__global__ void SW_GPU(double* memory,long int const m,long int const n, double const d, double const e, long int const N,const char* s1, const char* s2,int* semaphore)
{
    //extern __shared__ int s[];
    //printf("Bez: %d, blockDim: %d, n = %ld\n",(int)n,blockDim.x,n);
    int index = threadIdx.x + blockIdx.x * blockDim.x;
    int stride = blockDim.x * gridDim.x;
    //printf("index: %d Stride: %d N: %ld\n",index,stride,N);
    //printf("%f\n",(float)(m-1)/(blockDim.x/2));
    //printf("blockIdx.x-int((double)(n-1)/(blockDim.x/2)) = %f\n",blockIdx.x-int((double)(n-1)/(blockDim.x/2)));
    //printf("semaphore[blockIdx.x-int((double)(n-1)/(blockDim.x/2))] = %f \n",semaphore[blockIdx.x-int((double)(n-1)/(blockDim.x/2))]);
    //printf("index = %d, blockidx = %d, Semaphor = %d\n",index,blockIdx.x,semaphore[blockIdx.x]);
    //printf("gridDim: %d\n",gridDim.x);
    for (int i = index; i < N; i += stride)
    {
        __syncthreads();
        while(1)
        {

            //printf("index = %d i = %d, blockidx = %d, threadID = %d, Semaphor = %d, memory = %f\n",index,i,blockIdx.x,threadIdx.x,semaphore[blockIdx.x],memory[i-1]);
            if(i==0) break;
            //if(blockIdx.x<(int)(double)(n-1)/(blockDim.x/2) && semaphore[blockIdx.x-1]==1) break;
            //if(blockIdx.x>=(int)(double)(n-1)/(blockDim.x/2) && semaphore[blockIdx.x-1]==1 && semaphore[blockIdx.x-int((double)(n-1)/(blockDim.x/2))]==1) break;
            //if(blockIdx.x%(int)(double)(n-1)/(blockDim.x/2)==0 && semaphore[blockIdx.x-int((double)(n-1)/(blockDim.x/2))]==1) break;

            if(i<(int)n && semaphore[(i-1)%gridDim.x]>0)
            {
                semaphore[(i-1)%gridDim.x]--;
                break;
            }
            if(i>=(int)n && semaphore[(i-1)%gridDim.x]>0 && semaphore[(i-int(n))%gridDim.x]>0)
            {
                semaphore[(i-int(n))%gridDim.x]--;
                semaphore[(i-1)%gridDim.x]--;
                break;
            }
            if(i%(int)n==0 && semaphore[(i-int(n))%gridDim.x]>0)
            {
                semaphore[(i-int(n))%gridDim.x]--;
                break;
            }
        }
        if(i%n!=0 && i > n)
        {
            double simil;
            if(s1[find_index_left(i,n)]==s2[find_index_upper(i,n)]) simil = 1;
            else simil = -3;
            //printf("Hello from block %d, thread %d, index %d, Memory is %f\n", blockIdx.x, threadIdx.x,index,memory[index]);
            //printf("Index: %d\n", i);
            //printf("memory[i-n-1] = %f\n", memory[i-n-1]);
            //printf("find_L:\n");
            //printf("find_ind_l = %d\n",find_index_left(i,n));
            //printf("s1[find_index_left(i,n)] = %c\n",s1[find_index_left(i,n)]);
            //printf("s1 finished\n");
            //printf("s2[find_index_upper(i,n)] = %c\n",s2[find_index_upper(i,n)]);
            //printf("s2 finished\n");
            //printf("sim: = %f\n",simil);
            //printf("sim finished\n");
            //printf("Index: %d, memory[i-n-1] = %f, sim: %d find_ind_l = %d, find_ind_u = %d, memory[i-n] = %f, memory[i-1] = %f\n",i, memory[i-n-1],simil, find_index_left(i,n), find_index_upper(i,n), memory[i-n], memory[i-1]);
            memory[i]=max((double)0,max(memory[i-n-1]+simil,max(memory[i-n] - d,memory[i-1] - d)));
        }
        semaphore[(blockIdx.x)%gridDim.x]=2;
    }


    //semaphore[blockIdx.x+1]=1;
    //semaphore[blockIdx.x+n]=1;


    return;
}