Merge pull request #817 from wernsaar/develop
added optimized zaxpy kernel for POWER8
This commit is contained in:
commit
8519e4ed9f
|
@ -113,7 +113,7 @@ ZTRSMKERNEL_RT = ../generic/trsm_kernel_RT.c
|
|||
#SAXPYKERNEL = ../arm/axpy.c
|
||||
DAXPYKERNEL = daxpy.c
|
||||
#CAXPYKERNEL = ../arm/zaxpy.c
|
||||
#ZAXPYKERNEL = ../arm/zaxpy.c
|
||||
ZAXPYKERNEL = zaxpy.c
|
||||
#
|
||||
#SCOPYKERNEL = ../arm/copy.c
|
||||
#DCOPYKERNEL = ../arm/copy.c
|
||||
|
|
|
@ -0,0 +1,140 @@
|
|||
/***************************************************************************
|
||||
Copyright (c) 2013-2016, 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.
|
||||
*****************************************************************************/
|
||||
|
||||
/**************************************************************************************
|
||||
* 2016/03/23 Werner Saar (wernsaar@googlemail.com)
|
||||
* BLASTEST : OK
|
||||
* CTEST : OK
|
||||
* TEST : OK
|
||||
* LAPACK-TEST : OK
|
||||
**************************************************************************************/
|
||||
|
||||
#include "common.h"
|
||||
|
||||
|
||||
#if defined(POWER8)
|
||||
#include "zaxpy_microk_power8.c"
|
||||
#endif
|
||||
|
||||
|
||||
#ifndef HAVE_KERNEL_4
|
||||
|
||||
static void zaxpy_kernel_4(BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha)
|
||||
{
|
||||
BLASLONG register i = 0;
|
||||
BLASLONG register ix = 0;
|
||||
FLOAT da_r = alpha[0];
|
||||
FLOAT da_i = alpha[1];
|
||||
|
||||
|
||||
while(i < n)
|
||||
{
|
||||
#if !defined(CONJ)
|
||||
y[ix] += ( da_r * x[ix] - da_i * x[ix+1] ) ;
|
||||
y[ix+1] += ( da_r * x[ix+1] + da_i * x[ix] ) ;
|
||||
y[ix+2] += ( da_r * x[ix+2] - da_i * x[ix+3] ) ;
|
||||
y[ix+3] += ( da_r * x[ix+3] + da_i * x[ix+2] ) ;
|
||||
#else
|
||||
y[ix] += ( da_r * x[ix] + da_i * x[ix+1] ) ;
|
||||
y[ix+1] -= ( da_r * x[ix+1] - da_i * x[ix] ) ;
|
||||
y[ix+2] += ( da_r * x[ix+2] + da_i * x[ix+3] ) ;
|
||||
y[ix+3] -= ( da_r * x[ix+3] - da_i * x[ix+2] ) ;
|
||||
#endif
|
||||
|
||||
ix+=4 ;
|
||||
i+=2 ;
|
||||
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
#endif
|
||||
|
||||
int CNAME(BLASLONG n, BLASLONG dummy0, BLASLONG dummy1, FLOAT da_r, FLOAT da_i, FLOAT *x, BLASLONG inc_x, FLOAT *y, BLASLONG inc_y, FLOAT *dummy, BLASLONG dummy2)
|
||||
{
|
||||
BLASLONG i=0;
|
||||
BLASLONG ix=0,iy=0;
|
||||
FLOAT da[4];
|
||||
|
||||
if ( n <= 0 ) return(0);
|
||||
|
||||
if ( (inc_x == 1) && (inc_y == 1) )
|
||||
{
|
||||
|
||||
BLASLONG n1 = n & -16;
|
||||
|
||||
if ( n1 )
|
||||
{
|
||||
da[0] = da_r;
|
||||
da[1] = da_r;
|
||||
da[2] = da_i;
|
||||
da[3] = da_i;
|
||||
zaxpy_kernel_4(n1, x, y , da );
|
||||
ix = 2 * n1;
|
||||
}
|
||||
i = n1;
|
||||
while(i < n)
|
||||
{
|
||||
#if !defined(CONJ)
|
||||
y[ix] += ( da_r * x[ix] - da_i * x[ix+1] ) ;
|
||||
y[ix+1] += ( da_r * x[ix+1] + da_i * x[ix] ) ;
|
||||
#else
|
||||
y[ix] += ( da_r * x[ix] + da_i * x[ix+1] ) ;
|
||||
y[ix+1] -= ( da_r * x[ix+1] - da_i * x[ix] ) ;
|
||||
#endif
|
||||
i++ ;
|
||||
ix += 2;
|
||||
|
||||
}
|
||||
return(0);
|
||||
|
||||
|
||||
}
|
||||
|
||||
inc_x *=2;
|
||||
inc_y *=2;
|
||||
|
||||
while(i < n)
|
||||
{
|
||||
|
||||
#if !defined(CONJ)
|
||||
y[iy] += ( da_r * x[ix] - da_i * x[ix+1] ) ;
|
||||
y[iy+1] += ( da_r * x[ix+1] + da_i * x[ix] ) ;
|
||||
#else
|
||||
y[iy] += ( da_r * x[ix] + da_i * x[ix+1] ) ;
|
||||
y[iy+1] -= ( da_r * x[ix+1] - da_i * x[ix] ) ;
|
||||
#endif
|
||||
ix += inc_x ;
|
||||
iy += inc_y ;
|
||||
i++ ;
|
||||
|
||||
}
|
||||
return(0);
|
||||
|
||||
}
|
||||
|
||||
|
|
@ -0,0 +1,250 @@
|
|||
/***************************************************************************
|
||||
Copyright (c) 2013-2016, 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.
|
||||
*****************************************************************************/
|
||||
|
||||
/**************************************************************************************
|
||||
* 2016/03/23 Werner Saar (wernsaar@googlemail.com)
|
||||
* BLASTEST : OK
|
||||
* CTEST : OK
|
||||
* TEST : OK
|
||||
* LAPACK-TEST : OK
|
||||
**************************************************************************************/
|
||||
|
||||
|
||||
#define HAVE_KERNEL_4 1
|
||||
static void zaxpy_kernel_4( BLASLONG n, FLOAT *x, FLOAT *y , FLOAT *alpha) __attribute__ ((noinline));
|
||||
|
||||
static void zaxpy_kernel_4( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha)
|
||||
{
|
||||
|
||||
|
||||
BLASLONG i = n;
|
||||
BLASLONG o16 = 16;
|
||||
BLASLONG o32 = 32;
|
||||
BLASLONG o48 = 48;
|
||||
FLOAT *x1=x;
|
||||
FLOAT *y1=y;
|
||||
FLOAT *y2=y+1;
|
||||
BLASLONG pre = 384;
|
||||
|
||||
#if !defined(CONJ)
|
||||
FLOAT mvec[2] = { -1.0, 1.0 };
|
||||
#else
|
||||
FLOAT mvec[2] = { 1.0, -1.0 };
|
||||
#endif
|
||||
|
||||
|
||||
__asm__ __volatile__
|
||||
(
|
||||
|
||||
"lxsdx 34, 0 , %4 \n\t" // alpha_r
|
||||
"lxsdx 35, %5, %4 \n\t" // alpha_i
|
||||
"xxspltd 32, 34, 0 \n\t"
|
||||
"xxspltd 33, 35, 0 \n\t"
|
||||
|
||||
"lxvd2x 36, 0, %9 \n\t" // mvec
|
||||
|
||||
#if !defined(CONJ)
|
||||
"xvmuldp 33, 33 , 36 \n\t" // alpha_i * mvec
|
||||
#else
|
||||
"xvmuldp 32, 32 , 36 \n\t" // alpha_r * mvec
|
||||
#endif
|
||||
|
||||
"addi %8, %8, -8 \n\t"
|
||||
|
||||
"dcbt %2, %10 \n\t"
|
||||
"dcbt %3, %10 \n\t"
|
||||
|
||||
|
||||
"lxvd2x 40, 0, %2 \n\t" // x0
|
||||
"lxvd2x 41, %5, %2 \n\t" // x1
|
||||
"lxvd2x 42, %6, %2 \n\t" // x2
|
||||
"lxvd2x 43, %7, %2 \n\t" // x3
|
||||
|
||||
"lxvd2x 48, 0, %3 \n\t" // y0
|
||||
"lxvd2x 49, %5, %3 \n\t" // y1
|
||||
"lxvd2x 50, %6, %3 \n\t" // y2
|
||||
"lxvd2x 51, %7, %3 \n\t" // y3
|
||||
|
||||
"xxswapd 56, 40 \n\t" // exchange real and imag part
|
||||
"xxswapd 57, 41 \n\t" // exchange real and imag part
|
||||
"xxswapd 58, 42 \n\t" // exchange real and imag part
|
||||
"xxswapd 59, 43 \n\t" // exchange real and imag part
|
||||
|
||||
"addi %2, %2, 64 \n\t"
|
||||
"addi %3, %3, 64 \n\t"
|
||||
|
||||
"lxvd2x 44, 0, %2 \n\t" // x4
|
||||
"lxvd2x 45, %5, %2 \n\t" // x5
|
||||
"lxvd2x 46, %6, %2 \n\t" // x6
|
||||
"lxvd2x 47, %7, %2 \n\t" // x7
|
||||
|
||||
"lxvd2x 52, 0, %3 \n\t" // y4
|
||||
"lxvd2x 53, %5, %3 \n\t" // y5
|
||||
"lxvd2x 54, %6, %3 \n\t" // y6
|
||||
"lxvd2x 55, %7, %3 \n\t" // y7
|
||||
|
||||
"xxswapd 60, 44 \n\t" // exchange real and imag part
|
||||
"xxswapd 61, 45 \n\t" // exchange real and imag part
|
||||
"xxswapd 62, 46 \n\t" // exchange real and imag part
|
||||
"xxswapd 63, 47 \n\t" // exchange real and imag part
|
||||
|
||||
"addi %2, %2, 64 \n\t"
|
||||
"addi %3, %3, 64 \n\t"
|
||||
|
||||
"addic. %0 , %0 , -8 \n\t"
|
||||
"ble 2f \n\t"
|
||||
|
||||
".align 5 \n\t"
|
||||
"1: \n\t"
|
||||
|
||||
"dcbt %2, %10 \n\t"
|
||||
"dcbt %3, %10 \n\t"
|
||||
|
||||
"xvmaddadp 48, 40, 32 \n\t" // alpha_r * x0_r , alpha_r * x0_i
|
||||
"xvmaddadp 49, 41, 32 \n\t"
|
||||
"lxvd2x 40, 0, %2 \n\t" // x0
|
||||
"lxvd2x 41, %5, %2 \n\t" // x1
|
||||
"xvmaddadp 50, 42, 32 \n\t"
|
||||
"xvmaddadp 51, 43, 32 \n\t"
|
||||
"lxvd2x 42, %6, %2 \n\t" // x2
|
||||
"lxvd2x 43, %7, %2 \n\t" // x3
|
||||
|
||||
"xvmaddadp 52, 44, 32 \n\t"
|
||||
"addi %2, %2, 64 \n\t"
|
||||
"xvmaddadp 53, 45, 32 \n\t"
|
||||
"lxvd2x 44, 0, %2 \n\t" // x4
|
||||
"lxvd2x 45, %5, %2 \n\t" // x5
|
||||
"xvmaddadp 54, 46, 32 \n\t"
|
||||
"xvmaddadp 55, 47, 32 \n\t"
|
||||
"lxvd2x 46, %6, %2 \n\t" // x6
|
||||
"lxvd2x 47, %7, %2 \n\t" // x7
|
||||
|
||||
"xvmaddadp 48, 56, 33 \n\t" // alpha_i * x0_i , alpha_i * x0_r
|
||||
"addi %2, %2, 64 \n\t"
|
||||
"xvmaddadp 49, 57, 33 \n\t"
|
||||
"xvmaddadp 50, 58, 33 \n\t"
|
||||
"xvmaddadp 51, 59, 33 \n\t"
|
||||
|
||||
"xvmaddadp 52, 60, 33 \n\t"
|
||||
"xvmaddadp 53, 61, 33 \n\t"
|
||||
"xvmaddadp 54, 62, 33 \n\t"
|
||||
"xvmaddadp 55, 63, 33 \n\t"
|
||||
|
||||
"stxvd2x 48, 0, %8 \n\t"
|
||||
"stxvd2x 49, %5, %8 \n\t"
|
||||
"stxvd2x 50, %6, %8 \n\t"
|
||||
"stxvd2x 51, %7, %8 \n\t"
|
||||
|
||||
"addi %8, %8, 64 \n\t"
|
||||
|
||||
"stxvd2x 52, 0, %8 \n\t"
|
||||
"stxvd2x 53, %5, %8 \n\t"
|
||||
"stxvd2x 54, %6, %8 \n\t"
|
||||
"stxvd2x 55, %7, %8 \n\t"
|
||||
|
||||
"addi %8, %8, 64 \n\t"
|
||||
|
||||
"xxswapd 56, 40 \n\t" // exchange real and imag part
|
||||
"xxswapd 57, 41 \n\t" // exchange real and imag part
|
||||
"lxvd2x 48, 0, %3 \n\t" // y0
|
||||
"lxvd2x 49, %5, %3 \n\t" // y1
|
||||
"xxswapd 58, 42 \n\t" // exchange real and imag part
|
||||
"xxswapd 59, 43 \n\t" // exchange real and imag part
|
||||
"lxvd2x 50, %6, %3 \n\t" // y2
|
||||
"lxvd2x 51, %7, %3 \n\t" // y3
|
||||
|
||||
"xxswapd 60, 44 \n\t" // exchange real and imag part
|
||||
"addi %3, %3, 64 \n\t"
|
||||
"xxswapd 61, 45 \n\t" // exchange real and imag part
|
||||
"lxvd2x 52, 0, %3 \n\t" // y4
|
||||
"lxvd2x 53, %5, %3 \n\t" // y5
|
||||
"xxswapd 62, 46 \n\t" // exchange real and imag part
|
||||
"xxswapd 63, 47 \n\t" // exchange real and imag part
|
||||
"lxvd2x 54, %6, %3 \n\t" // y6
|
||||
"lxvd2x 55, %7, %3 \n\t" // y7
|
||||
|
||||
"addi %3, %3, 64 \n\t"
|
||||
|
||||
"addic. %0 , %0 , -8 \n\t"
|
||||
"bgt 1b \n\t"
|
||||
|
||||
"2: \n\t"
|
||||
|
||||
"xvmaddadp 48, 40, 32 \n\t" // alpha_r * x0_r , alpha_r * x0_i
|
||||
"xvmaddadp 49, 41, 32 \n\t"
|
||||
"xvmaddadp 50, 42, 32 \n\t"
|
||||
"xvmaddadp 51, 43, 32 \n\t"
|
||||
|
||||
"xvmaddadp 52, 44, 32 \n\t"
|
||||
"xvmaddadp 53, 45, 32 \n\t"
|
||||
"xvmaddadp 54, 46, 32 \n\t"
|
||||
"xvmaddadp 55, 47, 32 \n\t"
|
||||
|
||||
"xvmaddadp 48, 56, 33 \n\t" // alpha_i * x0_i , alpha_i * x0_r
|
||||
"xvmaddadp 49, 57, 33 \n\t"
|
||||
"xvmaddadp 50, 58, 33 \n\t"
|
||||
"xvmaddadp 51, 59, 33 \n\t"
|
||||
|
||||
"xvmaddadp 52, 60, 33 \n\t"
|
||||
"xvmaddadp 53, 61, 33 \n\t"
|
||||
"xvmaddadp 54, 62, 33 \n\t"
|
||||
"xvmaddadp 55, 63, 33 \n\t"
|
||||
|
||||
|
||||
"stxvd2x 48, 0, %8 \n\t"
|
||||
"stxvd2x 49, %5, %8 \n\t"
|
||||
"stxvd2x 50, %6, %8 \n\t"
|
||||
"stxvd2x 51, %7, %8 \n\t"
|
||||
|
||||
"addi %8, %8, 64 \n\t"
|
||||
|
||||
"stxvd2x 52, 0, %8 \n\t"
|
||||
"stxvd2x 53, %5, %8 \n\t"
|
||||
"stxvd2x 54, %6, %8 \n\t"
|
||||
"stxvd2x 55, %7, %8 \n\t"
|
||||
|
||||
"addi %8, %8, 64 \n\t"
|
||||
|
||||
:
|
||||
:
|
||||
"r" (i), // 0
|
||||
"r" (n), // 1
|
||||
"r" (x1), // 2
|
||||
"r" (y1), // 3
|
||||
"r" (alpha), // 4
|
||||
"r" (o16), // 5
|
||||
"r" (o32), // 6
|
||||
"r" (o48), // 7
|
||||
"r" (y2), // 8
|
||||
"r" (mvec), // 9
|
||||
"r" (pre) // 10
|
||||
: "cr0", "%0", "%2" , "%3", "%8", "memory"
|
||||
);
|
||||
|
||||
}
|
||||
|
||||
|
Loading…
Reference in New Issue