Compare stl::find_if vs simple search

STLvsLoop.cpp

#include <cstdint>
#include <tuple>
#include <vector>
#include <algorithm>
#include <string>
#include <random>
#include <chrono>
#include <iostream>

typedef uint32_t Hash;
std::vector<std::tuple<Hash, int, int, int>> addonImages;

bool isPresentinAddonImages_STL(Hash hash) {
	return addonImages.end() != std::find_if(addonImages.begin(), addonImages.end(), [&hash](const std::tuple<Hash, int, int, int>& element) {
		return std::get<0>(element) == hash;
	});
}

bool isPresentinAddonImages_loop(Hash hash) {
	for (const auto& addonImage : addonImages) {
		if (std::get<0>(addonImage) == hash)
			return true;
	}
	return false;
}

Hash joaat(std::string s) {
	std::transform(s.begin(), s.end(), s.begin(), ::tolower);

	Hash hash = 0;
	for (int i = 0; i < s.size(); i++) {
		hash += s[i];
		hash += (hash << 10);
		hash ^= (hash >> 6);
	}
	hash += (hash << 3);
	hash ^= (hash >> 11);
	hash += (hash << 15);
	return hash;
}

int main() {
	Hash needle = joaat("needle");
	std::mt19937 rng;
	rng.seed(std::random_device()());
	std::uniform_int_distribution<std::mt19937::result_type> dist(1, 9999); // distribution in range [1, 6]

	for (int i = 0; i < 1000; i++) {
		Hash hash = joaat("notneedle" + std::to_string(dist(rng)));
		addonImages.push_back(std::make_tuple(hash, 1, 480, 270));
	}
	addonImages.push_back(std::make_tuple(needle, 1, 480, 270));
	for (int i = 0; i < 1000; i++) {
		Hash hash = joaat("notneedle" + std::to_string(dist(rng)));
		addonImages.push_back(std::make_tuple(hash, 1, 480, 270));
	}
	std::cout << "Elements: " << addonImages.size() << "\n";
	auto t1 = std::chrono::high_resolution_clock::now();
	bool res = isPresentinAddonImages_STL(needle);
	auto t2 = std::chrono::high_resolution_clock::now();
	std::cout << "STL:\nDid ";
	if (!res) {
		std::cout << "not ";
	}
	std::cout << "find the needle in "
		<< std::chrono::duration_cast<std::chrono::nanoseconds>(t2 - t1).count()
		<< " ns!\n";

	t1 = std::chrono::high_resolution_clock::now();
	res = isPresentinAddonImages_loop(needle);
	t2 = std::chrono::high_resolution_clock::now();
	std::cout << "loop:\nDid";
	if (!res) {
		std::cout << "not ";
	}
	std::cout << "find the needle in "
		<< std::chrono::duration_cast<std::chrono::nanoseconds>(t2 - t1).count()
		<< " ns!\n";


    return 0;
}

for (auto addonImage : addonImages)
should be:
for (const auto& addonImage : addonImages)
this makes the most difference.

@Inori

Thanks for the suggestion. Wew, this was 3 years ago, I don't even remember what I ended up doing. Running the "benchmark" now the difference is minimal anyway, with no clear winner with either a reference or by value. Plus, the collection size usually is less than 1000 anyway, or other issues might occur ;)

On my machine, under MSVC x64 Release build, passing by value is 3 times slower than passing by reference on average.
After the fix, no significant difference anymore.
I'm just curious about the performence of find_if, and searching on Google brings me here.

Ah yeah, under debug mode the difference is significant, I forgot to mention I tested with a release type build this time.