amd/blis
1
0
Fork 0
mirror of https://github.com/amd/blis.git synced 2026-05-11 17:50:00 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge "Code cleanup in 6xk DGEMM pack Kernel" into amd-staging-rome-2.2

Browse Source
This commit is contained in:
Kiran Varaganti
2020-06-01 05:08:58 -04:00
committed by Gerrit Code Review
parent 5e0ad13f8e 3ebd5f8aa0
commit c8f3cec5f7
1 changed files with 17 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files

39
kernels/zen/1m/bli_packm_zen_int.c
View File

@@ -62,9 +62,8 @@ void bli_dpackm_8xk_nn_zen
cntx_t* restrict cntx
)
{
double* restrict kappa_cast = kappa;
double* restrict alpha1 = a;
double* restrict pi1 = p;
double* restrict alpha1 = a;
double* restrict pi1 = p;
dim_t n_iter = n / 2;
dim_t n_left = n % 2;
@@ -121,21 +120,22 @@ void bli_dpackm_8xk_nn_zen
double* restrict p_cast = p;
// (*kappa_cast == 1.0) for GEMM
PRAGMA_SIMD
PRAGMA_SIMD
for (dim_t j = 0; j < n; ++j)
for (dim_t i = 0; i < cdim; ++i)
p_cast[i + j*ldp] = a_cast[i + j*lda];
/* if ( cdim < mnr ) */
const dim_t i = cdim;
const dim_t m_edge = 8 - cdim;
const dim_t n_edge = n_max;
// double* restrict p_cast = p;
double* restrict p_edge = p_cast + (i) * 1;
PRAGMA_SIMD
for (dim_t j = 0; j < n_edge; ++j)
for (dim_t i = 0; i < m_edge; ++i)
*(p_edge + i * 1 + j*ldp) = 0.0;
*(p_edge + i + j*ldp) = 0.0;
}
if (n < n_max)
@@ -146,6 +146,7 @@ void bli_dpackm_8xk_nn_zen
double* restrict p_cast = p;
double* restrict p_edge = p_cast + (j)*ldp;
PRAGMA_SIMD
for (dim_t j = 0; j < n_edge; ++j)
for (dim_t i = 0; i < m_edge; ++i)
*(p_edge + i + j*ldp) = 0.0;
@@ -166,7 +167,6 @@ void bli_dpackm_6xk_nn_zen
cntx_t* restrict cntx
)
{
double* restrict kappa_cast = kappa;
double* restrict alpha1 = a;
double* restrict pi1 = p;
@@ -192,26 +192,20 @@ void bli_dpackm_6xk_nn_zen
double* restrict a_cast = a;
double* restrict p_cast = p;
if ((*kappa_cast) == 0.0)
{
for (dim_t j = 0; j < n; ++j)
for (dim_t i = 0; i < cdim; ++i)
p_cast[i + j*ldp] = 0.0;
}
else
{
// (*kappa_cast == 1.0) for GEMM
// a will be in row-major, inca != 1 and lda = 1
for (dim_t i = 0; i < cdim; ++i)
for(dim_t j = 0; j < n; ++j)
p_cast[i + j*ldp] = a_cast[i * inca + j]; // i * inca + j * lda, lda = 1
}
/* if ( cdim < mnr ) */
// (*kappa_cast == 1.0) for GEMM
// a will be in row-major, inca != 1 and lda = 1
PRAGMA_SIMD
for (dim_t i = 0; i < cdim; ++i)
for(dim_t j = 0; j < n; ++j)
p_cast[i + j*ldp] = a_cast[i * inca + j]; // i * inca + j * lda, lda = 1
const dim_t m_edge = 6 - cdim;
const dim_t n_edge = n_max;
// double* restrict p_cast = p;
double* restrict p_edge = p_cast + (cdim) * 1;
PRAGMA_SIMD
for (dim_t j = 0; j < n_edge; ++j)
for (dim_t i = 0; i < m_edge; ++i)
*(p_edge + i + j*ldp) = 0.0;
@@ -225,6 +219,7 @@ void bli_dpackm_6xk_nn_zen
double* restrict p_cast = p;
double* restrict p_edge = p_cast + (j)*ldp;
PRAGMA_SIMD
for (dim_t j = 0; j < n_edge; ++j)
for (dim_t i = 0; i < m_edge; ++i)
*(p_edge + i + j*ldp) = 0.0;