diff mbox series

[FFmpeg-devel,1/4] lavc/vp9dsp: R-V V mc bilin h v

Message ID tencent_BB59F1D25C0BD67E9806CFB6AAADC5E1D006@qq.com
State New
Headers show
Series [FFmpeg-devel,1/4] lavc/vp9dsp: R-V V mc bilin h v | expand

Checks

Context Check Description
yinshiyou/make_loongarch64 success Make finished
yinshiyou/make_fate_loongarch64 success Make fate finished
andriy/make_x86 success Make finished
andriy/make_fate_x86 success Make fate finished

Commit Message

uk7b@foxmail.com Aug. 3, 2024, 9:50 a.m. UTC
From: sunyuechi <sunyuechi@iscas.ac.cn>

                                                     C908   X60
vp9_avg_bilin_4h_8bpp_c                            :    5.5    4.7
vp9_avg_bilin_4h_8bpp_rvv_i32                      :    1.7    1.5
vp9_avg_bilin_4v_8bpp_c                            :    5.5    4.7
vp9_avg_bilin_4v_8bpp_rvv_i32                      :    1.5    1.2
vp9_avg_bilin_8h_8bpp_c                            :   20.0   17.7
vp9_avg_bilin_8h_8bpp_rvv_i32                      :    3.0    2.7
vp9_avg_bilin_8v_8bpp_c                            :   20.7   18.7
vp9_avg_bilin_8v_8bpp_rvv_i32                      :    3.0    2.7
vp9_avg_bilin_16h_8bpp_c                           :   78.2   69.7
vp9_avg_bilin_16h_8bpp_rvv_i32                     :    7.0    6.2
vp9_avg_bilin_16v_8bpp_c                           :   98.5   73.2
vp9_avg_bilin_16v_8bpp_rvv_i32                     :    7.0    6.0
vp9_avg_bilin_32h_8bpp_c                           :  325.5  275.5
vp9_avg_bilin_32h_8bpp_rvv_i32                     :   23.0   20.5
vp9_avg_bilin_32v_8bpp_c                           :  342.2  290.0
vp9_avg_bilin_32v_8bpp_rvv_i32                     :   21.7   19.5
vp9_avg_bilin_64h_8bpp_c                           : 1263.7 1095.7
vp9_avg_bilin_64h_8bpp_rvv_i32                     :   91.2   81.2
vp9_avg_bilin_64v_8bpp_c                           : 1331.7 1155.2
vp9_avg_bilin_64v_8bpp_rvv_i32                     :   91.2   81.0
vp9_put_bilin_4h_8bpp_c                            :    4.5    4.0
vp9_put_bilin_4h_8bpp_rvv_i32                      :    1.0    1.0
vp9_put_bilin_4v_8bpp_c                            :    4.7    4.2
vp9_put_bilin_4v_8bpp_rvv_i32                      :    1.0    1.0
vp9_put_bilin_8h_8bpp_c                            :   16.7   15.0
vp9_put_bilin_8h_8bpp_rvv_i32                      :    2.2    2.0
vp9_put_bilin_8v_8bpp_c                            :   17.5   15.7
vp9_put_bilin_8v_8bpp_rvv_i32                      :    2.2    2.0
vp9_put_bilin_16h_8bpp_c                           :   65.2   58.0
vp9_put_bilin_16h_8bpp_rvv_i32                     :    6.0    5.5
vp9_put_bilin_16v_8bpp_c                           :   69.2   61.7
vp9_put_bilin_16v_8bpp_rvv_i32                     :    5.7    5.2
vp9_put_bilin_32h_8bpp_c                           :  273.2  229.0
vp9_put_bilin_32h_8bpp_rvv_i32                     :   19.7   17.7
vp9_put_bilin_32v_8bpp_c                           :  290.5  243.7
vp9_put_bilin_32v_8bpp_rvv_i32                     :   18.7   16.7
vp9_put_bilin_64h_8bpp_c                           : 1040.5  910.5
vp9_put_bilin_64h_8bpp_rvv_i32                     :   82.5   73.0
vp9_put_bilin_64v_8bpp_c                           : 1108.5  971.0
vp9_put_bilin_64v_8bpp_rvv_i32                     :   82.2   73.2
---
 libavcodec/riscv/vp9_mc_rvv.S  | 116 +++++++++++++++++++++++++++++++++
 libavcodec/riscv/vp9dsp.h      |  12 ++--
 libavcodec/riscv/vp9dsp_init.c |  21 ++++++
 3 files changed, 143 insertions(+), 6 deletions(-)
diff mbox series

Patch

diff --git a/libavcodec/riscv/vp9_mc_rvv.S b/libavcodec/riscv/vp9_mc_rvv.S
index 8d776661d9..817cc58b5e 100644
--- a/libavcodec/riscv/vp9_mc_rvv.S
+++ b/libavcodec/riscv/vp9_mc_rvv.S
@@ -54,6 +54,122 @@  func ff_vp9_avg\len\()_rvv, zve32x
 endfunc
 .endm
 
+.macro bilin_load_h dst, op, mn
+        addi            t5, a2, 1
+        vle8.v          v8, (a2)
+        vle8.v          v0, (t5)
+        vwmulu.vx       v16, v0, \mn
+        vwmaccsu.vx     v16, t1, v8
+        vwadd.wx        v16, v16, t4
+        vnsra.wi        v16, v16, 4
+        vadd.vv         \dst, v16, v8
+.ifc \op,avg
+        vle8.v          v16, (a0)
+        vaaddu.vv       \dst, \dst, v16
+.endif
+.endm
+
+.macro bilin_h_v op, type, mn
+func ff_\op\()_vp9_bilin_64\type\()_rvv, zve32x
+        lpad    0
+        vsetvlstatic8   64, t0, 64
+.ifc \op,avg
+        csrwi           vxrm, 0
+.endif
+        li              t4, 8
+        neg             t1, \mn
+1:
+        addi            a4, a4, -1
+.ifc \type,v
+        add             t5, a2, a3
+.else
+        addi            t5, a2, 1
+.endif
+        vle8.v          v8, (a2)
+        vle8.v          v0, (t5)
+        vwmulu.vx       v16, v0, \mn
+        vwmaccsu.vx     v16, t1, v8
+        vwadd.wx        v16, v16, t4
+        vnsra.wi        v16, v16, 4
+        vadd.vv         v0, v16, v8
+.ifc \op,avg
+        vle8.v          v16, (a0)
+        vaaddu.vv       v0, v0, v16
+.endif
+        vse8.v          v0, (a0)
+        add             a2, a2, a3
+        add             a0, a0, a1
+        bnez            a4, 1b
+        ret
+
+.Lbilin_\type\op:
+.ifc \op,avg
+        csrwi           vxrm, 0
+.endif
+        li              t4, 8
+        neg             t1, \mn
+1:
+        addi            a4, a4, -2
+        add             t6, a0, a1
+        add             t0, a2, a3
+        vle8.v          v8, (a2)
+        vle8.v          v4, (t0)
+.ifc \type,v
+        add             t2, t0, a3
+        vwmulu.vx       v16, v4, \mn
+.else
+        addi            t3, a2, 1
+        addi            t2, t0, 1
+        vle8.v          v0, (t3)
+        vwmulu.vx       v16, v0, \mn
+.endif
+        vle8.v          v12, (t2)
+        vwmulu.vx       v20, v12, \mn
+        vwmaccsu.vx     v16, t1, v8
+        vwmaccsu.vx     v20, t1, v4
+        vwadd.wx        v16, v16, t4
+        vwadd.wx        v20, v20, t4
+        vnsra.wi        v16, v16, 4
+        vnsra.wi        v20, v20, 4
+        vadd.vv         v0, v16, v8
+        vadd.vv         v12, v20, v4
+.ifc \op,avg
+        vle8.v          v16, (a0)
+        vle8.v          v20, (t6)
+        vaaddu.vv       v0, v0, v16
+        vaaddu.vv       v12, v12, v20
+.endif
+        vse8.v          v0, (a0)
+        vse8.v          v12, (t6)
+        add             a2, t0, a3
+        add             a0, t6, a1
+        bnez            a4, 1b
+
+        ret
+endfunc
+.endm
+
 .irp len, 64, 32, 16, 8, 4
         copy_avg \len
 .endr
+
+bilin_h_v  put, h, a5
+bilin_h_v  avg, h, a5
+bilin_h_v  put, v, a6
+bilin_h_v  avg, v, a6
+
+.macro func_bilin_h_v len, op, type
+func ff_\op\()_vp9_bilin_\len\()\type\()_rvv, zve32x
+        lpad    0
+        vsetvlstatic8   \len, t0, 64
+        j               .Lbilin_\type\()\op
+endfunc
+.endm
+
+.irp len, 32, 16, 8, 4
+        .irp op, put, avg
+                .irp type, h, v
+                        func_bilin_h_v \len, \op, \type
+                .endr
+        .endr
+.endr
diff --git a/libavcodec/riscv/vp9dsp.h b/libavcodec/riscv/vp9dsp.h
index ff8431591c..8fb326dae0 100644
--- a/libavcodec/riscv/vp9dsp.h
+++ b/libavcodec/riscv/vp9dsp.h
@@ -113,27 +113,27 @@  void ff_avg_8tap_##type##_##SIZE##hv_rvv(uint8_t *dst, ptrdiff_t dststride,  \
                                          int h, int mx, int my);
 
 #define VP9_BILINEAR_RISCV_RVV_FUNC(SIZE)                                   \
-void ff_put_bilin_##SIZE##h_rvv(uint8_t *dst, ptrdiff_t dststride,         \
+void ff_put_vp9_bilin_##SIZE##h_rvv(uint8_t *dst, ptrdiff_t dststride,     \
                                 const uint8_t *src, ptrdiff_t srcstride,   \
                                 int h, int mx, int my);                    \
                                                                            \
-void ff_put_bilin_##SIZE##v_rvv(uint8_t *dst, ptrdiff_t dststride,         \
+void ff_put_vp9_bilin_##SIZE##v_rvv(uint8_t *dst, ptrdiff_t dststride,     \
                                 const uint8_t *src, ptrdiff_t srcstride,   \
                                 int h, int mx, int my);                    \
                                                                            \
-void ff_put_bilin_##SIZE##hv_rvv(uint8_t *dst, ptrdiff_t dststride,        \
+void ff_put_vp9_bilin_##SIZE##hv_rvv(uint8_t *dst, ptrdiff_t dststride,    \
                                  const uint8_t *src, ptrdiff_t srcstride,  \
                                  int h, int mx, int my);                   \
                                                                            \
-void ff_avg_bilin_##SIZE##h_rvv(uint8_t *dst, ptrdiff_t dststride,         \
+void ff_avg_vp9_bilin_##SIZE##h_rvv(uint8_t *dst, ptrdiff_t dststride,     \
                                 const uint8_t *src, ptrdiff_t srcstride,   \
                                 int h, int mx, int my);                    \
                                                                            \
-void ff_avg_bilin_##SIZE##v_rvv(uint8_t *dst, ptrdiff_t dststride,         \
+void ff_avg_vp9_bilin_##SIZE##v_rvv(uint8_t *dst, ptrdiff_t dststride,     \
                                 const uint8_t *src, ptrdiff_t srcstride,   \
                                 int h, int mx, int my);                    \
                                                                            \
-void ff_avg_bilin_##SIZE##hv_rvv(uint8_t *dst, ptrdiff_t dststride,        \
+void ff_avg_vp9_bilin_##SIZE##hv_rvv(uint8_t *dst, ptrdiff_t dststride,    \
                                  const uint8_t *src, ptrdiff_t srcstride,  \
                                  int h, int mx, int my);
 
diff --git a/libavcodec/riscv/vp9dsp_init.c b/libavcodec/riscv/vp9dsp_init.c
index 2034e1c976..83dbe1b5d9 100644
--- a/libavcodec/riscv/vp9dsp_init.c
+++ b/libavcodec/riscv/vp9dsp_init.c
@@ -63,6 +63,27 @@  static av_cold void vp9dsp_mc_init_riscv(VP9DSPContext *dsp, int bpp)
     init_fpel(3, 8);
     init_fpel(4, 4);
 
+    dsp->mc[0][FILTER_BILINEAR ][0][0][1] = ff_put_vp9_bilin_64v_rvv;
+    dsp->mc[0][FILTER_BILINEAR ][0][1][0] = ff_put_vp9_bilin_64h_rvv;
+    dsp->mc[0][FILTER_BILINEAR ][1][0][1] = ff_avg_vp9_bilin_64v_rvv;
+    dsp->mc[0][FILTER_BILINEAR ][1][1][0] = ff_avg_vp9_bilin_64h_rvv;
+    dsp->mc[1][FILTER_BILINEAR ][0][0][1] = ff_put_vp9_bilin_32v_rvv;
+    dsp->mc[1][FILTER_BILINEAR ][0][1][0] = ff_put_vp9_bilin_32h_rvv;
+    dsp->mc[1][FILTER_BILINEAR ][1][0][1] = ff_avg_vp9_bilin_32v_rvv;
+    dsp->mc[1][FILTER_BILINEAR ][1][1][0] = ff_avg_vp9_bilin_32h_rvv;
+    dsp->mc[2][FILTER_BILINEAR ][0][0][1] = ff_put_vp9_bilin_16v_rvv;
+    dsp->mc[2][FILTER_BILINEAR ][0][1][0] = ff_put_vp9_bilin_16h_rvv;
+    dsp->mc[2][FILTER_BILINEAR ][1][0][1] = ff_avg_vp9_bilin_16v_rvv;
+    dsp->mc[2][FILTER_BILINEAR ][1][1][0] = ff_avg_vp9_bilin_16h_rvv;
+    dsp->mc[3][FILTER_BILINEAR ][0][0][1] = ff_put_vp9_bilin_8v_rvv;
+    dsp->mc[3][FILTER_BILINEAR ][0][1][0] = ff_put_vp9_bilin_8h_rvv;
+    dsp->mc[3][FILTER_BILINEAR ][1][0][1] = ff_avg_vp9_bilin_8v_rvv;
+    dsp->mc[3][FILTER_BILINEAR ][1][1][0] = ff_avg_vp9_bilin_8h_rvv;
+    dsp->mc[4][FILTER_BILINEAR ][0][0][1] = ff_put_vp9_bilin_4v_rvv;
+    dsp->mc[4][FILTER_BILINEAR ][0][1][0] = ff_put_vp9_bilin_4h_rvv;
+    dsp->mc[4][FILTER_BILINEAR ][1][0][1] = ff_avg_vp9_bilin_4v_rvv;
+    dsp->mc[4][FILTER_BILINEAR ][1][1][0] = ff_avg_vp9_bilin_4h_rvv;
+
 #undef init_fpel
     }
 #endif