const void *buffer, size_t size)
{
(void)size;
return *reinterpret_cast<const int*>(buffer);
}
void test_int()
{
// Prepare test data (random integers) to avoid string allocation overhead in the loops
std::vector<int> keys(NUM_ELEMENTS);
std::vector<int> values(NUM_ELEMENTS);
for (int i = 0; i < NUM_ELEMENTS; ++i) {
keys[i] = i;
values[i] = i * 2;
}
// --- Custom implementation ---
HashMap hm = hm_create_ex(sizeof(int), sizeof(int), hash_int);
// Insertion
auto start = now();
for (int i = 0; i < NUM_ELEMENTS; ++i) {
hm_put(&hm, &keys[i], &values[i]);
}
auto end = now();
double custom_put = time_diff_nano(end, start);
// Lookup
start = now();
for (int i = 0; i < NUM_ELEMENTS; ++i) {
int* val = (int*)hm_get(&hm, &keys[i]);
if (!val || *val != values[i]) {
std::cerr << "Error in custom lookup!" << std::endl;
return;
}
}
end = now();
double custom_get = time_diff_nano(end, start);
// Deletion
start = now();
for (int i = 0; i < NUM_ELEMENTS; ++i) {
hm_remove(&hm, &keys[i]);
}
end = now();
double custom_del = time_diff_nano(end, start);
hm_free(&hm);
// --- Benchmark std::unordered_map ---
std::unordered_map<int, int> std_map;
// Insertion
start = now();
for (int i = 0; i < NUM_ELEMENTS; ++i) {
std_map[keys[i]] = values[i];
}
end = now();
double std_put = time_diff_nano(end, start);
// Lookup
start = now();
for (int i = 0; i < NUM_ELEMENTS; ++i) {
auto it = std_map.find(keys[i]);
if (it == std_map.end() || it->second != values[i]) {
std::cerr << "Error in std lookup!" << std::endl;
}
}
end = now();
double std_get = time_diff_nano(end, start);
// Deletion
start = now();
for (int i = 0; i < NUM_ELEMENTS; ++i) {
std_map.erase(keys[i]);
}
end = now();
double std_del = time_diff_nano(end, start);
// Output Results
print_head();
print_div();
print_row("Insert", custom_put, std_put);
print_row("Lookup", custom_get, std_get);
print_row("Delete", custom_del, std_del);
print_div();
}
int main()
{
std::cout << "Benchmarking " << NUM_ELEMENTS << " (integer keys)" << std::endl << std::endl;
test_int();
std::cout << std::endl;
std::cout << "Benchmarking " << NUM_ELEMENTS << " (string keys)" << std::endl << std::endl;
test_str();
return 0;
}