OpenBLAS/kernel/loongarch64/znrm2.S

305 lines
7.0 KiB
ArmAsm

/***************************************************************************
Copyright (c) 2021, 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.
*****************************************************************************/
#define ASSEMBLER
#include "common.h"
#define N $r4
#define X $r5
#define INCX $r6
#define XX $r7
#define I $r17
#define TEMP $r18
#define a1 $f10
#define a2 $f11
#define a3 $f12
#define a4 $f13
#define a5 $f14
#define a6 $f15
#define a7 $f16
#define a8 $f17
#define t1 $f0
#define t2 $f1
#define t3 $f2
#define t4 $f3
#define s1 $f22
#define s2 $f8
#define s3 $f23
#define s4 $f9
#define ALPHA $f4
#define max $f5
PROLOGUE
#ifdef F_INTERFACE
LDINT N, 0(N)
LDINT INCX, 0(INCX)
#endif
MTC s1, $r0
bge $r0, N, .L999
slli.d INCX, INCX, ZBASE_SHIFT
beq $r0, INCX, .L999
move XX, X
MOV s2, s1
srai.d I, N, 2
MOV s3, s1
MOV s4, s1
bge $r0, I, .L15
LD a1, X, 0 * SIZE
LD a2, X, 1 * SIZE
add.d X, X, INCX
LD a3, X, 0 * SIZE
LD a4, X, 1 * SIZE
add.d X, X, INCX
LD a5, X, 0 * SIZE
LD a6, X, 1 * SIZE
add.d X, X, INCX
LD a7, X, 0 * SIZE
LD a8, X, 1 * SIZE
addi.d I, I, -1
add.d X, X, INCX
bge $r0, I, .L13
.align 3
.L12:
FABS t1, a1
LD a1, X, 0 * SIZE
FABS t2, a2
NOP
FABS t3, a3
LD a2, X, 1 * SIZE
FABS t4, a4
add.d X, X, INCX
CMPLT $fcc0, s1, t1
LD a3, X, 0 * SIZE
CMPLT $fcc1, s2, t2
NOP
CMPLT $fcc2, s3, t3
LD a4, X, 1 * SIZE
CMPLT $fcc3, s4, t4
add.d X, X, INCX
CMOVT s1, s1, t1, $fcc0
CMOVT s2, s2, t2, $fcc1
CMOVT s3, s3, t3, $fcc2
CMOVT s4, s4, t4, $fcc3
FABS t1, a5
LD a5, X, 0 * SIZE
FABS t2, a6
NOP
FABS t3, a7
LD a6, X, 1 * SIZE
FABS t4, a8
add.d X, X, INCX
CMPLT $fcc0, s1, t1
LD a7, X, 0 * SIZE
CMPLT $fcc1, s2, t2
NOP
CMPLT $fcc2, s3, t3
LD a8, X, 1 * SIZE
CMPLT $fcc3, s4, t4
add.d X, X, INCX
CMOVT s1, s1, t1, $fcc0
addi.d I, I, -1
CMOVT s2, s2, t2, $fcc1
CMOVT s3, s3, t3, $fcc2
CMOVT s4, s4, t4, $fcc3
blt $r0, I, .L12
.align 3
.L13:
FABS t1, a1
FABS t2, a2
FABS t3, a3
FABS t4, a4
CMPLT $fcc0, s1, t1
CMPLT $fcc1, s2, t2
CMPLT $fcc2, s3, t3
CMPLT $fcc3, s4, t4
CMOVT s1, s1, t1, $fcc0
CMOVT s2, s2, t2, $fcc1
CMOVT s3, s3, t3, $fcc2
CMOVT s4, s4, t4, $fcc3
FABS t1, a5
FABS t2, a6
FABS t3, a7
FABS t4, a8
CMPLT $fcc0, s1, t1
CMPLT $fcc1, s2, t2
CMPLT $fcc2, s3, t3
CMPLT $fcc3, s4, t4
CMOVT s1, s1, t1, $fcc0
CMOVT s2, s2, t2, $fcc1
CMOVT s3, s3, t3, $fcc2
CMOVT s4, s4, t4, $fcc3
.align 3
.L15:
andi I, N, 3
bge $r0, I, .L100
.align 3
.L16:
LD a1, X, 0 * SIZE
LD a2, X, 1 * SIZE
addi.d I, I, -1
FABS t1, a1
FABS t2, a2
CMPLT $fcc0, s1, t1
CMPLT $fcc1, s2, t2
CMOVT s1, s1, t1, $fcc0
CMOVT s2, s2, t2, $fcc1
add.d X, X, INCX
blt $r0, I, .L16
.align 3
.L100:
CMPLT $fcc0, s1, s2
CMPLT $fcc1, s3, s4
CMOVT s1, s1, s2, $fcc0
CMOVT s3, s3, s4, $fcc1
CMPLT $fcc0, s1, s3
CMOVT s1, s1, s3, $fcc0
lu12i.w TEMP, 0x3f800
movgr2fr.d a1, $r0
movgr2fr.w ALPHA, TEMP
CMPEQ $fcc0, s1, a1
fcvt.d.s ALPHA, ALPHA
bcnez $fcc0, .L999
fdiv.d ALPHA, ALPHA, s1
MOV max, s1
MOV s1, a1
MOV s2, a1
MOV s3, a1
MOV s4, a1
srai.d I, N, 2
bge $r0, I, .L105
LD a1, XX, 0 * SIZE
LD a2, XX, 1 * SIZE
add.d XX, XX, INCX
LD a3, XX, 0 * SIZE
LD a4, XX, 1 * SIZE
add.d XX, XX, INCX
LD a5, XX, 0 * SIZE
LD a6, XX, 1 * SIZE
add.d XX, XX, INCX
LD a7, XX, 0 * SIZE
LD a8, XX, 1 * SIZE
addi.d I, I, -1
add.d XX, XX, INCX
bge $r0, I, .L104
.align 3
.L103:
MUL t1, ALPHA, a1
LD a1, XX, 0 * SIZE
MUL t2, ALPHA, a2
addi.d I, I, -1
MUL t3, ALPHA, a3
LD a2, XX, 1 * SIZE
MUL t4, ALPHA, a4
add.d XX, XX, INCX
MADD s1, t1, t1, s1
LD a3, XX, 0 * SIZE
MADD s2, t2, t2, s2
NOP
MADD s3, t3, t3, s3
LD a4, XX, 1 * SIZE
MADD s4, t4, t4, s4
add.d XX, XX, INCX
MUL t1, ALPHA, a5
LD a5, XX, 0 * SIZE
MUL t2, ALPHA, a6
NOP
MUL t3, ALPHA, a7
LD a6, XX, 1 * SIZE
MUL t4, ALPHA, a8
add.d XX, XX, INCX
MADD s1, t1, t1, s1
LD a7, XX, 0 * SIZE
MADD s2, t2, t2, s2
LD a8, XX, 1 * SIZE
MADD s3, t3, t3, s3
add.d XX, XX, INCX
MADD s4, t4, t4, s4
blt $r0, I, .L103
.align 3
.L104:
MUL t1, ALPHA, a1
MUL t2, ALPHA, a2
MUL t3, ALPHA, a3
MUL t4, ALPHA, a4
MADD s1, t1, t1, s1
MADD s2, t2, t2, s2
MADD s3, t3, t3, s3
MADD s4, t4, t4, s4
MUL t1, ALPHA, a5
MUL t2, ALPHA, a6
MUL t3, ALPHA, a7
MUL t4, ALPHA, a8
MADD s1, t1, t1, s1
MADD s2, t2, t2, s2
MADD s3, t3, t3, s3
MADD s4, t4, t4, s4
.align 3
.L105:
andi I, N, 3
bge $r0, I, .L998
.align 3
.L106:
LD a1, XX, 0 * SIZE
LD a2, XX, 1 * SIZE
addi.d I, I, -1
MUL t1, ALPHA, a1
MUL t2, ALPHA, a2
MADD s1, t1, t1, s1
add.d XX, XX, INCX
MADD s2, t2, t2, s2
blt $r0, I, .L106
.align 3
.L998:
ADD s1, s1, s2
ADD s3, s3, s4
ADD s1, s1, s3
fsqrt.d s1, s1
move $r4, $r17
MUL $f0, max, s1
jirl $r0, $r1, 0x0
.align 3
.L999:
move $r4, $r17
fmov.d $f0, $f22
jirl $r0, $r1, 0x0
EPILOGUE