OpenBLAS/kernel/loongarch64/dgemm_ncopy_8_lsx.S

286 lines
8.9 KiB
ArmAsm

/*******************************************************************************
Copyright (c) 2023, 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"
#include "loongarch64_asm.S"
/* Function parameters */
#define M $r4 // param 1: m
#define N $r5 // param 2: n
#define SRC $r6 // param 3: src
#define LDA $r7 // param 4: lda
#define DST $r8 // param 5: dst
#define I $r9
#define J $r10
#define S1 $r12
#define S2 $r13
#define S3 $r14
#define S4 $r15
#define S5 $r16
#define S6 $r17
#define S7 $r18
#define S8 $r19
#define TD $r20
#define TS $r21
#define TL $r7
#define T0 $r6
#define ZERO $r0
#define F0 $f0
#define F1 $f1
#define F2 $f2
#define F3 $f3
#define F4 $f4
#define F5 $f5
#define F6 $f6
#define F7 $f7
/* LSX vectors */
#define U0 $vr0
#define U1 $vr1
#define U2 $vr2
#define U3 $vr3
#define U4 $vr4
#define U5 $vr5
#define U6 $vr6
#define U7 $vr7
#define D0 $vr8
#define D1 $vr9
#define D2 $vr10
#define D3 $vr11
#define D4 $vr12
#define D5 $vr13
#define D6 $vr14
#define D7 $vr15
PROLOGUE
push_if_used 0, 0
move TD, DST
move TS, SRC
slli.d TL, LDA, 0x03
slli.d T0, TL, 0x01
srai.d J, N, 0x03
beq J, ZERO, .L_N4
.L_J1:
move S1, TS
add.d S2, TS, TL
srai.d I, M, 0x03
add.d S3, S2, TL
addi.d J, J, -1
add.d S4, S3, TL
add.d S5, S3, T0
add.d S6, S4, T0
add.d S7, S5, T0
add.d S8, S6, T0
add.d TS, S7, T0
beq I, ZERO, .L_I7
.L_I1:
GLD v, , U0, S1, 0x00, U1, S2, 0x00, U2, S3, 0x00, U3, S4, 0x00, \
U4, S5, 0x00, U5, S6, 0x00, U6, S7, 0x00, U7, S8, 0x00
GINTERLACE v, d, D0, D4, U1, U0
GINTERLACE v, d, D1, D5, U3, U2
GINTERLACE v, d, D2, D6, U5, U4
GINTERLACE v, d, D3, D7, U7, U6
GST v, , D0, TD, 0x00, D1, TD, 0x10, D2, TD, 0x20, D3, TD, 0x30, \
D4, TD, 0x40, D5, TD, 0x50, D6, TD, 0x60, D7, TD, 0x70
addi.d TD, TD, 0x80
GLD v, , U0, S1, 0x10, U1, S2, 0x10, U2, S3, 0x10, U3, S4, 0x10, \
U4, S5, 0x10, U5, S6, 0x10, U6, S7, 0x10, U7, S8, 0x10
GINTERLACE v, d, D0, D4, U1, U0
GINTERLACE v, d, D1, D5, U3, U2
GINTERLACE v, d, D2, D6, U5, U4
GINTERLACE v, d, D3, D7, U7, U6
GST v, , D0, TD, 0x00, D1, TD, 0x10, D2, TD, 0x20, D3, TD, 0x30, \
D4, TD, 0x40, D5, TD, 0x50, D6, TD, 0x60, D7, TD, 0x70
addi.d TD, TD, 0x80
GLD v, , U0, S1, 0x20, U1, S2, 0x20, U2, S3, 0x20, U3, S4, 0x20, \
U4, S5, 0x20, U5, S6, 0x20, U6, S7, 0x20, U7, S8, 0x20
GINTERLACE v, d, D0, D4, U1, U0
GINTERLACE v, d, D1, D5, U3, U2
GINTERLACE v, d, D2, D6, U5, U4
GINTERLACE v, d, D3, D7, U7, U6
GST v, , D0, TD, 0x00, D1, TD, 0x10, D2, TD, 0x20, D3, TD, 0x30, \
D4, TD, 0x40, D5, TD, 0x50, D6, TD, 0x60, D7, TD, 0x70
addi.d TD, TD, 0x80
GLD v, , U0, S1, 0x30, U1, S2, 0x30, U2, S3, 0x30, U3, S4, 0x30, \
U4, S5, 0x30, U5, S6, 0x30, U6, S7, 0x30, U7, S8, 0x30
GINTERLACE v, d, D0, D4, U1, U0
GINTERLACE v, d, D1, D5, U3, U2
GINTERLACE v, d, D2, D6, U5, U4
GINTERLACE v, d, D3, D7, U7, U6
GST v, , D0, TD, 0x00, D1, TD, 0x10, D2, TD, 0x20, D3, TD, 0x30, \
D4, TD, 0x40, D5, TD, 0x50, D6, TD, 0x60, D7, TD, 0x70
addi.d TD, TD, 0x80
addi.d S1, S1, 0x40
addi.d S2, S2, 0x40
addi.d S3, S3, 0x40
addi.d S4, S4, 0x40
addi.d S5, S5, 0x40
addi.d S6, S6, 0x40
addi.d S7, S7, 0x40
addi.d S8, S8, 0x40
addi.d I, I, -1
blt ZERO, I, .L_I1
.L_I7:
andi I, M, 0x07
beq I, ZERO, .L_I0
.L_II1: /* I-- */
fld.d F0, S1, 0x00
fld.d F1, S2, 0x00
fld.d F2, S3, 0x00
fld.d F3, S4, 0x00
fld.d F4, S5, 0x00
fld.d F5, S6, 0x00
fld.d F6, S7, 0x00
fld.d F7, S8, 0x00
fst.d F0, TD, 0x00
addi.d S1, S1, 0x08
fst.d F1, TD, 0x08
addi.d S2, S2, 0x08
fst.d F2, TD, 0x10
addi.d S3, S3, 0x08
fst.d F3, TD, 0x18
addi.d S4, S4, 0x08
fst.d F4, TD, 0x20
addi.d S5, S5, 0x08
fst.d F5, TD, 0x28
addi.d S6, S6, 0x08
fst.d F6, TD, 0x30
addi.d S7, S7, 0x08
fst.d F7, TD, 0x38
addi.d S8, S8, 0x08
addi.d TD, TD, 0x40
addi.d I, I, -1
blt ZERO, I, .L_II1
.L_I0:
blt ZERO, J, .L_J1
.L_N4:
andi J, N, 0x04
beq ZERO, J, .L_N2
move S1, TS
add.d S2, TS, TL
srai.d I, M, 0x02
add.d S3, S2, TL
add.d S4, S2, T0
add.d TS, S3, T0
beq I, ZERO, .L_I3
.L_4I1: /* I-- */
GLD v, , U0, S1, 0x00, U1, S2, 0x00, U2, S3, 0x00, U3, S4, 0x00
GINTERLACE v, d, D0, D2, U1, U0
GINTERLACE v, d, D1, D3, U3, U2
GST v, , D0, TD, 0x00, D1, TD, 0x10, D2, TD, 0x20, D3, TD, 0x30
addi.d TD, TD, 0x40
GLD v, , U0, S1, 0x10, U1, S2, 0x10, U2, S3, 0x10, U3, S4, 0x10
GINTERLACE v, d, D0, D2, U1, U0
GINTERLACE v, d, D1, D3, U3, U2
GST v, , D0, TD, 0x00, D1, TD, 0x10, D2, TD, 0x20, D3, TD, 0x30
addi.d S1, S1, 0x20
addi.d S2, S2, 0x20
addi.d S3, S3, 0x20
addi.d S4, S4, 0x20
addi.d TD, TD, 0x40
addi.d I, I, -1
blt ZERO, I, .L_4I1
.L_I3:
andi I, M, 0x03
beq I, ZERO, .L_N2
.L_4II1:
fld.d F0, S1, 0x00
fld.d F1, S2, 0x00
fld.d F2, S3, 0x00
fld.d F3, S4, 0x00
fst.d F0, TD, 0x00
addi.d S1, S1, 0x08
fst.d F1, TD, 0x08
addi.d S2, S2, 0x08
fst.d F2, TD, 0x10
addi.d S3, S3, 0x08
fst.d F3, TD, 0x18
addi.d S4, S4, 0x08
addi.d TD, TD, 0x20
addi.d I, I, -1
blt ZERO, I, .L_4II1
.L_N2:
andi J, N, 0x02
beq ZERO, J, .L_N1
move S1, TS
add.d S2, TS, TL
srai.d I, M, 0x01
add.d TS, S2, TL
beq I, ZERO, .L_NI1
.L_2I1: /* I-- */
GLD v, , U0, S1, 0x00, U1, S2, 0x00
GINTERLACE v, d, D0, D1, U1, U0
GST v, , D0, TD, 0x00, D1, TD, 0x10
addi.d S1, S1, 0x10
addi.d S2, S2, 0x10
addi.d TD, TD, 0x20
addi.d I, I, -1
blt ZERO, I, .L_2I1
.L_NI1:
andi I, M, 0x01
beq I, ZERO, .L_N1
fld.d F0, S1, 0x00
fld.d F1, S2, 0x00
fst.d F0, TD, 0x00
addi.d S1, S1, 0x08
fst.d F1, TD, 0x08
addi.d S2, S2, 0x08
addi.d TD, TD, 0x10
.L_N1:
andi J, N, 0x01
beq ZERO, J, .L_N0
move S1, TS
beq ZERO, M, .L_N0
.L_M1:
fld.d F0, S1, 0x00
addi.d S1, S1, 0x08
fst.d F0, TD, 0x00
addi.d TD, TD, 0x08
addi.d M, M, -1
blt ZERO, M, .L_M1
.L_N0:
pop_if_used 0, 0
jirl $r0, $r1, 0x00
EPILOGUE