Merge pull request #1996 from quickwritereader/develop
NBMAX=4096 for gemvn, added sgemvn 8x8 for future
This commit is contained in:
Martin Kroeker
GitHub
commit
729e925174
|
|
@ -28,7 +28,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
|||
|
||||
#include "common.h"
|
||||
|
||||
#define NBMAX 2048
|
||||
#define NBMAX 4096
|
||||
|
||||
static void sgemv_kernel_4x8(BLASLONG n, FLOAT **ap, FLOAT *xo, FLOAT *y, BLASLONG lda4, FLOAT *alpha)
|
||||
{
|
||||
|
|
|
|||
|
|
@ -0,0 +1,513 @@
|
|||
/***************************************************************************
|
||||
Copyright (c) 2019, The OpenBLAS Project
|
||||
All rights reserved.
|
||||
Redistribution and use in source and binary forms, with or without
|
||||
modification, are permitted provided that the following conditions are
|
||||
met:
|
||||
1. Redistributions of source code must retain the above copyright
|
||||
notice, this list of conditions and the following disclaimer.
|
||||
2. Redistributions in binary form must reproduce the above copyright
|
||||
notice, this list of conditions and the following disclaimer in
|
||||
the documentation and/or other materials provided with the
|
||||
distribution.
|
||||
3. Neither the name of the OpenBLAS project nor the names of
|
||||
its contributors may be used to endorse or promote products
|
||||
derived from this software without specific prior written permission.
|
||||
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
|
||||
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
||||
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
||||
ARE DISCLAIMED. IN NO EVENT SHALL THE OPENBLAS PROJECT OR CONTRIBUTORS BE
|
||||
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
||||
DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
|
||||
SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
|
||||
CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
|
||||
OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE
|
||||
USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
||||
*****************************************************************************/
|
||||
|
||||
|
||||
/****Note***
|
||||
UnUsed kernel
|
||||
This kernel works. But it was not competitive enough to be added in production
|
||||
It could be used and tested in future or could provide barebone for switching to inline assembly
|
||||
*/
|
||||
|
||||
#include "common.h"
|
||||
|
||||
#define NBMAX 4096
|
||||
|
||||
static void sgemv_kernel_8x8(BLASLONG n, FLOAT **ap, FLOAT *xo, FLOAT *y, BLASLONG lda4, FLOAT *alpha)
|
||||
{
|
||||
|
||||
BLASLONG i;
|
||||
FLOAT *a0,*a1,*a2,*a3,*b0,*b1,*b2,*b3;
|
||||
FLOAT x0,x1,x2,x3,x4,x5,x6,x7;
|
||||
a0 = ap[0];
|
||||
a1 = ap[1];
|
||||
a2 = ap[2];
|
||||
a3 = ap[3];
|
||||
b0 = a0 + lda4 ;
|
||||
b1 = a1 + lda4 ;
|
||||
b2 = a2 + lda4 ;
|
||||
b3 = a3 + lda4 ;
|
||||
x0 = xo[0] * *alpha;
|
||||
x1 = xo[1] * *alpha;
|
||||
x2 = xo[2] * *alpha;
|
||||
x3 = xo[3] * *alpha;
|
||||
x4 = xo[4] * *alpha;
|
||||
x5 = xo[5] * *alpha;
|
||||
x6 = xo[6] * *alpha;
|
||||
x7 = xo[7] * *alpha;
|
||||
__vector float* va0 = (__vector float*)a0;
|
||||
__vector float* va1 = (__vector float*)a1;
|
||||
__vector float* va2 = (__vector float*)a2;
|
||||
__vector float* va3 = (__vector float*)a3;
|
||||
__vector float* vb0 = (__vector float*)b0;
|
||||
__vector float* vb1 = (__vector float*)b1;
|
||||
__vector float* vb2 = (__vector float*)b2;
|
||||
__vector float* vb3 = (__vector float*)b3;
|
||||
|
||||
register __vector float v_x0 = {x0,x0,x0,x0};
|
||||
register __vector float v_x1 = {x1,x1,x1,x1};
|
||||
register __vector float v_x2 = {x2,x2,x2,x2};
|
||||
register __vector float v_x3 = {x3,x3,x3,x3};
|
||||
register __vector float v_x4 = {x4,x4,x4,x4};
|
||||
register __vector float v_x5 = {x5,x5,x5,x5};
|
||||
register __vector float v_x6 = {x6,x6,x6,x6};
|
||||
register __vector float v_x7 = {x7,x7,x7,x7};
|
||||
__vector float* v_y =(__vector float*)y;
|
||||
|
||||
for ( i=0; i< n/4; i+=2)
|
||||
{
|
||||
register __vector float vy_1=v_y[i];
|
||||
register __vector float vy_2=v_y[i+1];
|
||||
register __vector float va0_1=va0[i] ;
|
||||
register __vector float va0_2=va0[i+1] ;
|
||||
register __vector float va1_1=va1[i] ;
|
||||
register __vector float va1_2=va1[i+1] ;
|
||||
register __vector float va2_1=va2[i] ;
|
||||
register __vector float va2_2=va2[i+1] ;
|
||||
register __vector float va3_1=va3[i] ;
|
||||
register __vector float va3_2=va3[i+1] ;
|
||||
register __vector float vb0_1=vb0[i] ;
|
||||
register __vector float vb0_2=vb0[i+1] ;
|
||||
register __vector float vb1_1=vb1[i] ;
|
||||
register __vector float vb1_2=vb1[i+1] ;
|
||||
register __vector float vb2_1=vb2[i] ;
|
||||
register __vector float vb2_2=vb2[i+1] ;
|
||||
register __vector float vb3_1=vb3[i] ;
|
||||
register __vector float vb3_2=vb3[i+1] ;
|
||||
vy_1 += v_x0 * va0_1 + v_x1 * va1_1 + v_x2 * va2_1 + v_x3 * va3_1 ;
|
||||
vy_1 += v_x4 * vb0_1 + v_x5 * vb1_1 + v_x6 * vb2_1 + v_x7 * vb3_1 ;
|
||||
vy_2 += v_x0 * va0_2 + v_x1 * va1_2 + v_x2 * va2_2 + v_x3 * va3_2 ;
|
||||
vy_2 += v_x4 * vb0_2 + v_x5 * vb1_2 + v_x6 * vb2_2 + v_x7 * vb3_2 ;
|
||||
v_y[i] =vy_1;
|
||||
v_y[i+1] =vy_2;
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
static void sgemv_kernel_8x4(BLASLONG n, FLOAT **ap, FLOAT *xo, FLOAT *y, FLOAT *alpha)
|
||||
{
|
||||
BLASLONG i;
|
||||
FLOAT x0,x1,x2,x3;
|
||||
x0 = xo[0] * *alpha;
|
||||
x1 = xo[1] * *alpha;
|
||||
x2 = xo[2] * *alpha;
|
||||
x3 = xo[3] * *alpha;
|
||||
__vector float v_x0 = {x0,x0,x0,x0};
|
||||
__vector float v_x1 = {x1,x1,x1,x1};
|
||||
__vector float v_x2 = {x2,x2,x2,x2};
|
||||
__vector float v_x3 = {x3,x3,x3,x3};
|
||||
__vector float* v_y =(__vector float*)y;
|
||||
__vector float* va0 = (__vector float*)ap[0];
|
||||
__vector float* va1 = (__vector float*)ap[1];
|
||||
__vector float* va2 = (__vector float*)ap[2];
|
||||
__vector float* va3 = (__vector float*)ap[3];
|
||||
|
||||
for ( i=0; i< n/4; i+=2 )
|
||||
{
|
||||
register __vector float vy_1=v_y[i];
|
||||
register __vector float vy_2=v_y[i+1];
|
||||
register __vector float va0_1=va0[i] ;
|
||||
register __vector float va0_2=va0[i+1] ;
|
||||
register __vector float va1_1=va1[i] ;
|
||||
register __vector float va1_2=va1[i+1] ;
|
||||
register __vector float va2_1=va2[i] ;
|
||||
register __vector float va2_2=va2[i+1] ;
|
||||
register __vector float va3_1=va3[i] ;
|
||||
register __vector float va3_2=va3[i+1] ;
|
||||
vy_1 += v_x0 * va0_1 + v_x1 * va1_1 + v_x2 * va2_1 + v_x3 * va3_1 ;
|
||||
vy_2 += v_x0 * va0_2 + v_x1 * va1_2 + v_x2 * va2_2 + v_x3 * va3_2 ;
|
||||
v_y[i] =vy_1;
|
||||
v_y[i+1] =vy_2;
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
static void sgemv_kernel_8x2( BLASLONG n, FLOAT **ap, FLOAT *x, FLOAT *y, FLOAT *alpha)
|
||||
{
|
||||
|
||||
BLASLONG i;
|
||||
FLOAT x0,x1;
|
||||
x0 = x[0] * *alpha;
|
||||
x1 = x[1] * *alpha;
|
||||
__vector float v_x0 = {x0,x0,x0,x0};
|
||||
__vector float v_x1 = {x1,x1,x1,x1};
|
||||
__vector float* v_y =(__vector float*)y;
|
||||
__vector float* va0 = (__vector float*)ap[0];
|
||||
__vector float* va1 = (__vector float*)ap[1];
|
||||
|
||||
for ( i=0; i< n/4; i+=2 )
|
||||
{
|
||||
v_y[i] += v_x0 * va0[i] + v_x1 * va1[i] ;
|
||||
v_y[i+1] += v_x0 * va0[i+1] + v_x1 * va1[i+1] ;
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
|
||||
static void sgemv_kernel_8x1(BLASLONG n, FLOAT *ap, FLOAT *x, FLOAT *y, FLOAT *alpha)
|
||||
{
|
||||
|
||||
BLASLONG i;
|
||||
FLOAT x0 ;
|
||||
x0 = x[0] * *alpha;
|
||||
__vector float v_x0 = {x0,x0,x0,x0};
|
||||
__vector float* v_y =(__vector float*)y;
|
||||
__vector float* va0 = (__vector float*)ap;
|
||||
|
||||
for ( i=0; i< n/4; i+=2 )
|
||||
{
|
||||
v_y[i] += v_x0 * va0[i] ;
|
||||
v_y[i+1] += v_x0 * va0[i+1] ;
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
static void add_y(BLASLONG n, FLOAT *src, FLOAT *dest, BLASLONG inc_dest)
|
||||
{
|
||||
BLASLONG i;
|
||||
|
||||
for ( i=0; i<n; i++ ){
|
||||
*dest += *src;
|
||||
src++;
|
||||
dest += inc_dest;
|
||||
}
|
||||
return;
|
||||
|
||||
|
||||
}
|
||||
|
||||
int CNAME(BLASLONG m, BLASLONG n, BLASLONG dummy1, FLOAT alpha, FLOAT *a, BLASLONG lda, FLOAT *x, BLASLONG inc_x, FLOAT *y, BLASLONG inc_y, FLOAT *buffer)
|
||||
{
|
||||
BLASLONG i;
|
||||
FLOAT *a_ptr;
|
||||
FLOAT *x_ptr;
|
||||
FLOAT *y_ptr;
|
||||
FLOAT *ap[4];
|
||||
BLASLONG n1;
|
||||
BLASLONG m1;
|
||||
BLASLONG m2;
|
||||
BLASLONG m3;
|
||||
BLASLONG n2;
|
||||
BLASLONG lda4 = lda << 2;
|
||||
BLASLONG lda8 = lda << 3;
|
||||
FLOAT xbuffer[8],*ybuffer;
|
||||
|
||||
if ( m < 1 ) return(0);
|
||||
if ( n < 1 ) return(0);
|
||||
|
||||
ybuffer = buffer;
|
||||
|
||||
if ( inc_x == 1 )
|
||||
{
|
||||
n1 = n >> 3 ;
|
||||
n2 = n & 7 ;
|
||||
}
|
||||
else
|
||||
{
|
||||
n1 = n >> 2 ;
|
||||
n2 = n & 3 ;
|
||||
|
||||
}
|
||||
|
||||
m3 = m & 7 ;
|
||||
m1 = m - m3;
|
||||
m2 = (m & (NBMAX-1)) - m3 ;
|
||||
|
||||
|
||||
y_ptr = y;
|
||||
|
||||
BLASLONG NB = NBMAX;
|
||||
|
||||
while ( NB == NBMAX )
|
||||
{
|
||||
|
||||
m1 -= NB;
|
||||
if ( m1 < 0)
|
||||
{
|
||||
if ( m2 == 0 ) break;
|
||||
NB = m2;
|
||||
}
|
||||
|
||||
a_ptr = a;
|
||||
x_ptr = x;
|
||||
|
||||
ap[0] = a_ptr;
|
||||
ap[1] = a_ptr + lda;
|
||||
ap[2] = ap[1] + lda;
|
||||
ap[3] = ap[2] + lda;
|
||||
|
||||
if ( inc_y != 1 )
|
||||
memset(ybuffer,0,NB*4);
|
||||
else
|
||||
ybuffer = y_ptr;
|
||||
|
||||
if ( inc_x == 1 )
|
||||
{
|
||||
|
||||
|
||||
for( i = 0; i < n1 ; i++)
|
||||
{
|
||||
sgemv_kernel_8x8(NB,ap,x_ptr,ybuffer,lda4,&alpha);
|
||||
ap[0] += lda8;
|
||||
ap[1] += lda8;
|
||||
ap[2] += lda8;
|
||||
ap[3] += lda8;
|
||||
a_ptr += lda8;
|
||||
x_ptr += 8;
|
||||
}
|
||||
|
||||
|
||||
if ( n2 & 4 )
|
||||
{
|
||||
sgemv_kernel_8x4(NB,ap,x_ptr,ybuffer,&alpha);
|
||||
ap[0] += lda4;
|
||||
ap[1] += lda4;
|
||||
ap[2] += lda4;
|
||||
ap[3] += lda4;
|
||||
a_ptr += lda4;
|
||||
x_ptr += 4;
|
||||
}
|
||||
|
||||
if ( n2 & 2 )
|
||||
{
|
||||
sgemv_kernel_8x2(NB,ap,x_ptr,ybuffer,&alpha);
|
||||
a_ptr += lda*2;
|
||||
x_ptr += 2;
|
||||
}
|
||||
|
||||
|
||||
if ( n2 & 1 )
|
||||
{
|
||||
sgemv_kernel_8x1(NB,a_ptr,x_ptr,ybuffer,&alpha);
|
||||
a_ptr += lda;
|
||||
x_ptr += 1;
|
||||
}
|
||||
|
||||
|
||||
}
|
||||
else
|
||||
{
|
||||
|
||||
for( i = 0; i < n1 ; i++)
|
||||
{
|
||||
xbuffer[0] = x_ptr[0];
|
||||
x_ptr += inc_x;
|
||||
xbuffer[1] = x_ptr[0];
|
||||
x_ptr += inc_x;
|
||||
xbuffer[2] = x_ptr[0];
|
||||
x_ptr += inc_x;
|
||||
xbuffer[3] = x_ptr[0];
|
||||
x_ptr += inc_x;
|
||||
sgemv_kernel_8x4(NB,ap,xbuffer,ybuffer,&alpha);
|
||||
ap[0] += lda4;
|
||||
ap[1] += lda4;
|
||||
ap[2] += lda4;
|
||||
ap[3] += lda4;
|
||||
a_ptr += lda4;
|
||||
}
|
||||
|
||||
for( i = 0; i < n2 ; i++)
|
||||
{
|
||||
xbuffer[0] = x_ptr[0];
|
||||
x_ptr += inc_x;
|
||||
sgemv_kernel_8x1(NB,a_ptr,xbuffer,ybuffer,&alpha);
|
||||
a_ptr += lda;
|
||||
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
a += NB;
|
||||
if ( inc_y != 1 )
|
||||
{
|
||||
add_y(NB,ybuffer,y_ptr,inc_y);
|
||||
y_ptr += NB * inc_y;
|
||||
}
|
||||
else
|
||||
y_ptr += NB ;
|
||||
|
||||
}
|
||||
|
||||
|
||||
if ( m3 & 4 )
|
||||
{
|
||||
a_ptr = a;
|
||||
x_ptr = x;
|
||||
FLOAT temp0 = 0.0;
|
||||
FLOAT temp1 = 0.0;
|
||||
FLOAT temp2 = 0.0;
|
||||
FLOAT temp3 = 0.0;
|
||||
if ( lda == 4 && inc_x ==1 )
|
||||
{
|
||||
|
||||
for( i = 0; i < ( n & -4 ); i+=4 )
|
||||
{
|
||||
|
||||
temp0 += a_ptr[0] * x_ptr[0] + a_ptr[4] * x_ptr[1];
|
||||
temp1 += a_ptr[1] * x_ptr[0] + a_ptr[5] * x_ptr[1];
|
||||
temp2 += a_ptr[2] * x_ptr[0] + a_ptr[6] * x_ptr[1];
|
||||
temp3 += a_ptr[3] * x_ptr[0] + a_ptr[7] * x_ptr[1];
|
||||
|
||||
temp0 += a_ptr[8] * x_ptr[2] + a_ptr[12] * x_ptr[3];
|
||||
temp1 += a_ptr[9] * x_ptr[2] + a_ptr[13] * x_ptr[3];
|
||||
temp2 += a_ptr[10] * x_ptr[2] + a_ptr[14] * x_ptr[3];
|
||||
temp3 += a_ptr[11] * x_ptr[2] + a_ptr[15] * x_ptr[3];
|
||||
|
||||
a_ptr += 16;
|
||||
x_ptr += 4;
|
||||
}
|
||||
|
||||
for( ; i < n; i++ )
|
||||
{
|
||||
temp0 += a_ptr[0] * x_ptr[0];
|
||||
temp1 += a_ptr[1] * x_ptr[0];
|
||||
temp2 += a_ptr[2] * x_ptr[0];
|
||||
temp3 += a_ptr[3] * x_ptr[0] ;
|
||||
a_ptr +=4;
|
||||
x_ptr ++;
|
||||
}
|
||||
|
||||
}
|
||||
else
|
||||
{
|
||||
|
||||
for( i = 0; i < n; i++ )
|
||||
{
|
||||
temp0 += a_ptr[0] * x_ptr[0];
|
||||
temp1 += a_ptr[1] * x_ptr[0];
|
||||
temp2 += a_ptr[2] * x_ptr[0];
|
||||
temp3 += a_ptr[3] * x_ptr[0];
|
||||
a_ptr += lda;
|
||||
x_ptr += inc_x;
|
||||
|
||||
|
||||
}
|
||||
|
||||
}
|
||||
y_ptr[0] += alpha * temp0;
|
||||
y_ptr += inc_y;
|
||||
y_ptr[0] += alpha * temp1;
|
||||
y_ptr += inc_y;
|
||||
y_ptr[0] += alpha * temp2;
|
||||
y_ptr += inc_y;
|
||||
y_ptr[0] += alpha * temp3;
|
||||
y_ptr += inc_y;
|
||||
a += 4;
|
||||
}
|
||||
|
||||
|
||||
if ( m3 & 2 )
|
||||
{
|
||||
a_ptr = a;
|
||||
x_ptr = x;
|
||||
FLOAT temp0 = 0.0;
|
||||
FLOAT temp1 = 0.0;
|
||||
if ( lda == 2 && inc_x ==1 )
|
||||
{
|
||||
|
||||
for( i = 0; i < (n & -4) ; i+=4 )
|
||||
{
|
||||
temp0 += a_ptr[0] * x_ptr[0] + a_ptr[2] * x_ptr[1];
|
||||
temp1 += a_ptr[1] * x_ptr[0] + a_ptr[3] * x_ptr[1];
|
||||
temp0 += a_ptr[4] * x_ptr[2] + a_ptr[6] * x_ptr[3];
|
||||
temp1 += a_ptr[5] * x_ptr[2] + a_ptr[7] * x_ptr[3];
|
||||
a_ptr += 8;
|
||||
x_ptr += 4;
|
||||
|
||||
}
|
||||
|
||||
|
||||
for( ; i < n; i++ )
|
||||
{
|
||||
temp0 += a_ptr[0] * x_ptr[0];
|
||||
temp1 += a_ptr[1] * x_ptr[0];
|
||||
a_ptr += 2;
|
||||
x_ptr ++;
|
||||
}
|
||||
|
||||
}
|
||||
else
|
||||
{
|
||||
|
||||
for( i = 0; i < n; i++ )
|
||||
{
|
||||
temp0 += a_ptr[0] * x_ptr[0];
|
||||
temp1 += a_ptr[1] * x_ptr[0];
|
||||
a_ptr += lda;
|
||||
x_ptr += inc_x;
|
||||
|
||||
|
||||
}
|
||||
|
||||
}
|
||||
y_ptr[0] += alpha * temp0;
|
||||
y_ptr += inc_y;
|
||||
y_ptr[0] += alpha * temp1;
|
||||
y_ptr += inc_y;
|
||||
a += 2;
|
||||
}
|
||||
|
||||
if ( m3 & 1 )
|
||||
{
|
||||
a_ptr = a;
|
||||
x_ptr = x;
|
||||
FLOAT temp = 0.0;
|
||||
if ( lda == 1 && inc_x ==1 )
|
||||
{
|
||||
|
||||
for( i = 0; i < (n & -4); i+=4 )
|
||||
{
|
||||
temp += a_ptr[i] * x_ptr[i] + a_ptr[i+1] * x_ptr[i+1] + a_ptr[i+2] * x_ptr[i+2] + a_ptr[i+3] * x_ptr[i+3];
|
||||
|
||||
}
|
||||
|
||||
for( ; i < n; i++ )
|
||||
{
|
||||
temp += a_ptr[i] * x_ptr[i];
|
||||
}
|
||||
|
||||
}
|
||||
else
|
||||
{
|
||||
|
||||
for( i = 0; i < n; i++ )
|
||||
{
|
||||
temp += a_ptr[0] * x_ptr[0];
|
||||
a_ptr += lda;
|
||||
x_ptr += inc_x;
|
||||
}
|
||||
|
||||
}
|
||||
y_ptr[0] += alpha * temp;
|
||||
|
||||
|
||||
}
|
||||
|
||||
|
||||
return(0);
|
||||
}
|
||||
|
||||
|
||||
|
|
@ -25,9 +25,16 @@ OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE
|
|||
USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
||||
*****************************************************************************/
|
||||
|
||||
|
||||
/****Note***
|
||||
UnUsed kernel
|
||||
This kernel works. But it was not competitive enough to be added in production
|
||||
It could be used and tested in future or could be used as base for switching to inline assembly
|
||||
*/
|
||||
|
||||
#include "common.h"
|
||||
#include <stdio.h>
|
||||
#define NBMAX 2048
|
||||
#define NBMAX 4096
|
||||
|
||||
#include <altivec.h>
|
||||
|
||||
|
|
|
|||
Loading…
Reference in New Issue