Why the execution time of same function definition within a class is slower more than 10x time?

user_365251 0
Viewed 134

Not sure which kind of optimization the compiler does, but why within a class a same function definition is slower than the same called as global method?

#include <iostream>
#include <chrono>

#define MAX_BUFFER 256
const int whileLoops = 1024 * 1024 * 10;

void TracedFunction(int blockSize) {
    std::chrono::high_resolution_clock::time_point pStart;
    std::chrono::high_resolution_clock::time_point pEnd;

    double A[MAX_BUFFER];
    double B[MAX_BUFFER];
    double C[MAX_BUFFER];

    // fill A/B
    for (int sampleIndex = 0; sampleIndex < MAX_BUFFER; sampleIndex++) {
        A[sampleIndex] = sampleIndex;
        B[sampleIndex] = sampleIndex + 1000.0;
    }

    // same traced function
    pStart = std::chrono::high_resolution_clock::now();

    int whileCounter = 0;
    while (whileCounter < whileLoops) {
        for (int sampleIndex = 0; sampleIndex < blockSize; sampleIndex++) {
            double value = A[sampleIndex] + B[sampleIndex];

            C[sampleIndex] = value;
        }

        whileCounter++;
    }

    pEnd = std::chrono::high_resolution_clock::now();
    std::cout << "execution time: " << std::chrono::duration_cast<std::chrono::milliseconds>(pEnd - pStart).count() << " ms" << " | fake result: " << A[19] << " " << B[90] << " " << C[129] << std::endl;
}

class OptimizeProcess
{
public:
    std::chrono::high_resolution_clock::time_point pStart;
    std::chrono::high_resolution_clock::time_point pEnd;

    double A[MAX_BUFFER];
    double B[MAX_BUFFER];
    double C[MAX_BUFFER];

    OptimizeProcess() {
        // fill A/B
        for (int sampleIndex = 0; sampleIndex < MAX_BUFFER; sampleIndex++) {
            A[sampleIndex] = sampleIndex;
            B[sampleIndex] = sampleIndex + 1000.0;
        }
    }

    void TracedFunction(int blockSize) {
        // same traced function
        pStart = std::chrono::high_resolution_clock::now();

        int whileCounter = 0;
        while (whileCounter < whileLoops) {
            for (int sampleIndex = 0; sampleIndex < blockSize; sampleIndex++) {
                double value = A[sampleIndex] + B[sampleIndex];

                C[sampleIndex] = value;
            }

            whileCounter++;
        }

        pEnd = std::chrono::high_resolution_clock::now();
        std::cout << "execution time: " << std::chrono::duration_cast<std::chrono::milliseconds>(pEnd - pStart).count() << " ms" << " | fake result: " << A[19] << " " << B[90] << " " << C[129] << std::endl;
    }
};

int main() {
    int blockSize = MAX_BUFFER;

    // outside class
    TracedFunction(blockSize);

    // within class
    OptimizeProcess p1;
    p1.TracedFunction(blockSize);

    std::cout << std::endl;
    system("pause");

    return 0;
}

Tried with MSVC, /Oi /Ot.

~80ms vs 1200ms. Is there loop unrolling using blockSize as constant at compile-time?

Not sure, since I've tried to set blockSize random with:

std::mt19937_64 gen{ std::random_device()() };
std::uniform_real_distribution<double> dis{ 0.0, 1.0 };

int blockSize = dis(gen) * 255 + 1;

Same results...

compiler-optimization optimization function c++ performance
1 Answers

If you compile with the maximum optimization flag of GCC, i.e. O3, then you will get similar execution times.

There is no difference in the aspect of executing a function within or not a class, w.r.t. execution time.

The only difference that I see, is when and how you create your arrays. In the first function, the arrays are automatic variables of the function. In the within function, the arrays are data members of the class.

That can play a role in certain cases. Make the arrays global (create them only once), and you will see no difference in your execution times (regardless of using O1, O2 or O3).

Note: Compile with O2, and you will get a faster execution time for the within function (that's the other way around of what you mention). To be precise a x1.35 speedup, as you can see in the Live Demo.

Nevertheless, remember than when optimization is done right, with O3 in this case, you shouldn't see any significant differences whatsoever!

user_24113200
Related