diff mbox series

[FFmpeg-devel,1/7] lavc/vp9dsp: R-V mc copy_avg

Message ID CAEa-L+sr-7vAHBVr5_A_q6-fKT3p=TUbdijyJhjVUwQ=sVmc-A@mail.gmail.com
State New
Headers show
Series [FFmpeg-devel,1/7] lavc/vp9dsp: R-V mc copy_avg | expand

Checks

Context Check Description
andriy/configure_x86 warning Failed to apply patch
yinshiyou/configure_loongarch64 warning Failed to apply patch

Commit Message

flow gg March 22, 2024, 6:03 a.m. UTC
(This should be used after applying these patches)

```
[FFmpeg-devel] [PATCH 1/4] lavc/vp9dsp: R-V V ipred dc
1-4
```

Comments

flow gg March 22, 2024, 6:12 a.m. UTC | #1
It might be a bit inconvenient to find the patches related to vp8, vp9 that
were sent earlier. Here, I've placed them in a zip file in this reply

flow gg <hlefthleft@gmail.com> 于2024年3月22日周五 14:03写道:

> (This should be used after applying these patches)
>
> ```
> [FFmpeg-devel] [PATCH 1/4] lavc/vp9dsp: R-V V ipred dc
> 1-4
> ```
>
Rémi Denis-Courmont March 26, 2024, 6:30 p.m. UTC | #2
Hi,

Le perjantaina 22. maaliskuuta 2024, 8.12.41 EET flow gg a écrit :
> It might be a bit inconvenient to find the patches related to vp8, vp9 that
> were sent earlier. Here, I've placed them in a zip file in this reply

ZIP files are not particularly convenient. The only easy way for me to review 
is if patches are sent with git-send-email. (As for merging, the easiest is of 
course to pull them from a git repository.)

This would not be a problem if FFmpeg used a web forge, but hell will freeze 
before that.
flow gg March 27, 2024, 2:37 a.m. UTC | #3
Hi, here's the github link (https://github.com/hleft/FFmpeg/tree/vp8vp9)

Rémi Denis-Courmont <remi@remlab.net> 于2024年3月27日周三 02:30写道:

> Hi,
>
> Le perjantaina 22. maaliskuuta 2024, 8.12.41 EET flow gg a écrit :
> > It might be a bit inconvenient to find the patches related to vp8, vp9
> that
> > were sent earlier. Here, I've placed them in a zip file in this reply
>
> ZIP files are not particularly convenient. The only easy way for me to
> review
> is if patches are sent with git-send-email. (As for merging, the easiest
> is of
> course to pull them from a git repository.)
>
> This would not be a problem if FFmpeg used a web forge, but hell will
> freeze
> before that.
>
> --
> レミ・デニ-クールモン
> http://www.remlab.net/
>
>
>
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel@ffmpeg.org
> https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
> To unsubscribe, visit link above, or email
> ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".
>
diff mbox series

Patch

From ea81872215165ff859a0b5b2e003c5c678ea8ed0 Mon Sep 17 00:00:00 2001
From: sunyuechi <sunyuechi@iscas.ac.cn>
Date: Thu, 21 Mar 2024 22:01:18 +0800
Subject: [PATCH 1/7] lavc/vp9dsp: R-V mc copy_avg

vp9_avg4_8bpp_c: 1.2
vp9_avg4_8bpp_rvv_i64: 1.0
vp9_avg8_8bpp_c: 3.7
vp9_avg8_8bpp_rvv_i64: 1.5
vp9_avg16_8bpp_c: 14.7
vp9_avg16_8bpp_rvv_i64: 3.5
vp9_avg32_8bpp_c: 57.7
vp9_avg32_8bpp_rvv_i64: 10.0
vp9_avg64_8bpp_c: 229.0
vp9_avg64_8bpp_rvv_i64: 31.7
vp9_put4_8bpp_c: 0.7
vp9_put4_8bpp_rvi: 0.2
vp9_put8_8bpp_c: 2.5
vp9_put8_8bpp_rvi: 0.5
vp9_put16_8bpp_c: 16.5
vp9_put16_8bpp_rvv_i64: 1.7
vp9_put32_8bpp_c: 37.2
vp9_put32_8bpp_rvv_i64: 5.7
vp9_put64_8bpp_c: 91.2
vp9_put64_8bpp_rvv_i64: 19.7
---
 libavcodec/riscv/Makefile      |  4 ++-
 libavcodec/riscv/vp9_mc_rvi.S  | 43 +++++++++++++++++++++++
 libavcodec/riscv/vp9_mc_rvv.S  | 64 ++++++++++++++++++++++++++++++++++
 libavcodec/riscv/vp9dsp_init.c | 47 +++++++++++++++++++++++++
 4 files changed, 157 insertions(+), 1 deletion(-)
 create mode 100644 libavcodec/riscv/vp9_mc_rvi.S
 create mode 100644 libavcodec/riscv/vp9_mc_rvv.S

diff --git a/libavcodec/riscv/Makefile b/libavcodec/riscv/Makefile
index c237e60800..dce1236b84 100644
--- a/libavcodec/riscv/Makefile
+++ b/libavcodec/riscv/Makefile
@@ -61,6 +61,8 @@  RVV-OBJS-$(CONFIG_VC1DSP) += riscv/vc1dsp_rvv.o
 OBJS-$(CONFIG_VP8DSP) += riscv/vp8dsp_init.o
 RVV-OBJS-$(CONFIG_VP8DSP) += riscv/vp8dsp_rvv.o
 OBJS-$(CONFIG_VP9_DECODER) += riscv/vp9dsp_init.o
-RVV-OBJS-$(CONFIG_VP9_DECODER) += riscv/vp9_intra_rvv.o
+RV-OBJS-$(CONFIG_VP9_DECODER) += riscv/vp9_mc_rvi.o
+RVV-OBJS-$(CONFIG_VP9_DECODER) += riscv/vp9_intra_rvv.o \
+                                  riscv/vp9_mc_rvv.o
 OBJS-$(CONFIG_VORBIS_DECODER) += riscv/vorbisdsp_init.o
 RVV-OBJS-$(CONFIG_VORBIS_DECODER) += riscv/vorbisdsp_rvv.o
diff --git a/libavcodec/riscv/vp9_mc_rvi.S b/libavcodec/riscv/vp9_mc_rvi.S
new file mode 100644
index 0000000000..03d8dbbbae
--- /dev/null
+++ b/libavcodec/riscv/vp9_mc_rvi.S
@@ -0,0 +1,43 @@ 
+/*
+ * 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"
+
+func ff_copy8_rvi
+1:
+        addi         a4, a4, -1
+        ld           t4, (a2)
+        sd           t4, (a0)
+        add          a2, a2, a3
+        add          a0, a0, a1
+        bnez         a4, 1b
+        ret
+endfunc
+
+func ff_copy4_rvi
+1:
+        addi         a4, a4, -1
+        lw           t4, (a2)
+        sw           t4, (a0)
+        add          a2, a2, a3
+        add          a0, a0, a1
+        bnez         a4, 1b
+        ret
+endfunc
diff --git a/libavcodec/riscv/vp9_mc_rvv.S b/libavcodec/riscv/vp9_mc_rvv.S
new file mode 100644
index 0000000000..ba9ec3431f
--- /dev/null
+++ b/libavcodec/riscv/vp9_mc_rvv.S
@@ -0,0 +1,64 @@ 
+/*
+ * 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 copy_avg len type
+.ifc \type,avg
+        csrwi           vxrm, 0
+.endif
+.ifc \len,64
+        li              t5, 64
+        vsetvli         t0, t5, e8, m4, ta, ma
+.elseif \len == 32
+        li              t5, 32
+        vsetvli         t0, t5, e8, m2, ta, ma
+.elseif \len == 16
+        vsetivli        t0, 16, e8, m1, ta, ma
+.elseif \len == 8
+        vsetivli        t0, 8, e8, mf2, ta, ma
+.elseif \len == 4
+        vsetivli        t0, 4, e8, mf4, ta, ma
+.endif
+1:
+        addi            a4, a4, -1
+        vle8.v          v8, (a2)
+.ifc \type,avg
+        vle8.v          v16, (a0)
+        vaaddu.vv       v8, v8, v16
+.endif
+        vse8.v          v8, (a0)
+        add             a2, a2, a3
+        add             a0, a0, a1
+        bnez            a4, 1b
+        ret
+.endm
+
+.irp len 64, 32, 16
+func ff_copy\len\()_rvv, zve32x
+        copy_avg \len copy
+endfunc
+.endr
+
+.irp len 64, 32, 16, 8, 4
+func ff_avg\len\()_rvv, zve32x
+        copy_avg \len avg
+endfunc
+.endr
diff --git a/libavcodec/riscv/vp9dsp_init.c b/libavcodec/riscv/vp9dsp_init.c
index f08c8f6a42..c602c38bb2 100644
--- a/libavcodec/riscv/vp9dsp_init.c
+++ b/libavcodec/riscv/vp9dsp_init.c
@@ -65,7 +65,54 @@  static av_cold void vp9dsp_intrapred_init_rvv(VP9DSPContext *dsp, int bpp)
     #endif
 }
 
+static av_cold void vp9dsp_mc_init_rvv(VP9DSPContext *dsp, int bpp)
+{
+#if HAVE_RV
+    int flags = av_get_cpu_flags();
+
+    if (flags & AV_CPU_FLAG_RVI) {
+        dsp->mc[3][FILTER_8TAP_SMOOTH][0][0][0] = ff_copy8_rvi;
+        dsp->mc[3][FILTER_8TAP_REGULAR][0][0][0] = ff_copy8_rvi;
+        dsp->mc[3][FILTER_8TAP_SHARP][0][0][0] = ff_copy8_rvi;
+        dsp->mc[3][FILTER_BILINEAR][0][0][0] = ff_copy8_rvi;
+        dsp->mc[4][FILTER_8TAP_SMOOTH][0][0][0] = ff_copy4_rvi;
+        dsp->mc[4][FILTER_8TAP_REGULAR][0][0][0] = ff_copy4_rvi;
+        dsp->mc[4][FILTER_8TAP_SHARP][0][0][0] = ff_copy4_rvi;
+        dsp->mc[4][FILTER_BILINEAR][0][0][0] = ff_copy4_rvi;
+    }
+
+#if HAVE_RVV
+    if (bpp == 8 && flags & AV_CPU_FLAG_RVV_I64 && ff_get_rv_vlenb() >= 16) {
+
+#define init_fpel(idx1, idx2, sz, type)                                    \
+    dsp->mc[idx1][FILTER_8TAP_SMOOTH ][idx2][0][0] = ff_##type##sz##_rvv;  \
+    dsp->mc[idx1][FILTER_8TAP_REGULAR][idx2][0][0] = ff_##type##sz##_rvv;  \
+    dsp->mc[idx1][FILTER_8TAP_SHARP  ][idx2][0][0] = ff_##type##sz##_rvv;  \
+    dsp->mc[idx1][FILTER_BILINEAR    ][idx2][0][0] = ff_##type##sz##_rvv
+
+#define init_copy_avg(idx, sz)    \
+    init_fpel(idx, 0, sz, copy);  \
+    init_fpel(idx, 1, sz, avg)
+
+#define init_avg(idx, sz)  \
+    init_fpel(idx, 1, sz, avg)
+
+    init_copy_avg(0, 64);
+    init_copy_avg(1, 32);
+    init_copy_avg(2, 16);
+    init_avg(3, 8);
+    init_avg(4, 4);
+
+#undef init_copy_avg
+#undef init_avg
+#undef init_fpel
+    }
+#endif
+#endif
+}
+
 av_cold void ff_vp9dsp_init_riscv(VP9DSPContext *dsp, int bpp, int bitexact)
 {
     vp9dsp_intrapred_init_rvv(dsp, bpp);
+    vp9dsp_mc_init_rvv(dsp, bpp);
 }
-- 
2.44.0