ffmpeg/libavcodec/riscv/bswapdsp_rvb.S
Rémi Denis-Courmont f0ef11ea83 lavc/bswapdsp: RISC-V B bswap_buf
Simply taking the Zbb REV8 instruction into use in a simple loop gives
some significant savings:

bswap_buf_c: 1081.0
bswap_buf_rvb_b: 771.0

But we can also use the 64-bit REV8 as a pseudo-SIMD instruction with
just one additional shift, and one fewer load, effectively doubling the
bandwidth. Consequently, this patch is useful even if the compile-time
target has Zbb enabled for C code:

bswap_buf_c: 1081.0
bswap_buf_rvb_b: 341.0  (this patch)

On the other hand, this approach fails miserably for bswap16_buf as the
ratio of shifts and stores becomes unfavorable compared to naïve C:

bswap16_buf_c: 1542.0
bswap16_buf_rvb_b: 1803.7

Unrolling to process 128 bits (4 samples) at a time actually worsens
performance ever so slightly:

bswap_buf_c: 1081.0
bswap_buf_rvb_b: 408.5
2022-10-05 08:26:19 +02:00

69 lines
2.0 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 "config.h"
#include "libavutil/riscv/asm.S"
#if (__riscv_xlen >= 64)
func ff_bswap32_buf_rvb, zbb
andi t0, a1, 4
beqz t0, 1f
/* Align a1 (input) to 64-bit */
lwu t0, (a1)
addi a0, a0, 4
rev8 t0, t0
addi a2, a2, -1
srli t0, t0, __riscv_xlen - 32
addi a1, a1, 4
sw t0, -4(a0)
1:
andi a3, a2, -2
sh2add a2, a2, a0
beqz a3, 3f
sh2add a3, a3, a0
2: /* 2 elements (64 bits) at a time on a 64-bit boundary */
ld t0, (a1)
addi a0, a0, 8
rev8 t0, t0
#if (__riscv_xlen == 64)
srli t2, t0, 32
sw t0, -4(a0)
#else
srli t1, t0, __riscv_xlen - 64
srli t2, t0, __riscv_xlen - 32
sw t1, -4(a0)
#endif
addi a1, a1, 8
sw t2, -8(a0)
bne a0, a3, 2b
3:
beq a0, a2, 5f
4: /* Process last element */
lwu t0, (a1)
addi a0, a0, 4
rev8 t0, t0
addi a1, a1, 4
srli t0, t0, __riscv_xlen - 32
sw t0, -4(a0)
5:
ret
endfunc
#endif