Skip to content

skylakex sgemv_t kernel fix #3348

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
martin-frbg merged 2 commits into from
Aug 25, 2021
Merged

skylakex sgemv_t kernel fix #3348

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
e9acb46
sgemv: skylakex: bug fix for sgemv_t kernel in corner case
guowangy Aug 25, 2021
dbbb391
sgemv: skylakex: fix build warning
guowangy Aug 25, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 0 additions & 3 deletions kernel/x86_64/sgemv_n_4.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -302,9 +302,6 @@ int CNAME(BLASLONG m, BLASLONG n, BLASLONG dummy1, FLOAT alpha, FLOAT *a, BLASLO
FLOAT * xbuffer_align = x;
FLOAT * ybuffer_align = y;

FLOAT * xbuffer = NULL;
FLOAT * ybuffer = NULL;

if (inc_x != 1) {
xbuffer_align = buffer;
for(BLASLONG i=0; i<n; i++) {
Expand Down
2 changes: 1 addition & 1 deletion kernel/x86_64/sgemv_t_4.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,7 +38,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#include "sgemv_t_microk_haswell-4.c"
#elif defined (SKYLAKEX) || defined (COOPERLAKE)
#include "sgemv_t_microk_haswell-4.c"
/*#include "sgemv_t_microk_skylakex.c"*/
#include "sgemv_t_microk_skylakex.c"
#endif

#if defined(STEAMROLLER) || defined(EXCAVATOR)
Expand Down
27 changes: 14 additions & 13 deletions kernel/x86_64/sgemv_t_microk_skylakex_template.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -93,7 +93,7 @@ static int sgemv_kernel_t_1(BLASLONG m, float alpha, float *a, float *x, float *
}

if (tag_m_32x != m) {
for (BLASLONG idx_m = tag_m_64x; idx_m < tag_m_16x; idx_m+=32) {
for (BLASLONG idx_m = tag_m_32x; idx_m < tag_m_16x; idx_m+=16) {
matrixArray_0 = _mm512_loadu_ps(&a[idx_m + 0]);

_mm512_storeu_ps(&y[idx_m + 0], _mm512_fmadd_ps(matrixArray_0, ALPHAXVECTOR, _mm512_loadu_ps(&y[idx_m + 0])));
Expand Down Expand Up @@ -145,8 +145,8 @@ static int sgemv_kernel_t_2(BLASLONG m, float alpha, float *a, float *x, float *
}
if (tag_m_32x != m) {
for (BLASLONG idx_m = tag_m_32x; idx_m < tag_m_16x; idx_m+=16) {
m0 = _mm512_loadu_ps(&a[idx_m]);
m1 = _mm512_loadu_ps(&a[idx_m + 16]);
m0 = _mm512_loadu_ps(&a[idx_m*2]);
m1 = _mm512_loadu_ps(&a[idx_m*2 + 16]);
col1_1 = _mm512_permutex2var_ps(m0, idx_base_0, m1);
col1_2 = _mm512_permutex2var_ps(m0, idx_base_1, m1);
_mm512_storeu_ps(&y[idx_m], _mm512_add_ps(_mm512_fmadd_ps(x2Array, col1_2, _mm512_mul_ps(col1_1, x1Array)), _mm512_loadu_ps(&y[idx_m])));
Expand All @@ -157,7 +157,7 @@ static int sgemv_kernel_t_2(BLASLONG m, float alpha, float *a, float *x, float *
__mmask8 load_mask = *((__mmask8*) &load_mask_value);
x1Array = _mm512_broadcast_f32x2(_mm_maskz_loadu_ps(load_mask, x));
for (BLASLONG idx_m = tag_m_16x; idx_m < tag_m_8x; idx_m+=8) {
m0 = _mm512_loadu_ps(&a[idx_m]);
m0 = _mm512_loadu_ps(&a[idx_m*2]);
m1 = _mm512_mul_ps(_mm512_mul_ps(m0, x1Array), ALPHAVECTOR);
m2 = _mm512_permutexvar_ps(_mm512_set_epi32(15, 13, 11, 9, 7, 5, 3, 1, 14, 12, 10, 8, 6, 4, 2, 0), m1);
__m256 ret = _mm256_add_ps(_mm512_extractf32x8_ps(m2, 1), _mm512_extractf32x8_ps(m2, 0));
Expand All @@ -166,12 +166,12 @@ static int sgemv_kernel_t_2(BLASLONG m, float alpha, float *a, float *x, float *
}

if (tag_m_8x != m) {
unsigned short tail_mask_value = (((unsigned int)0xffff) >> (16-((m-tag_m_8x)*2)&15));
unsigned short tail_mask_value = (((unsigned int)0xffff) >> (16-(((m-tag_m_8x)*2)&15)));
__mmask16 a_mask = *((__mmask16*) &tail_mask_value);
unsigned char y_mask_value = (((unsigned char)0xff) >> (8-(m-tag_m_8x)));
__mmask8 y_mask = *((__mmask8*) &y_mask_value);

m0 = _mm512_maskz_loadu_ps(a_mask, &a[tag_m_8x]);
m0 = _mm512_maskz_loadu_ps(a_mask, &a[tag_m_8x*2]);
m1 = _mm512_mul_ps(_mm512_mul_ps(m0, x1Array), ALPHAVECTOR);
m2 = _mm512_permutexvar_ps(_mm512_set_epi32(15, 13, 11, 9, 7, 5, 3, 1, 14, 12, 10, 8, 6, 4, 2, 0), m1);
__m256 ret = _mm256_add_ps(_mm512_extractf32x8_ps(m2, 1), _mm512_extractf32x8_ps(m2, 0));
Expand Down Expand Up @@ -322,7 +322,7 @@ static int sgemv_kernel_t_4(BLASLONG m, float alpha, float *a, float *x, float *
{
BLASLONG tag_m_4x = m & (~3);
BLASLONG tag_m_2x = m & (~1);
__m512 m0, m1, m2;
__m512 m0, m1;
__m256 m256_0, m256_1, c256_1, c256_2;
__m128 c1, c2, c3, c4, ret;
__m128 xarray = _mm_maskz_loadu_ps(0x0f, x);
Expand All @@ -346,7 +346,7 @@ static int sgemv_kernel_t_4(BLASLONG m, float alpha, float *a, float *x, float *
c3 = _mm256_extractf32x4_ps(c256_2, 0);
c4 = _mm256_extractf32x4_ps(c256_2, 1);

ret = _mm_maskz_add_ps(0xff, _mm_maskz_add_ps(0xff, _mm_maskz_add_ps(0xff, c1, c2), _mm_maskz_add_ps(0xff, c3, c4)), _mm_maskz_loadu_ps(0xff, y));
ret = _mm_maskz_add_ps(0xff, _mm_maskz_add_ps(0xff, _mm_maskz_add_ps(0xff, c1, c2), _mm_maskz_add_ps(0xff, c3, c4)), _mm_maskz_loadu_ps(0xff, &y[idx_m]));
_mm_mask_storeu_ps(&y[idx_m], 0xff, ret);
}

Expand Down Expand Up @@ -958,6 +958,7 @@ static int sgemv_kernel_t_7(BLASLONG m, float alpha, float *a, float *x, float *
c256_1 = _mm512_extractf32x8_ps(tmp0, 1);

c256_0 = _mm256_add_ps(c256_0, c256_1);
c256_0 = _mm256_mul_ps(c256_0, alpha256);

__m128 c128_0 = _mm256_extractf32x4_ps(c256_0, 0);
__m128 c128_1 = _mm256_extractf32x4_ps(c256_0, 1);
Expand Down Expand Up @@ -1016,9 +1017,10 @@ static int sgemv_kernel_t_8(BLASLONG m, float alpha, float *a, float *x, float *
__m512 m0, m1, m2, m3;
__m256 r0, r1, r2, r3, r4, r5, r6, r7, tmp0, tmp1, tmp2, tmp3;
__m128 c128_0, c128_1, c128_2, c128_3;
__m128 alpha128 = _mm_set1_ps(alpha);
__m256 alpha256 = _mm256_set1_ps(alpha);

__m256 x256 = _mm256_loadu_ps(x);
x256 = _mm256_mul_ps(x256, alpha256);
__m512 x512 = _mm512_broadcast_f32x8(x256);

for(BLASLONG idx_m=0; idx_m<tag_m_8x; idx_m+=8) {
Expand Down Expand Up @@ -1053,8 +1055,8 @@ static int sgemv_kernel_t_8(BLASLONG m, float alpha, float *a, float *x, float *

c128_0 = _mm_add_ps(c128_0, c128_1);
c128_2 = _mm_add_ps(c128_2, c128_3);
_mm_storeu_ps(&y[idx_m], _mm_fmadd_ps(c128_0, alpha128, _mm_loadu_ps(&y[idx_m])));
_mm_storeu_ps(&y[idx_m+4], _mm_fmadd_ps(c128_2, alpha128, _mm_loadu_ps(&y[idx_m+4])));
_mm_storeu_ps(&y[idx_m], _mm_add_ps(c128_0, _mm_loadu_ps(&y[idx_m])));
_mm_storeu_ps(&y[idx_m+4], _mm_add_ps(c128_2, _mm_loadu_ps(&y[idx_m+4])));
}

if (tag_m_8x !=m ){
Expand All @@ -1078,7 +1080,7 @@ static int sgemv_kernel_t_8(BLASLONG m, float alpha, float *a, float *x, float *
c128_1 = _mm256_extractf32x4_ps(tmp1, 1);

c128_0 = _mm_add_ps(c128_0, c128_1);
_mm_storeu_ps(&y[idx_m], _mm_fmadd_ps(c128_0, alpha128, _mm_loadu_ps(&y[idx_m])));
_mm_storeu_ps(&y[idx_m], _mm_add_ps(c128_0, _mm_loadu_ps(&y[idx_m])));

}

Expand All @@ -1094,7 +1096,6 @@ static int sgemv_kernel_t_8(BLASLONG m, float alpha, float *a, float *x, float *
c128_1 = _mm256_extractf32x4_ps(tmp0, 1);

c128_0 = _mm_add_ps(c128_0, c128_1);
c128_0 = _mm_mul_ps(c128_0, alpha128);

_mm_storeu_ps(ret, c128_0);
y[idx_m] += (ret[0]+ret[1]);
Expand Down