[FFmpeg-devel] [PATCH 06/10] lavc/vp8dsp: R-V V put_epel v

[uk7b at foxmail.com]

From: sunyuechi <sunyuechi at iscas.ac.cn>

C908:
vp8_put_epel4_v4_c: 11.0
vp8_put_epel4_v4_rvv_i32: 5.0
vp8_put_epel4_v6_c: 16.5
vp8_put_epel4_v6_rvv_i32: 6.2
vp8_put_epel8_v4_c: 43.7
vp8_put_epel8_v4_rvv_i32: 11.2
vp8_put_epel8_v6_c: 68.7
vp8_put_epel8_v6_rvv_i32: 13.2
vp8_put_epel16_v4_c: 92.5
vp8_put_epel16_v4_rvv_i32: 13.7
vp8_put_epel16_v6_c: 135.7
vp8_put_epel16_v6_rvv_i32: 16.5
---
 libavcodec/riscv/vp8dsp_init.c |  7 ++++++
 libavcodec/riscv/vp8dsp_rvv.S  | 46 +++++++++++++++++++++++-----------
 2 files changed, 39 insertions(+), 14 deletions(-)

diff --git a/libavcodec/riscv/vp8dsp_init.c b/libavcodec/riscv/vp8dsp_init.c
index 6614d661f7..2f123b67fe 100644
--- a/libavcodec/riscv/vp8dsp_init.c
+++ b/libavcodec/riscv/vp8dsp_init.c
@@ -85,6 +85,13 @@ av_cold void ff_vp78dsp_init_riscv(VP8DSPContext *c)
         c->put_vp8_epel_pixels_tab[0][0][1] = ff_put_vp8_epel16_h4_rvv;
         c->put_vp8_epel_pixels_tab[1][0][1] = ff_put_vp8_epel8_h4_rvv;
         c->put_vp8_epel_pixels_tab[2][0][1] = ff_put_vp8_epel4_h4_rvv;
+
+        c->put_vp8_epel_pixels_tab[0][2][0] = ff_put_vp8_epel16_v6_rvv;
+        c->put_vp8_epel_pixels_tab[1][2][0] = ff_put_vp8_epel8_v6_rvv;
+        c->put_vp8_epel_pixels_tab[2][2][0] = ff_put_vp8_epel4_v6_rvv;
+        c->put_vp8_epel_pixels_tab[0][1][0] = ff_put_vp8_epel16_v4_rvv;
+        c->put_vp8_epel_pixels_tab[1][1][0] = ff_put_vp8_epel8_v4_rvv;
+        c->put_vp8_epel_pixels_tab[2][1][0] = ff_put_vp8_epel4_v4_rvv;
     }
 #endif
 }
diff --git a/libavcodec/riscv/vp8dsp_rvv.S b/libavcodec/riscv/vp8dsp_rvv.S
index 84e8ec61de..440a965ddd 100644
--- a/libavcodec/riscv/vp8dsp_rvv.S
+++ b/libavcodec/riscv/vp8dsp_rvv.S
@@ -234,9 +234,13 @@ const subpel_filters
         .byte 0,  -1,  12, 123,  -6, 0
 endconst
 
-.macro epel_filter size
+.macro epel_filter size type
         lla             t2, subpel_filters
+.ifc \type,v
+        addi            t0, a6, -1
+.elseif \type == h
         addi            t0, a5, -1
+.endif
         li              t1, 6
         mul             t0, t0, t1
         add             t0, t0, t2
@@ -249,19 +253,25 @@ endconst
 .endif
 .endm
 
-.macro epel_load dst len size
-        addi            t6, a2, -1
-        addi            a7, a2, 1
+.macro epel_load dst len size type
+.ifc \type,v
+        mv              a5, a3
+.else
+        li              a5, 1
+.endif
+        sub             t6, a2, a5
+        add             a7, a2, a5
+.if \from_mem
         vle8.v          v24, (a2)
         vle8.v          v22, (t6)
         vle8.v          v26, (a7)
-        addi            a7, a7, 1
+        add             a7, a7, a5
         vle8.v          v28, (a7)
         vwmulu.vx       v16, v24, t2
         vwmulu.vx       v20, v26, t3
 .ifc \size,6
-        addi            t6, t6, -1
-        addi            a7, a7, 1
+        sub             t6, t6, a5
+        add             a7, a7, a5
         vle8.v          v24, (t6)
         vle8.v          v26, (a7)
         vwmaccu.vx      v16, t0, v24
@@ -293,13 +303,13 @@ endconst
         vnclipu.wi      \dst, v24, 0
 .endm
 
-.macro epel_load_inc dst len size
-        epel_load       \dst \len \size
+.macro epel_load_inc dst len size type
+        epel_load       \dst \len \size \type
         add             a2, a2, a3
 .endm
 
-.macro epel len size
-        epel_filter     \size
+.macro epel len size type
+        epel_filter     \size \type
 
 .ifc \len,4
         vsetivli        zero, 4, e8, mf4, ta, ma
@@ -311,7 +321,7 @@ endconst
 
 1:
         addi            a4, a4, -1
-        epel_load_inc   v30 \len \size
+        epel_load_inc   v30 \len \size \type
         vse8.v          v30, (a0)
         add             a0, a0, a1
         bnez            a4, 1b
@@ -321,10 +331,18 @@ endconst
 
 .irp len 16,8,4
 func ff_put_vp8_epel\len\()_h6_rvv, zve32x
-        epel \len 6
+        epel \len 6 h
 endfunc
 
 func ff_put_vp8_epel\len\()_h4_rvv, zve32x
-        epel \len 4
+        epel \len 4 h
+endfunc
+
+func ff_put_vp8_epel\len\()_v6_rvv, zve32x
+        epel \len 6 v
+endfunc
+
+func ff_put_vp8_epel\len\()_v4_rvv, zve32x
+        epel \len 4 v
 endfunc
 .endr
-- 
2.45.0