Benchmark Windows file system metadata query

benchmark_windows_fs_metadata_query.cpp

#define WIN32_LEAN_AND_MEAN
#define NOMINMAX
#include <Windows.h>
#include <array>
#include <cstdio>
#include <chrono>
#include <thread>
#include <vector>

constexpr int FileCount = 1024;
constexpr int IterationCount = 100;

const wchar_t* g_Root;
std::array<std::wstring, FileCount> g_Files;

struct State
{
	std::thread Thread;
	std::chrono::microseconds ElapsedMicroseconds;
};

void benchmark(State* pState)
{
	const auto start = std::chrono::high_resolution_clock::now();

	WIN32_FILE_ATTRIBUTE_DATA data{};
	WIN32_FIND_DATA findData{};
	for (int i = 0; i < IterationCount; i++)
	{
		for (auto& f : g_Files)
		{
			if (!GetFileAttributesExW(f.c_str(), GetFileExInfoStandard, &data))
			{
				std::printf("error: GetFileAttributesExW failed: %d\n", GetLastError());
				return;
			}

			//auto hFile = CreateFileW(f.c_str(), GENERIC_READ, FILE_SHARE_READ, nullptr, OPEN_EXISTING, 0, nullptr);
			//if (hFile == INVALID_HANDLE_VALUE)
			//{
			//	std::printf("error: CreateFileW failed: %d\n", GetLastError());
			//	return;
			//}
			//CloseHandle(hFile);

			//auto hFind = FindFirstFileExW(f.c_str(), FindExInfoBasic, &findData, FindExSearchNameMatch, nullptr, 0);
			//if (hFind == INVALID_HANDLE_VALUE)
			//{
			//	std::printf("error: FindFirstFileExW failed: %d\n", GetLastError());
			//	return;
			//}
			//FindClose(hFind);
		}
	}

	pState->ElapsedMicroseconds = std::chrono::duration_cast<std::chrono::microseconds>(std::chrono::high_resolution_clock::now() - start);
}

int wmain(int argc, wchar_t* argv[])
{
	if (argc < 3)
	{
		std::printf("exe <Root> <ThreadCount>\n");
		exit(1);
	}

	g_Root = argv[1];
	int threadCount = _wtoi(argv[2]);

	if (threadCount <= 0)
	{
		std::printf("Invalid thread count: %ls\n", argv[2]);
		exit(1);
	}

	if (!CreateDirectoryW(g_Root, nullptr) && GetLastError() != ERROR_ALREADY_EXISTS)
	{
		std::printf("fatal error: CreateDirectoryW failed: %ls, %d\n", g_Root, GetLastError());
		exit(1);
	}

	wchar_t buf[MAX_PATH + 10];
	for (int i = 0; i < FileCount; i++)
	{
		wsprintf(buf, L"%ls\\File%04d", g_Root, i);
		g_Files[i] = buf;

		auto hFile = CreateFileW(g_Files[i].c_str(), 0, 0, nullptr, CREATE_ALWAYS, 0, nullptr);
		if (hFile == INVALID_HANDLE_VALUE)
		{
			std::printf("fatal error: CreateFileW failed: %ls, %d\n", g_Files[i].c_str(), GetLastError());
			exit(1);
		}
		CloseHandle(hFile);
	}

	long totalInvocation = static_cast<long>(threadCount) * IterationCount * FileCount;
	std::printf("File#      : %d\n", FileCount);
	std::printf("Thread#    : %d\n", threadCount);
	std::printf("Iteration# : %d\n", IterationCount);
	std::printf("Invocation#: %ld\n", totalInvocation);

	std::vector<State> states;
	for (int i = 0; i < threadCount; i++)
	{
		states.push_back(State{});
	}

	for (auto& s : states)
	{
		s.Thread = std::thread(benchmark, &s);
	}

	std::chrono::microseconds totalMicroseconds{};
	for (auto& s : states)
	{
		s.Thread.join();
		totalMicroseconds += s.ElapsedMicroseconds;
	}

	auto totalMilliseconds = std::chrono::duration_cast<std::chrono::milliseconds>(totalMicroseconds);
	std::printf("%lld ms\n", totalMilliseconds.count());
	auto millisecondsPerInvocation = static_cast<double>(totalMicroseconds.count()) / totalInvocation;
	std::printf("%.3f us per invocation\n", millisecondsPerInvocation);
}

benchmark_windows_fs_metadata_query_single_threaded.cpp

// When a directory has a large number of files (>1000 or so), 
// iterating over files in it is sometimes up to 10 times slower.
//
// This benchmark creates 10000 files and then performs GetFileAttributesExW on each file.
//
// Actual workloads will be some massively parallel operation where:
//
// 1. A build system tests if input files are updated or not.
// 2. Clang preprocesses and tries to locate header files from include search directories.
//    (Not MSVC which is optimized for Windows file system.)

#define WIN32_LEAN_AND_MEAN
#define NOMINMAX
#include <Windows.h>
#include <array>
#include <cstdio>
#include <chrono>
#include <thread>
#include <vector>

/*
	F:\TestDir>F:\source\ConsoleApplication1\x64\Release\ConsoleApplication1.exe F:\TestDir\ 1
	File#      : 10000
	Iteration# : 10
	Invocation#: 100000
	1516 ms
	15.168 us per invocation

	F:\TestDir>F:\source\ConsoleApplication1\x64\Release\ConsoleApplication1.exe F:\TestDir\ 1
	File#      : 10000
	Iteration# : 10
	Invocation#: 100000
	325 ms
	3.257 us per invocation

	F:\TestDir>F:\source\ConsoleApplication1\x64\Release\ConsoleApplication1.exe F:\TestDir\ 1
	File#      : 10000
	Iteration# : 10
	Invocation#: 100000
	1063 ms
	10.635 us per invocation

	F:\TestDir>F:\source\ConsoleApplication1\x64\Release\ConsoleApplication1.exe F:\TestDir\ 1
	File#      : 10000
	Iteration# : 10
	Invocation#: 100000
	313 ms
	3.133 us per invocation
*/

constexpr int FileCount = 10000;
constexpr int IterationCount = 10;

int wmain(int argc, wchar_t* argv[])
{
	if (argc < 2)
	{
		std::printf("exe <Root>\n");
		return 1;
	}

	const wchar_t* root = argv[1];

	if (!CreateDirectoryW(root, nullptr) && GetLastError() != ERROR_ALREADY_EXISTS)
	{
		std::printf("fatal error: CreateDirectoryW failed: %ls, %d\n", root, GetLastError());
		exit(1);
	}

	std::vector<std::wstring> files;
	for (int i = 0; i < FileCount; i++)
	{
		wchar_t buf[MAX_PATH + 10];
		wsprintf(buf, L"%ls\\File%04d", root, i);
		files.push_back(buf);

		auto hFile = CreateFileW(buf, 0, 0, nullptr, CREATE_ALWAYS, 0, nullptr);
		if (hFile == INVALID_HANDLE_VALUE)
		{
			std::printf("fatal error: CreateFileW failed: %ls, %d\n", buf, GetLastError());
			exit(1);
		}
		CloseHandle(hFile);
	}

	constexpr long totalInvocation = IterationCount * FileCount;
	std::printf("File#      : %d\n", FileCount);
	std::printf("Iteration# : %d\n", IterationCount);
	std::printf("Invocation#: %ld\n", totalInvocation);

	{
		const auto start = std::chrono::high_resolution_clock::now();

		WIN32_FILE_ATTRIBUTE_DATA data{};
		WIN32_FIND_DATA findData{};
		for (int i = 0; i < IterationCount; i++)
		{
			for (auto& f : files)
			{
				if (!GetFileAttributesExW(f.c_str(), GetFileExInfoStandard, &data))
				{
					std::printf("error: GetFileAttributesExW failed: %d\n", GetLastError());
					return 1;
				}

				//auto hFile = CreateFileW(f.c_str(), GENERIC_READ, FILE_SHARE_READ, nullptr, OPEN_EXISTING, 0, nullptr);
				//if (hFile == INVALID_HANDLE_VALUE)
				//{
				//	std::printf("error: CreateFileW failed: %d\n", GetLastError());
				//	return;
				//}
				//CloseHandle(hFile);

				//auto hFind = FindFirstFileExW(f.c_str(), FindExInfoBasic, &findData, FindExSearchNameMatch, nullptr, 0);
				//if (hFind == INVALID_HANDLE_VALUE)
				//{
				//	std::printf("error: FindFirstFileExW failed: %d\n", GetLastError());
				//	return;
				//}
				//FindClose(hFind);
			}
		}

		auto elapsedMicroseconds = std::chrono::duration_cast<std::chrono::microseconds>(std::chrono::high_resolution_clock::now() - start);
		auto elapsedMilliseconds = std::chrono::duration_cast<std::chrono::milliseconds>(elapsedMicroseconds);
		std::printf("%lld ms\n", elapsedMilliseconds.count());
		auto microsecondsPerInvocation = static_cast<double>(elapsedMicroseconds.count()) / totalInvocation;
		std::printf("%.3f us per invocation\n", microsecondsPerInvocation);
	}
}