bl1_axpyv3b.c File Reference

(r)

Functions
void	bl1_saxpyv3b (int n, float *alpha1, float *alpha2, float *alpha3, float *x1, int inc_x1, float *x2, int inc_x2, float *x3, int inc_x3, float *y, int inc_y)
void	bl1_daxpyv3b (int n, double *alpha1, double *alpha2, double *alpha3, double *x1, int inc_x1, double *x2, int inc_x2, double *x3, int inc_x3, double *y, int inc_y)
	for (i=0;i< n_run;++i)
	if (n_left==1)
void	bl1_caxpyv3b (int n, scomplex *alpha1, scomplex *alpha2, scomplex *alpha3, scomplex *x1, int inc_x1, scomplex *x2, int inc_x2, scomplex *x3, int inc_x3, scomplex *y, int inc_y)
void	bl1_zaxpyv3b (int n, dcomplex *alpha1, dcomplex *alpha2, dcomplex *alpha3, dcomplex *x1, int inc_x1, dcomplex *x2, int inc_x2, dcomplex *x3, int inc_x3, dcomplex *y, int inc_y)

Variables
double *restrict	chi2
double *restrict	psi1 = y
double	alpha1_c = *alpha1
double	alpha2_c = *alpha2
double	temp1
double	temp2
int	i
int	n_run = n / 2
int	n_left = n % 2
int	twoinc_x1 = 2*inc_x1
int	twoinc_x2 = 2*inc_x2
int	twoinc_y = 2*inc_y
	chi1 = x1

Function Documentation

bl1_caxpyv3b()

void bl1_caxpyv3b	(	int	n,
		scomplex *	alpha1,
		scomplex *	alpha2,
		scomplex *	alpha3,
		scomplex *	x1,
		int	inc_x1,
		scomplex *	x2,
		int	inc_x2,
		scomplex *	x3,
		int	inc_x3,
		scomplex *	y,
		int	inc_y
	)

{
    bl1_abort();
}

References bl1_abort().

bl1_daxpyv3b()

void bl1_daxpyv3b	(	int	n,
		double *	alpha1,
		double *	alpha2,
		double *	alpha3,
		double *	x1,
		int	inc_x1,
		double *	x2,
		int	inc_x2,
		double *	x3,
		int	inc_x3,
		double *	y,
		int	inc_y
	)

{
    double*   restrict chi1;
    double*   restrict chi2;
    double*   restrict chi3;
    double*   restrict psi1;
    int       i;
 
    int       n_pre;
    int       n_run;
    int       n_left;
 
    v2df_t    a1v, a2v, a3v;
    v2df_t    x11v, x12v, x13v;
    v2df_t    x21v, x22v, x23v;
    v2df_t    y1v;
    v2df_t    y2v;
 
    if ( inc_x1 != 1 ||
         inc_x2 != 1 ||
         inc_x3 != 1 ||
         inc_y  != 1 ) bl1_abort();
 
    n_pre = 0;
    if ( ( unsigned long ) y % 16 != 0 )
    {
        if ( ( unsigned long ) x1 % 16 == 0 ||
             ( unsigned long ) x2 % 16 == 0 ||
             ( unsigned long ) x3 % 16 == 0 ) bl1_abort();
 
        n_pre = 1;
    }
 
    n_run       = ( n - n_pre ) / 4;
    n_left      = ( n - n_pre ) % 4;
 
    chi1 = x1;
    chi2 = x2;
    chi3 = x3;
    psi1 = y;
 
    if ( n_pre == 1 )
    {
        double   alpha1_c = *alpha1;
        double   alpha2_c = *alpha2;
        double   alpha3_c = *alpha3;
        double   chi11_c = *chi1;
        double   chi12_c = *chi2;
        double   chi13_c = *chi3;
 
        *psi1 += alpha1_c * chi11_c + alpha2_c * chi12_c + alpha3_c * chi13_c;
 
        chi1 += inc_x1;
        chi2 += inc_x2;
        chi3 += inc_x3;
        psi1 += inc_y;
    }
 
    a1v.v = _mm_loaddup_pd( ( double* )alpha1 );
    a2v.v = _mm_loaddup_pd( ( double* )alpha2 );
    a3v.v = _mm_loaddup_pd( ( double* )alpha3 );
 
    for ( i = 0; i < n_run; ++i )
    {
        x11v.v = _mm_load_pd( ( double* )chi1 );
        x12v.v = _mm_load_pd( ( double* )chi2 );
        x13v.v = _mm_load_pd( ( double* )chi3 );
        y1v.v  = _mm_load_pd( ( double* )psi1 );
 
        y1v.v += a1v.v * x11v.v + a2v.v * x12v.v + a3v.v * x13v.v;
 
        _mm_store_pd( ( double* )psi1, y1v.v );
 
        x21v.v = _mm_load_pd( ( double* )(chi1 + 2) );
        x22v.v = _mm_load_pd( ( double* )(chi2 + 2) );
        x23v.v = _mm_load_pd( ( double* )(chi3 + 2) );
        y2v.v  = _mm_load_pd( ( double* )(psi1 + 2) );
 
        y2v.v += a1v.v * x21v.v + a2v.v * x22v.v + a3v.v * x23v.v;
 
        _mm_store_pd( ( double* )(psi1 + 2), y2v.v );
 
        chi1 += 4;
        chi2 += 4;
        chi3 += 4;
        psi1 += 4;
    }
 
    if ( n_left > 0 )
    {
        double   alpha1_c = *alpha1;
        double   alpha2_c = *alpha2;
        double   alpha3_c = *alpha3;
 
        for ( i = 0; i < n_left; ++i )
        {
            double   chi11_c = *chi1;
            double   chi12_c = *chi2;
            double   chi13_c = *chi3;
 
            *psi1 += alpha1_c * chi11_c + alpha2_c * chi12_c + alpha3_c * chi13_c;
 
            chi1 += inc_x1;
            chi2 += inc_x2;
            chi3 += inc_x3;
            psi1 += inc_y;
        }
    }
}

References alpha1, alpha1_c, alpha2, alpha2_c, bl1_abort(), chi1, chi2, i, n_left, n_pre, n_run, psi1, v2df_t::v, and x1.

bl1_saxpyv3b()

void bl1_saxpyv3b	(	int	n,
		float *	alpha1,
		float *	alpha2,
		float *	alpha3,
		float *	x1,
		int	inc_x1,
		float *	x2,
		int	inc_x2,
		float *	x3,
		int	inc_x3,
		float *	y,
		int	inc_y
	)

{
    bl1_abort();
}

References bl1_abort().

bl1_zaxpyv3b()

void bl1_zaxpyv3b	(	int	n,
		dcomplex *	alpha1,
		dcomplex *	alpha2,
		dcomplex *	alpha3,
		dcomplex *	x1,
		int	inc_x1,
		dcomplex *	x2,
		int	inc_x2,
		dcomplex *	x3,
		int	inc_x3,
		dcomplex *	y,
		int	inc_y
	)

{
    bl1_abort();
}

References bl1_abort().

for()

for

(

)

    {
        double   chi11_c = *chi1;
        double   chi21_c = *(chi1 + inc_x1);
        double   chi12_c = *chi2;
        double   chi22_c = *(chi2 + inc_x2);
        double   psi1_c  = *psi1;
        double   psi2_c  = *(psi1 + inc_y);
 
        // psi1 = psi1 + alpha1 * chi11 + alpha2 * chi12;
        // psi2 = psi2 + alpha1 * chi21 + alpha2 * chi22;
        temp1 = alpha1_c * chi11_c + alpha2_c * chi12_c;
        temp2 = alpha1_c * chi21_c + alpha2_c * chi22_c;
 
        *psi1           = psi1_c + temp1;
        *(psi1 + inc_y) = psi2_c + temp2;
 
        chi1 += twoinc_x1;
        chi2 += twoinc_x2;
        psi1 += twoinc_y;
    }

References alpha1_c, alpha2_c, chi1, chi2, psi1, temp1, temp2, twoinc_x1, twoinc_x2, and twoinc_y.

if()

if

(

n_left

= = 1

)

    {
        double   chi11_c = *chi1;
        double   chi12_c = *chi2;
 
        // psi1 = psi1 + alpha1 * chi11 + alpha2 * chi12;
        temp1 = alpha1_c * chi11_c + alpha2_c * chi12_c;
 
        *psi1 = *psi1 + temp1;
    }

References alpha1_c, alpha2_c, chi1, chi2, psi1, and temp1.

Variable Documentation

alpha1_c

alpha1_c = *alpha1

Referenced by bl1_daxpyv3b(), for(), and if().

alpha2_c

alpha2_c = *alpha2

Referenced by bl1_daxpyv3b(), for(), and if().

chi1

chi1 = x1

Referenced by bl1_daxpyv3b(), for(), and if().

chi2

chi2

Initial value:

{
    double*   restrict chi1

Referenced by bl1_daxpyv3b(), for(), and if().

i

int i

Referenced by bl1_daxpyv3b().

n_left

int n_left = n % 2

Referenced by bl1_daxpyv3b().

n_run

int n_run = n / 2

Referenced by bl1_daxpyv3b().

psi1

psi1 = y

Referenced by bl1_daxpyv3b(), for(), and if().

temp1

double temp1

Referenced by for(), and if().

temp2

double temp2

Referenced by for().

twoinc_x1

int twoinc_x1 = 2*inc_x1

Referenced by for().

twoinc_x2

int twoinc_x2 = 2*inc_x2

Referenced by for().

twoinc_y

int twoinc_y = 2*inc_y

Referenced by for().