[FFmpeg-devel] [PATCH 1/4] lavc/riscv: Move VVC macro to h26x
flow gg
hlefthleft at gmail.com
Sun Nov 17 15:17:49 EET 2024
> Generally speaking, I think that moving code should be done in dedicated
> patches.
> You can branch here. The rest of the byte code is the same in all but one
> cases.
Updated this.
<uk7b at foxmail.com> 于2024年11月17日周日 21:17写道：
> From: sunyuechi <sunyuechi at iscas.ac.cn>
>
> ---
> libavcodec/riscv/h26x/asm.S | 127 ++++++++++++++++++++++++++++++
> libavcodec/riscv/vvc/vvc_mc_rvv.S | 117 ++-------------------------
> 2 files changed, 132 insertions(+), 112 deletions(-)
> create mode 100644 libavcodec/riscv/h26x/asm.S
>
> diff --git a/libavcodec/riscv/h26x/asm.S b/libavcodec/riscv/h26x/asm.S
> new file mode 100644
> index 0000000000..1b8453d825
> --- /dev/null
> +++ b/libavcodec/riscv/h26x/asm.S
> @@ -0,0 +1,127 @@
> +/*
> + * Copyright (c) 2024 Institue of Software Chinese Academy of Sciences
> (ISCAS).
> + *
> + * 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"
> +
> +.macro vsetvlstatic8 w, vlen
> + .if \w == 2 && \vlen == 128
> + vsetivli zero, \w, e8, mf8, ta, ma
> + .elseif \w <= 4 && \vlen == 128
> + vsetivli zero, \w, e8, mf4, ta, ma
> + .elseif \w <= 8 && \vlen == 128
> + vsetivli zero, \w, e8, mf2, ta, ma
> + .elseif \w <= 16 && \vlen == 128
> + vsetivli zero, \w, e8, m1, ta, ma
> + .elseif \w <= 32 && \vlen == 128
> + li t0, \w
> + vsetvli zero, t0, e8, m2, ta, ma
> + .elseif \w <= 4 && \vlen == 256
> + vsetivli zero, \w, e8, mf8, ta, ma
> + .elseif \w <= 8 && \vlen == 256
> + vsetivli zero, \w, e8, mf4, ta, ma
> + .elseif \w <= 16 && \vlen == 256
> + vsetivli zero, \w, e8, mf2, ta, ma
> + .elseif \w <= 32 && \vlen == 256
> + li t0, \w
> + vsetvli zero, t0, e8, m1, ta, ma
> + .elseif \w <= 64 && \vlen == 256
> + li t0, \w
> + vsetvli zero, t0, e8, m2, ta, ma
> + .else
> + li t0, \w
> + vsetvli zero, t0, e8, m4, ta, ma
> + .endif
> +.endm
> +
> +.macro vsetvlstatic16 w, vlen
> + .if \w == 2 && \vlen == 128
> + vsetivli zero, \w, e16, mf4, ta, ma
> + .elseif \w <= 4 && \vlen == 128
> + vsetivli zero, \w, e16, mf2, ta, ma
> + .elseif \w <= 8 && \vlen == 128
> + vsetivli zero, \w, e16, m1, ta, ma
> + .elseif \w <= 16 && \vlen == 128
> + vsetivli zero, \w, e16, m2, ta, ma
> + .elseif \w <= 32 && \vlen == 128
> + li t0, \w
> + vsetvli zero, t0, e16, m4, ta, ma
> + .elseif \w <= 4 && \vlen == 256
> + vsetivli zero, \w, e16, mf4, ta, ma
> + .elseif \w <= 8 && \vlen == 256
> + vsetivli zero, \w, e16, mf2, ta, ma
> + .elseif \w <= 16 && \vlen == 256
> + vsetivli zero, \w, e16, m1, ta, ma
> + .elseif \w <= 32 && \vlen == 256
> + li t0, \w
> + vsetvli zero, t0, e16, m2, ta, ma
> + .elseif \w <= 64 && \vlen == 256
> + li t0, \w
> + vsetvli zero, t0, e16, m4, ta, ma
> + .else
> + li t0, \w
> + vsetvli zero, t0, e16, m8, ta, ma
> + .endif
> +.endm
> +
> +.macro vsetvlstatic32 w, vlen
> + .if \w == 2
> + vsetivli zero, \w, e32, mf2, ta, ma
> + .elseif \w <= 4 && \vlen == 128
> + vsetivli zero, \w, e32, m1, ta, ma
> + .elseif \w <= 8 && \vlen == 128
> + vsetivli zero, \w, e32, m2, ta, ma
> + .elseif \w <= 16 && \vlen == 128
> + vsetivli zero, \w, e32, m4, ta, ma
> + .elseif \w <= 4 && \vlen == 256
> + vsetivli zero, \w, e32, mf2, ta, ma
> + .elseif \w <= 8 && \vlen == 256
> + vsetivli zero, \w, e32, m1, ta, ma
> + .elseif \w <= 16 && \vlen == 256
> + vsetivli zero, \w, e32, m2, ta, ma
> + .elseif \w <= 32 && \vlen == 256
> + li t0, \w
> + vsetvli zero, t0, e32, m4, ta, ma
> + .else
> + li t0, \w
> + vsetvli zero, t0, e32, m8, ta, ma
> + .endif
> +.endm
> +
> +.macro POW2_JMP_TABLE id, vlen
> +const jmp_table_\id\vlen
> + .4byte \id\()2\vlen\()f - jmp_table_\id\vlen
> + .4byte \id\()4\vlen\()f - jmp_table_\id\vlen
> + .4byte \id\()8\vlen\()f - jmp_table_\id\vlen
> + .4byte \id\()16\vlen\()f - jmp_table_\id\vlen
> + .4byte \id\()32\vlen\()f - jmp_table_\id\vlen
> + .4byte \id\()64\vlen\()f - jmp_table_\id\vlen
> + .4byte \id\()128\vlen\()f - jmp_table_\id\vlen
> +endconst
> +.endm
> +
> +.macro POW2_J vlen, id, w
> + clz t1, \w
> + neg t1, t1
> + lla t5, jmp_table_\id\vlen
> + sh2add t1, t1, t5
> + lw t1, ((__riscv_xlen-2)<<2)(t1)
> + add t1, t1, t5
> + jr t1
> +.endm
> diff --git a/libavcodec/riscv/vvc/vvc_mc_rvv.S
> b/libavcodec/riscv/vvc/vvc_mc_rvv.S
> index 1dcbaf7d5b..ecdd4843ff 100644
> --- a/libavcodec/riscv/vvc/vvc_mc_rvv.S
> +++ b/libavcodec/riscv/vvc/vvc_mc_rvv.S
> @@ -18,91 +18,7 @@
> * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
> 02110-1301 USA
> */
>
> -#include "libavutil/riscv/asm.S"
> -
> -.macro vsetvlstatic8 w, vlen
> - .if \w == 2 && \vlen == 128
> - vsetivli zero, \w, e8, mf8, ta, ma
> - .elseif \w <= 4 && \vlen == 128
> - vsetivli zero, \w, e8, mf4, ta, ma
> - .elseif \w <= 8 && \vlen == 128
> - vsetivli zero, \w, e8, mf2, ta, ma
> - .elseif \w <= 16 && \vlen == 128
> - vsetivli zero, \w, e8, m1, ta, ma
> - .elseif \w <= 32 && \vlen == 128
> - li t0, \w
> - vsetvli zero, t0, e8, m2, ta, ma
> - .elseif \w <= 4 && \vlen == 256
> - vsetivli zero, \w, e8, mf8, ta, ma
> - .elseif \w <= 8 && \vlen == 256
> - vsetivli zero, \w, e8, mf4, ta, ma
> - .elseif \w <= 16 && \vlen == 256
> - vsetivli zero, \w, e8, mf2, ta, ma
> - .elseif \w <= 32 && \vlen == 256
> - li t0, \w
> - vsetvli zero, t0, e8, m1, ta, ma
> - .elseif \w <= 64 && \vlen == 256
> - li t0, \w
> - vsetvli zero, t0, e8, m2, ta, ma
> - .else
> - li t0, \w
> - vsetvli zero, t0, e8, m4, ta, ma
> - .endif
> -.endm
> -
> -.macro vsetvlstatic16 w, vlen
> - .if \w == 2 && \vlen == 128
> - vsetivli zero, \w, e16, mf4, ta, ma
> - .elseif \w <= 4 && \vlen == 128
> - vsetivli zero, \w, e16, mf2, ta, ma
> - .elseif \w <= 8 && \vlen == 128
> - vsetivli zero, \w, e16, m1, ta, ma
> - .elseif \w <= 16 && \vlen == 128
> - vsetivli zero, \w, e16, m2, ta, ma
> - .elseif \w <= 32 && \vlen == 128
> - li t0, \w
> - vsetvli zero, t0, e16, m4, ta, ma
> - .elseif \w <= 4 && \vlen == 256
> - vsetivli zero, \w, e16, mf4, ta, ma
> - .elseif \w <= 8 && \vlen == 256
> - vsetivli zero, \w, e16, mf2, ta, ma
> - .elseif \w <= 16 && \vlen == 256
> - vsetivli zero, \w, e16, m1, ta, ma
> - .elseif \w <= 32 && \vlen == 256
> - li t0, \w
> - vsetvli zero, t0, e16, m2, ta, ma
> - .elseif \w <= 64 && \vlen == 256
> - li t0, \w
> - vsetvli zero, t0, e16, m4, ta, ma
> - .else
> - li t0, \w
> - vsetvli zero, t0, e16, m8, ta, ma
> - .endif
> -.endm
> -
> -.macro vsetvlstatic32 w, vlen
> - .if \w == 2
> - vsetivli zero, \w, e32, mf2, ta, ma
> - .elseif \w <= 4 && \vlen == 128
> - vsetivli zero, \w, e32, m1, ta, ma
> - .elseif \w <= 8 && \vlen == 128
> - vsetivli zero, \w, e32, m2, ta, ma
> - .elseif \w <= 16 && \vlen == 128
> - vsetivli zero, \w, e32, m4, ta, ma
> - .elseif \w <= 4 && \vlen == 256
> - vsetivli zero, \w, e32, mf2, ta, ma
> - .elseif \w <= 8 && \vlen == 256
> - vsetivli zero, \w, e32, m1, ta, ma
> - .elseif \w <= 16 && \vlen == 256
> - vsetivli zero, \w, e32, m2, ta, ma
> - .elseif \w <= 32 && \vlen == 256
> - li t0, \w
> - vsetvli zero, t0, e32, m4, ta, ma
> - .else
> - li t0, \w
> - vsetvli zero, t0, e32, m8, ta, ma
> - .endif
> -.endm
> +#include "libavcodec/riscv/h26x/asm.S"
>
> .macro avg w, vlen, id
> \id\w\vlen:
> @@ -157,35 +73,12 @@
> ret
> .endm
>
> -
> -.macro AVG_JMP_TABLE id, vlen
> -const jmp_table_\id\vlen
> - .4byte \id\()2\vlen\()f - jmp_table_\id\vlen
> - .4byte \id\()4\vlen\()f - jmp_table_\id\vlen
> - .4byte \id\()8\vlen\()f - jmp_table_\id\vlen
> - .4byte \id\()16\vlen\()f - jmp_table_\id\vlen
> - .4byte \id\()32\vlen\()f - jmp_table_\id\vlen
> - .4byte \id\()64\vlen\()f - jmp_table_\id\vlen
> - .4byte \id\()128\vlen\()f - jmp_table_\id\vlen
> -endconst
> -.endm
> -
> -.macro AVG_J vlen, id
> - clz t1, a4
> - neg t1, t1
> - lla t5, jmp_table_\id\vlen
> - sh2add t1, t1, t5
> - lw t1, ((__riscv_xlen-2)<<2)(t1)
> - add t1, t1, t5
> - jr t1
> -.endm
> -
> .macro func_avg vlen
> func ff_vvc_avg_8_rvv_\vlen\(), zve32x, zbb, zba
> lpad 0
> - AVG_JMP_TABLE 1, \vlen
> + POW2_JMP_TABLE 1, \vlen
> csrwi vxrm, 0
> - AVG_J \vlen, 1
> + POW2_J \vlen, 1, a4
> .irp w,2,4,8,16,32,64,128
> avg \w, \vlen, 1
> .endr
> @@ -265,7 +158,7 @@ func_avg 256
> .macro func_w_avg vlen
> func ff_vvc_w_avg_8_rvv_\vlen\(), zve32x, zbb, zba
> lpad 0
> - AVG_JMP_TABLE 2, \vlen
> + POW2_JMP_TABLE 2, \vlen
> csrwi vxrm, 0
> addi t6, a6, 7
> ld t3, (sp)
> @@ -275,7 +168,7 @@ func ff_vvc_w_avg_8_rvv_\vlen\(), zve32x, zbb, zba
> add t4, t4, t5
> addi t5, t6, -1 // shift - 1
> sll t4, t4, t5
> - AVG_J \vlen, 2
> + POW2_J \vlen, 2, a4
> .irp w,2,4,8,16,32,64,128
> w_avg \w, \vlen, 2
> .endr
> --
> 2.47.0
>
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
> To unsubscribe, visit link above, or email
> ffmpeg-devel-request at ffmpeg.org with subject "unsubscribe".
>
More information about the ffmpeg-devel
mailing list