TechOverflow Logo
TechOverflow

Benchmarking nanosecond timestamping: std::chrono vs clock_gettime(CLOCK_REALTIME)

In C/C++ there are two essential methods of timestamping with nanosecond resolution:

Method A: std::chrono::high_resolution_clock

uint64_t getCurrentNanoTimestampCpp() {
    return std::chrono::duration_cast<std::chrono::nanoseconds>(
        std::chrono::system_clock::now().time_since_epoch()
    ).count();
}

This method only requires C++11 and the C++ standard library. It is portable and works on all platforms that support C++11.

Method B: clock_gettime(CLOCK_REALTIME)


uint64_t getCurrentNanoTimestampC() {
    struct timespec ts;
    clock_gettime(CLOCK_REALTIME, &ts);
    return (uint64_t)ts.tv_sec * 1000000000LL + ts.tv_nsec;
}

This method is available on POSIX-compliant systems and is not portable to Windows.

Benchmarking

See below for the full code to benchmark the two methods.

Results

Results on Intel(R) Core(TM) i7-14700, Ubuntu with kernel 6.8.1-1018-realtime and g++ -fexpensive-optimizations -O3 -march=native -o benchmark_nanosecond_timestamping benchmark_nanosecond_timestamping.cpp

C clock_gettime average time per call: 13.603 ns
C++ chrono average time per call: 14.1544 ns

In other words:

Full benchmark code

#include <iostream>
#include <chrono>
#include <ctime>

uint64_t getCurrentNanoTimestampC() {
    struct timespec ts;
    clock_gettime(CLOCK_REALTIME, &ts);
    return (uint64_t)ts.tv_sec * 1000000000LL + ts.tv_nsec;
}

uint64_t getCurrentNanoTimestampCpp() {
    return std::chrono::duration_cast<std::chrono::nanoseconds>(
        std::chrono::system_clock::now().time_since_epoch()
    ).count();
}

void benchmarkFunction(uint64_t (*func)(), const std::string& name, int iterations) {
    auto start = std::chrono::high_resolution_clock::now();
    for (int i = 0; i < iterations; ++i) {
        func();
    }
    auto end = std::chrono::high_resolution_clock::now();
    
    std::chrono::duration<double, std::nano> duration = end - start;
    std::cout << name << " average time per call: " 
              << (duration.count() / iterations) << " ns" << std::endl;
}

int main() {
    constexpr int iterations = 10000000;
    
    benchmarkFunction(getCurrentNanoTimestampC, "C clock_gettime", iterations);
    benchmarkFunction(getCurrentNanoTimestampCpp, "C++ chrono", iterations);
    
    return 0;
}