OpenBLAS/kernel/loongarch64/dgemm_tcopy_4_lsx.S

281 lines
7.7 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 S0 $r11
#define S1 $r12
#define S2 $r13
#define S3 $r14
#define S4 $r15
#define P0 $r16
#define P1 $r17
#define P2 $r18
#define P3 $r19
#define T0 $r20
#define T1 $r23
#define TL $r7
#define ZERO $r0
#define F0 $f0
#define F1 $f1
#define F2 $f2
#define F3 $f3
/* 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
PROLOGUE
push_if_used 1, 0
move S0, SRC
move P0, DST
// Find P0, P2, P3
srai.d T0, N, 0x02
slli.d T0, T0, 0x02
srai.d T1, N, 0x01
slli.d T1, T1, 0x01
mul.d T0, M, T0
mul.d T1, M, T1
slli.d T0, T0, 0x03
slli.d T1, T1, 0x03
add.d P2, DST, T0
add.d P3, DST, T1
slli.d TL, LDA, 0x03
srai.d J, M, 0x02
slli.d T0, TL, 0x01
slli.d T1, M, 0x05
beq ZERO, J, .L_M3
.L_J1: /* J-- */
move S1, S0
add.d S2, S0, TL
add.d S3, S1, T0
add.d S4, S2, T0
add.d S0, S3, T0
move P1, P0
addi.d P0, P0, 0x80
srai.d I, N, 0x02
addi.d J, J, -1
beq ZERO, I, .L_N3
.L_I1: /* I-- */
vld U0, S1, 0x00
vld U1, S1, 0x10
vld U2, S2, 0x00
vld U3, S2, 0x10
vld U4, S3, 0x00
vld U5, S3, 0x10
vld U6, S4, 0x00
vld U7, S4, 0x10
vst U0, P1, 0x00
vst U1, P1, 0x10
vst U2, P1, 0x20
vst U3, P1, 0x30
vst U4, P1, 0x40
vst U5, P1, 0x50
vst U6, P1, 0x60
vst U7, P1, 0x70
addi.d S1, S1, 0x20
addi.d S2, S2, 0x20
addi.d S3, S3, 0x20
addi.d S4, S4, 0x20
add.d P1, P1, T1
addi.d I, I, -1
blt ZERO, I, .L_I1
.L_N3:
andi I, N, 0x02
beq ZERO, I, .L_N1
vld U0, S1, 0x00
vld U1, S2, 0x00
vld U2, S3, 0x00
vld U3, S4, 0x00
vst U0, P2, 0x00
vst U1, P2, 0x10
vst U2, P2, 0x20
vst U3, P2, 0x30
addi.d S1, S1, 0x10
addi.d S2, S2, 0x10
addi.d S3, S3, 0x10
addi.d S4, S4, 0x10
addi.d P2, P2, 0x40
.L_N1:
andi I, N, 0x01
beq ZERO, I, .L_N0
fld.d F0, S1, 0x00
fld.d F1, S2, 0x00
fld.d F2, S3, 0x00
fld.d F3, S4, 0x00
fst.d F0, P3, 0x00
fst.d F1, P3, 0x08
fst.d F2, P3, 0x10
fst.d F3, P3, 0x18
addi.d S1, S1, 0x08
addi.d S2, S2, 0x08
addi.d S3, S3, 0x08
addi.d S4, S4, 0x08
addi.d P3, P3, 0x20
.L_N0:
blt ZERO, J, .L_J1
.L_M3:
andi J, M, 0x02
beq ZERO, J, .L_M1
move S1, S0
add.d S2, S0, TL
add.d S0, S0, T0
move P1, P0
addi.d P0, P0, 0x40
srai.d I, N, 0x02
beq ZERO, I, .L_2N3
.L_2I1: /* I-- */
vld U0, S1, 0x00
vld U1, S1, 0x10
vld U2, S2, 0x00
vld U3, S2, 0x10
vst U0, P1, 0x00
vst U1, P1, 0x10
vst U2, P1, 0x20
vst U3, P1, 0x30
addi.d S1, S1, 0x20
addi.d S2, S2, 0x20
addi.d I, I, -1
add.d P1, P1, T1
blt ZERO, I, .L_2I1
.L_2N3:
andi I, N, 0x02
beq ZERO, I, .L_2N1
vld U0, S1, 0x00
vld U1, S2, 0x00
vst U0, P2, 0x00
vst U1, P2, 0x10
addi.d S1, S1, 0x10
addi.d S2, S2, 0x10
addi.d P2, P2, 0x20
.L_2N1:
addi.d I, N, 0x01
beq ZERO, I, .L_M1
fld.d F0, S1, 0x00
fld.d F1, S2, 0x00
fst.d F0, P3, 0x00
fst.d F1, P3, 0x08
addi.d S1, S1, 0x08
addi.d S2, S2, 0x08
addi.d P3, P3, 0x10
.L_M1:
andi J, M, 0x01
beq ZERO, J, .L_M0
move S1, S0
move P1, P0
srai.d I, N, 0x02
beq ZERO, I, .L_1N3
.L_1I1:
vld U0, S1, 0x00
vld U1, S1, 0x10
vst U0, P1, 0x00
vst U1, P1, 0x10
addi.d S1, S1, 0x20
addi.d I, I, -1
add.d P1, P1, T1
blt ZERO, I, .L_1I1
.L_1N3:
andi I, N, 0x02
beq I, ZERO, .L_1N1
fld.d F0, S1, 0x00
fld.d F1, S1, 0x08
fst.d F0, P2, 0x00
fst.d F1, P2, 0x08
addi.d S1, S1, 0x10
addi.d P2, P2, 0x10
.L_1N1:
andi I, N, 0x01
beq I, ZERO, .L_M0
fld.d F0, S1, 0x00
fst.d F0, P3, 0x00
.L_M0:
pop_if_used 1, 0
jirl $r0, $r1, 0x00
EPILOGUE