mirror of
https://github.com/amd/blis.git
synced 2026-05-11 01:30:00 +00:00
Conflicts: kernels/bgq/1/bli_axpyv_opt_var1.c kernels/bgq/1/bli_dotv_opt_var1.c
This commit is contained in:
@@ -37,16 +37,11 @@
|
||||
void bli_daxpyv_opt_var1(
|
||||
conj_t conjx,
|
||||
dim_t n,
|
||||
double* restrict alpha_in,
|
||||
double* restrict x_in, inc_t incx,
|
||||
double* restrict y_in, inc_t incy
|
||||
double* restrict alpha,
|
||||
double* restrict x, inc_t incx,
|
||||
double* restrict y, inc_t incy
|
||||
)
|
||||
{
|
||||
{
|
||||
double* restrict alpha = alpha_in;
|
||||
double* restrict x = x_in;
|
||||
double* restrict y = y_in;
|
||||
|
||||
if ( bli_zero_dim1( n ) ) return;
|
||||
|
||||
// If there is anything that would interfere with our use of aligned
|
||||
@@ -65,7 +60,7 @@ void bli_daxpyv_opt_var1(
|
||||
dim_t n_left = n % 4;
|
||||
|
||||
vector4double xv, yv, zv;
|
||||
vector4double alphav = vec_lds( 0 * sizeof(double), &alpha[0] );
|
||||
vector4double alphav = vec_lds( 0 * sizeof(double), (double*)alpha );
|
||||
|
||||
#pragma omp parallel for
|
||||
for ( dim_t i = 0; i < n_run; i++ )
|
||||
|
||||
@@ -38,16 +38,13 @@ void bli_ddotv_opt_var1(
|
||||
conj_t conjx,
|
||||
conj_t conjy,
|
||||
dim_t n,
|
||||
double* restrict x_in, inc_t incx,
|
||||
double* restrict y_in, inc_t incy,
|
||||
double* restrict rho_in
|
||||
double* restrict x, inc_t incx,
|
||||
double* restrict y, inc_t incy,
|
||||
double* restrict rho
|
||||
)
|
||||
{{
|
||||
double* restrict x = x_in;
|
||||
double* restrict y = y_in;
|
||||
double* rho = rho_in;
|
||||
{
|
||||
bool_t use_ref = FALSE;
|
||||
|
||||
bool_t use_ref = FALSE;
|
||||
// If the vector lengths are zero, set rho to zero and return.
|
||||
if ( bli_zero_dim1( n ) ) {
|
||||
PASTEMAC(d,set0s)( rho );
|
||||
|
||||
Reference in New Issue
Block a user