@@ -26,6 +26,7 @@
#include "libavcodec/vp8dsp.h"
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);
av_cold void ff_vp7dsp_init_riscv(VP8DSPContext *c)
{
@@ -37,6 +38,7 @@ av_cold void ff_vp7dsp_init_riscv(VP8DSPContext *c)
#if __riscv_xlen >= 64
c->vp8_luma_dc_wht = ff_vp7_luma_dc_wht_rvv;
#endif
+ c->vp8_idct_add = ff_vp7_idct_add_rvv;
}
#endif
}
@@ -97,4 +97,33 @@ func ff_vp7_luma_dc_wht_rvv, zve32x
vnclip.wi v7, v3, 18
jr t0
endfunc
+
+func ff_vp7_idct_add_rvv, zve32x
+ jal t0, 1b
+ csrwi vxrm, 2
+ vsetvli zero, zero, e8, mf4, ta, ma
+ vle8.v v12, (a0)
+ vle8.v v13, (t1)
+ vwaddu.wv v4, v4, v12
+ vle8.v v14, (t2)
+ vwaddu.wv v5, v5, v13
+ vle8.v v15, (t3)
+ vwaddu.wv v6, v6, v14
+ vwaddu.wv v7, v7, v15
+ vsetvli zero, zero, e16, mf2, ta, ma
+ vmax.vx v4, v4, zero
+ vmax.vx v5, v5, zero
+ vmax.vx v6, v6, zero
+ vmax.vx v7, v7, zero
+ vsetvli zero, zero, e8, mf4, ta, ma
+ vnclipu.wi v0, v4, 0
+ vnclipu.wi v1, v5, 0
+ vse8.v v0, (a0)
+ vnclipu.wi v2, v6, 0
+ vse8.v v1, (t1)
+ vnclipu.wi v3, v7, 0
+ vse8.v v2, (t2)
+ vse8.v v3, (t3)
+ ret
+endfunc
#endif