lavc/vp8dsp: add R-V V vp7_idct_dc_add4y

As with idct_dc_add, most of the code is shared with, and replaces, the
previous VP8 function. To improve performance, we break down the 16x4
matrix into 4 rows, rather than 4 squares. Thus strided loads and
stores are avoided, and the 4 DC calculations are vectored.
Unfortunately this requires a vector gather to splat the DC values, but
overall this is still a win for performance:

T-Head C908:
vp7_idct_dc_add4y_c:       7.2
vp7_idct_dc_add4y_rvv_i32: 2.2
vp8_idct_dc_add4y_c:       6.2
vp8_idct_dc_add4y_rvv_i32: 2.2 (before)
vp8_idct_dc_add4y_rvv_i32: 1.7

SpacemiT X60:
vp7_idct_dc_add4y_c:       6.2
vp7_idct_dc_add4y_rvv_i32: 2.0
vp8_idct_dc_add4y_c:       5.5
vp8_idct_dc_add4y_rvv_i32: 2.5 (before)
vp8_idct_dc_add4y_rvv_i32: 1.7

I also tried to provision the DC values using indexed loads. It ends up
slower overall, especially for VP7, as we then have to compute 16 DC's
instead of just 4.

libavcodec/riscv/vp7dsp_init.c

@@ -28,6 +28,7 @@
 void ff_vp7_luma_dc_wht_rvv(int16_t block[4][4][16], int16_t dc[16]);
 void ff_vp7_idct_add_rvv(uint8_t *dst, int16_t block[16], ptrdiff_t stride);
 void ff_vp78_idct_dc_add_rvv(uint8_t *, int16_t block[16], ptrdiff_t, int dc);
+void ff_vp7_idct_dc_add4y_rvv(uint8_t *dst, int16_t block[4][16], ptrdiff_t);
 
 static void ff_vp7_idct_dc_add_rvv(uint8_t *dst, int16_t block[16],
                                    ptrdiff_t stride)
@@ -49,6 +50,7 @@ av_cold void ff_vp7dsp_init_riscv(VP8DSPContext *c)
         c->vp8_idct_add = ff_vp7_idct_add_rvv;
 #endif
         c->vp8_idct_dc_add = ff_vp7_idct_dc_add_rvv;
+        c->vp8_idct_dc_add4y  = ff_vp7_idct_dc_add4y_rvv;
     }
 #endif
 }

libavcodec/riscv/vp7dsp_rvv.S

@@ -127,3 +127,19 @@ func ff_vp7_idct_add_rvv, zve32x
         ret
 endfunc
 #endif
+
+func ff_vp7_idct_dc_add4y_rvv, zve32x
+        li       t0, 32
+        vsetivli zero, 4, e16, mf2, ta, ma
+        li       t1, 23170
+        vlse16.v v8, (a1), t0 # block[0..3][0]
+        vwmul.vx v0, v8, t1
+        li       t2, 0x20000
+        vsetvli  zero, zero, e32, m1, ta, ma
+        vsra.vi  v0, v0, 14
+        vmul.vx  v0, v0, t1
+        vadd.vx  v0, v0, t2
+        vsetvli  zero, zero, e16, mf2, ta, ma
+        vnsra.wi v8, v0, 18   # 4x DC
+        tail     ff_vp78_idct_dc_add4y_rvv
+endfunc

libavcodec/riscv/vp8dsp_rvv.S

@@ -105,6 +105,7 @@ func ff_vp8_idct_dc_add_rvv, zve32x
         # fall through
 endfunc
 
+# a3 = DC
 func ff_vp78_idct_dc_add_rvv, zve32x
         csrwi      vxrm, 0
         vsetivli   zero, 4, e8, mf4, ta, ma
@@ -121,6 +122,41 @@ func ff_vp78_idct_dc_add_rvv, zve32x
         ret
 endfunc
 
+func ff_vp8_idct_dc_add4y_rvv, zve32x
+        li       t0, 32
+        vsetivli zero, 4, e16, mf2, ta, ma
+        vlse16.v v8, (a1), t0
+        vadd.vi  v8, v8, 4
+        vsra.vi  v8, v8, 3
+        # fall through
+endfunc
+
+        .variant_cc ff_vp78_idct_dc_add4y_rvv
+# v8 = [dc0, dc1, dc2, dc3]
+func ff_vp78_idct_dc_add4y_rvv, zve32x
+        vsetivli    zero, 16, e16, m2, ta, ma
+        vid.v       v4
+        vsrl.vi     v4, v4, 2
+        vrgather.vv v0, v8, v4 # replicate each DC four times
+        vsetvli     zero, zero, e8, m1, ta, ma
+        li          a4, 4
+1:
+        vle8.v      v8, (a0)
+        addi        a4, a4, -1
+        vwaddu.wv   v16, v0, v8
+        sh          zero, (a1)
+        vsetvli     zero, zero, e16, m2, ta, ma
+        vmax.vx     v16, v16, zero
+        addi        a1, a1, 32
+        vsetvli     zero, zero, e8, m1, ta, ma
+        vnclipu.wi  v8, v16, 0
+        vse8.v      v8, (a0)
+        add         a0, a0, a2
+        bnez        a4, 1b
+
+        ret
+endfunc
+
 .macro vp8_idct_dc_add
         vlse32.v      v0, (a0), a2
         lh            a5, 0(a1)
@@ -143,16 +179,6 @@ endfunc
         addi          a1, a1, 32
 .endm
 
-func ff_vp8_idct_dc_add4y_rvv, zve32x
-        vsetivli      zero, 4, e8, mf4, ta, ma
-        .rept 3
-        vp8_idct_dc_addy
-        .endr
-        vp8_idct_dc_add
-
-        ret
-endfunc
-
 func ff_vp8_idct_dc_add4uv_rvv, zve32x
         vsetivli      zero, 4, e8, mf4, ta, ma
         vp8_idct_dc_addy