// newmond.cpp : Defines the entry point for the console application.
//

#include "stdafx.h"

#include <xmmintrin.h>
#include <emmintrin.h>
#include <ctime>
#include <cstdlib>
#include <iostream>
using namespace std;

void IterateMandelbrot( __m128 a, __m128 b );
int IterateMandelbrot2( float a, float b );

int _tmain(int argc, _TCHAR* argv[])
{

	unsigned int start = clock();

	const float xres= 1.0f/(float)400 ;
	const float yres= 1.0f/(float)320 ;

	const __m128 ixres = _mm_set1_ps( xres );
    const __m128 iyres = _mm_set1_ps( yres );

    for( int j=0; j < 320; j++ )
    for( int i=0; i < 400; i+=4 )
    {
        __m128  a, b;
		
        a = _mm_set_ps( i+3, i+2, i+1, i+0 );
        a = _mm_mul_ps( a, ixres );
        a = _mm_mul_ps( a, _mm_set1_ps( 3.00f) );
        a = _mm_add_ps( a, _mm_set1_ps(-2.25f) );

        b = _mm_set1_ps( (float)j );
        b = _mm_mul_ps( b, iyres );
        b = _mm_mul_ps( b, _mm_set1_ps(-2.24f) );
        b = _mm_add_ps( b, _mm_set1_ps( 1.12f) );

	   IterateMandelbrot( a, b );
	   
    }
	std::cout << "Time taken in millisecs: SSE " << clock()-start<<endl;
	start = clock();
	
    for( int j=0; j < 320; j++ )
    for( int i=0; i < 400; i++ )
    {
		float a2,b2;
		a2=xres*i;
		b2=yres*j;

		a2=a2*3.00f-2.25f;
		b2=b2*-2.24f+1.12f;
		IterateMandelbrot2( a2, b2 );
	}
	std::cout << "Time taken in millisecs: FPU " << clock()-start<<endl;
	system("pause");


	return 0;
}

int IterateMandelbrot2( float a, float b )
{
    float x, y, x2, y2;

    x = x2 = 0.0f;
    y = y2 = 0.0f;

    // iterate f(Z) = Z^2 + C,  Z0 = 0
    for( int i=0; i< 512; i++ )
    {
        y = 2.0f*x*y+b;
        x = x2-y2+a;

        x2 = x*x;
        y2 = y*y;

        const float m2 = x2+y2;
        if( m2>4.0f )
            break;
    }

    // create color
    return 1;
}



void IterateMandelbrot( __m128 a, __m128 b )
{
    __m128  x, y, x2, y2, m2;
    __m128  co, ite;

    unsigned int i;

    const __m128 one = _mm_set1_ps(1.0f);
    const __m128 th  = _mm_set1_ps(4.0f);

    x   = _mm_setzero_ps();
    y   = _mm_setzero_ps();
    x2  = _mm_setzero_ps();
    y2  = _mm_setzero_ps();
    co  = _mm_setzero_ps();
    ite = _mm_setzero_ps();

    // iterate f(Z) = Z^2 + C,  Z0 = 0
    for( i=0; i < 512; i++ )
        {
        y  = _mm_mul_ps( x, y );
        y  = _mm_add_ps( _mm_add_ps(y,y),   b );
        x  = _mm_add_ps( _mm_sub_ps(x2,y2), a );

        x2 = _mm_mul_ps( x, x );
        y2 = _mm_mul_ps( y, y );

        m2 = _mm_add_ps(x2,y2);
        co = _mm_or_ps( co, _mm_cmpgt_ps( m2, th ) );


        ite = _mm_add_ps( ite, _mm_andnot_ps( co, one ) );
        if( _mm_movemask_ps( co )==0x0f )
		{
           break;
		}
        

    // create color
    //__m128i bb= _mm_cvtps_epi32( ite );
    //const __m128i gg = _mm_slli_si128( bb, 1 );
    //const __m128i rr = _mm_slli_si128( bb, 2 );
    //const __m128i color = _mm_or_si128( _mm_or_si128(rr,gg),bb ); 

   // return( bb );

}
}