1
0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-11-21 10:55:51 +02:00
FFmpeg/libswscale/riscv/rgb2rgb_rvv.S
Rémi Denis-Courmont c2b38619c0 swscale/rgb2rgb2: rework RISC-V V shuffle_bytes_{1230,3012}
This avoids strided loads.

Before:
shuffle_bytes_1230_rvv_i32: 308.7
shuffle_bytes_3012_rvv_i32: 308.7

After:
shuffle_bytes_1230_rvv_i32: 46.7
shuffle_bytes_3012_rvv_i32: 46.7
2023-07-21 22:18:02 +03:00

180 lines
4.3 KiB
ArmAsm

/*
* Copyright © 2022 Rémi Denis-Courmont.
*
* This file is part of FFmpeg.
*
* FFmpeg is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* FFmpeg is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include "libavutil/riscv/asm.S"
func ff_shuffle_bytes_0321_rvv, zve32x
li t1, 0x00ff00ff
j 1f
endfunc
func ff_shuffle_bytes_2103_rvv, zve32x
li t1, ~0x00ff00ff
1:
not t2, t1
srai a2, a2, 2
2:
vsetvli t0, a2, e32, m8, ta, ma
vle32.v v8, (a0)
sub a2, a2, t0
vand.vx v16, v8, t2
sh2add a0, t0, a0
vand.vx v8, v8, t1
vsrl.vi v24, v16, 16
vsll.vi v16, v16, 16
vor.vv v8, v8, v24
vor.vv v8, v16, v8
vse32.v v8, (a1)
sh2add a1, t0, a1
bnez a2, 2b
ret
endfunc
func ff_shuffle_bytes_1230_rvv, zve32x
li t1, 24
li t2, 8
j 3f
endfunc
func ff_shuffle_bytes_3012_rvv, zve32x
li t1, 8
li t2, 24
3:
srai a2, a2, 2
4:
vsetvli t0, a2, e32, m8, ta, ma
vle32.v v8, (a0)
sub a2, a2, t0
vsll.vx v16, v8, t1
sh2add a0, t0, a0
vsrl.vx v8, v8, t2
vor.vv v16, v16, v8
vse32.v v16, (a1)
sh2add a1, t0, a1
bnez a2, 4b
ret
endfunc
func ff_shuffle_bytes_3210_rvv, zve32x
addi t1, a0, 2
addi t2, a0, 1
addi t3, a0, 0
addi a0, a0, 3
srai a2, a2, 2
li t4, 4
1:
vsetvli t0, a2, e8, m1, ta, ma
sub a2, a2, t0
vlse8.v v8, (a0), t4
sh2add a0, t0, a0
vlse8.v v9, (t1), t4
sh2add t1, t0, t1
vlse8.v v10, (t2), t4
sh2add t2, t0, t2
vlse8.v v11, (t3), t4
sh2add t3, t0, t3
vsseg4e8.v v8, (a1)
sh2add a1, t0, a1
bnez a2, 1b
ret
endfunc
func ff_interleave_bytes_rvv, zve32x
1:
mv t0, a0
mv t1, a1
mv t2, a2
mv t3, a3
addi a4, a4, -1
2:
vsetvli t4, t3, e8, m1, ta, ma
sub t3, t3, t4
vle8.v v8, (t0)
add t0, t4, t0
vle8.v v9, (t1)
add t1, t4, t1
vsseg2e8.v v8, (t2)
sh1add t2, t4, t2
bnez t4, 2b
add a0, a0, a5
add a1, a1, a6
add a2, a2, a7
bnez a4, 1b
ret
endfunc
#if (__riscv_xlen == 64)
.macro yuy2_to_i422p v_y0, v_y1, v_u, v_v
addi sp, sp, -16
sd s0, (sp)
sd s1, 8(sp)
addi a4, a4, 1
lw s0, 16(sp)
srai a4, a4, 1 // pixel width -> chroma width
li s1, 2
1:
mv t4, a4
mv t3, a3
mv t0, a0
addi t6, a0, 1
mv t1, a1
mv t2, a2
addi a5, a5, -1
2:
vsetvli t5, t4, e8, m1, ta, ma
sub t4, t4, t5
vlseg4e8.v v8, (t3)
sh2add t3, t5, t3
vsse8.v \v_y0, (t0), s1
sh1add t0, t5, t0
vsse8.v \v_y1, (t6), s1
sh1add t6, t5, t6
vse8.v \v_u, (t1)
add t1, t5, t1
vse8.v \v_v, (t2)
add t2, t5, t2
bnez t4, 2b
add a3, a3, s0
add a0, a0, a6
add a1, a1, a7
add a2, a2, a7
bnez a5, 1b
ld s1, 8(sp)
ld s0, (sp)
addi sp, sp, 16
ret
.endm
func ff_uyvytoyuv422_rvv, zve32x
yuy2_to_i422p v9, v11, v8, v10
endfunc
func ff_yuyvtoyuv422_rvv, zve32x
yuy2_to_i422p v8, v10, v9, v11
endfunc
#endif