sse.c

#include "stdio.h"
#include "stdlib.h"
#include "./intrin/nmmintrin.h"

/*
__m128 _pix1;
__m128 _pix2;
__m128 _fade;
__m128 _pixSub;
__m128 _pixMul;
__m128 _pixRet;

__attribute__((aligned(16))) int _ret[4];
*/

int *meargeFunc(int pic1[4], int pic2[4], float fadeRate){
	
	//__attribute__((aligned(16))) float _fpix1[4];
	//__attribute__((aligned(16))) float _fpix2[4];
	__attribute__((aligned(16))) float _fadeRate[4] = {fadeRate,fadeRate,fadeRate,fadeRate};
	__attribute__((aligned(16))) float _fpix1[4] = {(float)pic1[0],(float)pic1[1],(float)pic1[2],(float)pic1[3]};
	__attribute__((aligned(16))) float _fpix2[4] = {(float)pic2[0],(float)pic2[1],(float)pic2[2],(float)pic2[3]};
	
	//_fpix1 = _mm_cvtepi16_ps(pic1)
	//_fpix2 = _mm_cvtepi16_ps(pic2)

	/*
	for(int i=0; i<4; i++){
		//_fpix1[i] = (float)pic1[i];
		//_fpix2[i] = (float)pic2[i];
		_fadeRate[i] = fadeRate;
		//printf("fpix1[%d]=%lf,fpix2[%d]=%lf\n",i,fpix1[i],i,fpix2[i]);
	}
	
	
	
	*/
	__m128 _pix1;
	__m128 _pix2;
	__m128 _fade;
	__m128 _pixSub;
	__m128 _pixMul;
	__m128 _pixRet;
	
	__attribute__((aligned(16))) int _ret[4];
	

	/*
	_pix1 = _mm_load_ps(_fpix1);
	_pix2 = _mm_load_ps(_fpix2);
	_fade = _mm_load_ps(_fadeRate);
	
	_pixSub = _mm_sub_ps(_pix1, _pix2);  //pixa - pixb
	_pixMul = _mm_mul_ps(_pixSub, _fade);//ans * fade
	_pixRet = _mm_add_ps(_pixMul, _pix2);//ans + pixb = resultPixel
	*/
	//printf("pixMearge:%f",pixMearge);

	_pix1 = _mm_loadu_ps(_fpix1);
	_pix2 = _mm_loadu_ps(_fpix2);
	_fade = _mm_load_ps(_fadeRate);
	
	_pixSub = _mm_sub_ps(_pix1, _pix2);  //pixa - pixb
	_pixMul = _mm_mul_ps(_pixSub, _fade);//ans * fade
	_pixRet = _mm_add_ps(_pixMul, _pix2);//ans + pixb = resultPixel

	__m128i _intRet;
	
	_intRet = _mm_cvttps_epi32(_pixRet);//convert float to int

	_mm_store_si128(_ret, _intRet);//store int type result

	
	int *p = _ret;
	/*	
	for(int i=0; i<4; i++){
		printf("*(p+%d)=%d\n",i, *(p+i));
	}
	
	*/
	return p;
}