test_inv_sqrt.cpp

test_inv_sqrt.cpp

// Compile it with -std=c++11 and -O3

#include <cmath>
#include <iostream>
#include <chrono>
#include <vector>
#include <cstdlib>
#include <cstdint>

float Q_rsqrt( float number )
{
    static_assert(sizeof(float)==sizeof(int32_t), "float isn't 32bit on your machine");
    long i;
    float x2, y;
    const float threehalfs = 1.5F;

    x2 = number * 0.5F;
    y  = number;
    i  = * ( int32_t * ) &y;                       // evil floating point bit level hacking
    i  = 0x5f3759df - ( i >> 1 );               // what the fuck? 
    y  = * ( float * ) &i;
    y  = y * ( threehalfs - ( x2 * y * y ) );   // 1st iteration
    //y  = y * ( threehalfs - ( x2 * y * y ) );   // 2nd iteration, this can be removed

    return y;
}

using namespace std;

int main()
{
    auto start_time = chrono::high_resolution_clock::now();
    vector<float> v;
    v.reserve(100000000);
    for (int i=0; i<100000000; ++i)
    {
        v.push_back(Q_rsqrt(0.1 * i));
    } //870ms
    cout << chrono::duration_cast< chrono::milliseconds >(chrono::high_resolution_clock::now() - start_time).count() << "ms"<<endl;

    v.clear();
    v.reserve(100000000);
    start_time = chrono::high_resolution_clock::now();
    for (int i=0; i<100000000; ++i)
    {
        v.push_back(1/sqrt(0.1 * i)); //235ms
    }
    cout << chrono::duration_cast< chrono::milliseconds >(chrono::high_resolution_clock::now() - start_time).count() << "ms"<<endl;
}