#ifndef SSE_H
#define SSE_H
#include<iostream>
#include<vector>
#include<immintrin.h>
using namespace std;
namespace SSE{
template<class T>
	class SSE1{
	public:
		SSE1(){
			Array1=NULL;
			Array2=NULL;
			Array3=NULL;
			Block=0;
			Count=0;
			Oflow=0;
		}
		void Alloc(int Length){
			int Buff=Length*sizeof(T);
			Array1=(T*) _mm_malloc(Buff,32);
			Array2=(T*) _mm_malloc(Buff,32);
			Array3=(T*) _mm_malloc(Buff,32);
		}
		void Check(vector<T>&Buff1,vector<T>&Buff2,vector<T>&Buff3){
			Block=16/sizeof(T);
			Count=Buff1.size()/Block;
			Oflow=Buff1.size()-Count*Block;
			for(int i=0;i<Oflow;i++){
				Buff1.push_back((T)0.0);
				Buff2.push_back((T)0.0);
				Buff3.push_back((T)0.0);
			}
			cout<<Buff1.size()<<endl;
		}
		void Add(vector<T>&In1,vector<T>&In2,vector<T>&In3){
			Block=16/sizeof(T);
			Count=In1.size()/Block;
			for(int k=0;k<In1.size();k++){
				Array1[k]=In1[k];
				Array2[k]=In2[k];
			}
			T* pIn1=Array1;
			T* pIn2=Array2;
			T*	pIn3=Array3;
			for(int i=0;i<Count;i++){
				__m128 Xmm_A=_mm_load_ps(pIn1);
				__m128 Xmm_B=_mm_load_ps(pIn2);
				__m128 Xmm_C=_mm_add_ps(Xmm_A,Xmm_B);
				_mm_store_ps(pIn3,Xmm_C);
				pIn1+=Block;
				pIn2+=Block;
				pIn3+=Block;
			}
			for(int j=0;j<In1.size();j++){In3[j]=Array3[j];}
		}
		void Clean(vector<T>&Cl1,vector<T>&Cl2,vector<T>&Cl3){
			for(int i=0;i<Oflow;i++){
				Cl3.erase(Cl3.begin()+Cl3.size()-1);
				Cl2.erase(Cl2.begin()+Cl2.size()-1);
				Cl1.erase(Cl1.begin()+Cl2.size()-1);
			}
		}
		void Free(){
			_mm_free(Array1);
			_mm_free(Array2);
			_mm_free(Array3);
		}
	private:
		T* Array1;
		T* Array2;
		T* Array3;
		int Block;
		int Count;
		int Oflow;
	};
	namespace Math{
		template<class U>
		void Add(vector<U>&A,vector<U>&B,vector<U>&C){
			if(A.size()==B.size()&&A.size()==C.size()){
				SSE1<U>S;
				S.Check(A,B,C);
				S.Alloc(A.size());
				S.Add(A,B,C);
				S.Clean(A,B,C);
				S.Free();
			}else{
				cout<<"Enter the Vectors with same length!"<<endl;
			}
		}
	}
}
#endif