Optimizing a branch for a known more-common path

Question

Please consider the following piece of code:

void error_handling();
bool method_impl();

bool method()
{
    const bool res = method_impl();
    if (res == f         


        

Answer 1

The underlying hardware already performs this optimizations. It will "fail" to predict it the first times, but after it will hit the correct option en.wikipedia.org/wiki/Branch_predictor.

You can try applying the GCC extension and check if it is faster with it or not, but I think you will barely see any difference with it and without it. The branch prediction is applied always, it is not something you enable

Answer 2

Without knowing the implementation of std::time() I wouldn't conclude much from this test. From your own results, it seems like it dominates the time in the loop.

FWIW, I use likely()/unlikely() liberally myself when tuning code. I do not want to change the structure of the code, but when reading the assembly, I'd like to see the common path be a straight line of untaken branches. The key here (for me) is the readability of the assembly. The fact that this is also what will be fastest is secondary (the fastest possible branch is a correctly predicted untaken branch).

Answer 3

You could suggest the compiler that the method_impl() will return true:

void error_handling();
bool method_impl();

bool method()
{
    const bool res = method_impl();
    if (__builtin_expect (res, 0) == false) {
        error_handling();
        return false;
    }
    return true;
}

This will work in GCC.

Answer 4

Following other answers' suggestions, I benchmarked the solutions. If you consider upvoting this answer, please upvote the others too.

Benchmark code

#include <iostream>
#include <iomanip>
#include <string>

// solutions
#include <ctime>

// benchmak
#include <limits>
#include <random>
#include <chrono>
#include <algorithm>
#include <functional>

//
// Solutions
//
namespace
{
    volatile std::time_t near_futur = -1;
    void error_handling() { std::cerr << "error\n"; }
    bool method_impl() { return std::time(NULL) != near_futur; }

    bool method_no_builtin()
    {
        const bool res = method_impl();
        if (res == false) {
            error_handling();
            return false;
        }
        return true;
    }

    bool method_builtin()
    {
        const bool res = method_impl();
        if (__builtin_expect(res, 1) == false) {
            error_handling();
            return false;
        }
        return true;
    }

    bool method_builtin_incorrect()
    {
        const bool res = method_impl();
        if (__builtin_expect(res, 0) == false) {
            error_handling();
            return false;
        }
        return true;
    }

    bool method_rewritten()
    {
        const bool res = method_impl();
        if (res == true) {
            return true;
        } else {
            error_handling();
            return false;
        }
    }
}

//
// benchmark
//
constexpr std::size_t BENCHSIZE = 10'000'000;
class Clock
{
    std::chrono::time_point<std::chrono::steady_clock> _start;

public:
    static inline std::chrono::time_point<std::chrono::steady_clock> now() { return std::chrono::steady_clock::now(); }

    Clock() : _start(now())
    {
    }

    template<class DurationUnit>
    std::size_t end()
    {
        return std::chrono::duration_cast<DurationUnit>(now() - _start).count();
    }
};

//
// Entry point
//
int main()
{
    {
        Clock clock;
        bool result = true;
        for (std::size_t i = 0 ; i < BENCHSIZE ; ++i)
        {
            result &= method_no_builtin();
            result &= method_no_builtin();
            result &= method_no_builtin();
            result &= method_no_builtin();
            result &= method_no_builtin();
            result &= method_no_builtin();
            result &= method_no_builtin();
            result &= method_no_builtin();
            result &= method_no_builtin();
            result &= method_no_builtin();
        }
        const double unit_time = clock.end<std::chrono::nanoseconds>() / static_cast<double>(BENCHSIZE);
        std::cout << std::setw(40) << "method_no_builtin(): " << std::setprecision(3) << unit_time << " ns\n";
    }
    {
        Clock clock;
        bool result = true;
        for (std::size_t i = 0 ; i < BENCHSIZE ; ++i)
        {
            result &= method_builtin();
            result &= method_builtin();
            result &= method_builtin();
            result &= method_builtin();
            result &= method_builtin();
            result &= method_builtin();
            result &= method_builtin();
            result &= method_builtin();
            result &= method_builtin();
            result &= method_builtin();
        }
        const double unit_time = clock.end<std::chrono::nanoseconds>() / static_cast<double>(BENCHSIZE);
        std::cout << std::setw(40) << "method_builtin(): " << std::setprecision(3) << unit_time << " ns\n";
    }
    {
        Clock clock;
        bool result = true;
        for (std::size_t i = 0 ; i < BENCHSIZE ; ++i)
        {
            result &= method_builtin_incorrect();
            result &= method_builtin_incorrect();
            result &= method_builtin_incorrect();
            result &= method_builtin_incorrect();
            result &= method_builtin_incorrect();
            result &= method_builtin_incorrect();
            result &= method_builtin_incorrect();
            result &= method_builtin_incorrect();
            result &= method_builtin_incorrect();
            result &= method_builtin_incorrect();
        }
        const double unit_time = clock.end<std::chrono::nanoseconds>() / static_cast<double>(BENCHSIZE);
        std::cout << std::setw(40) << "method_builtin_incorrect(): " << std::setprecision(3) << unit_time << " ns\n";
    }
    {
        Clock clock;
        bool result = true;
        for (std::size_t i = 0 ; i < BENCHSIZE ; ++i)
        {
            result &= method_rewritten();
            result &= method_rewritten();
            result &= method_rewritten();
            result &= method_rewritten();
            result &= method_rewritten();
            result &= method_rewritten();
            result &= method_rewritten();
            result &= method_rewritten();
            result &= method_rewritten();
            result &= method_rewritten();
        }
        const double unit_time = clock.end<std::chrono::nanoseconds>() / static_cast<double>(BENCHSIZE);
        std::cout << std::setw(40) << "method_rewritten(): " << std::setprecision(3) << unit_time << " ns\n";
    }
}

Benchmark results

g++ -std=c++14 -O2 -Wall -Wextra -Werror main.cpp

               method_no_builtin(): 42.8 ns
                  method_builtin(): 44.4 ns
        method_builtin_incorrect(): 51.4 ns
                method_rewritten(): 39.3 ns

Demo

g++ -std=c++14 -O3 -Wall -Wextra -Werror main.cpp

               method_no_builtin(): 32.3 ns
                  method_builtin(): 31.1 ns
        method_builtin_incorrect(): 35.6 ns
                method_rewritten(): 30.5 ns

Demo

Conclusion

The difference between those optimizations are too small to come to any conclusion other than: if there is a performance gain to find in optimizing a branch for a known more common path, this gain is too small to be worth the trouble and the loss in readability.