diff options
| author | Rémi Denis-Courmont <remi@remlab.net> | 2024-05-26 13:22:45 +0300 |
|---|---|---|
| committer | Rémi Denis-Courmont <remi@remlab.net> | 2024-05-29 16:57:02 +0300 |
| commit | fa3b153cb1eba0d68327c716842c9a38f95c1667 (patch) | |
| tree | 30a4711c9bb6adbd5e8c9bf0ce3556b6be88568f | |
| parent | 4a0e629b6f4b6fb612ce08a471fcf3c249d35eaa (diff) | |
| download | ffmpeg-fa3b153cb1eba0d68327c716842c9a38f95c1667.tar.gz | |
lavc/vp7dsp: R-V V vp7_idct_add
Most of the code is shared with DC, thanks to minor earlier changes.
vp7_idct_add_c: 5.2
vp7_idct_add_rvv_i32: 2.5
| -rw-r--r-- | libavcodec/riscv/vp7dsp_init.c | 2 | ||||
| -rw-r--r-- | libavcodec/riscv/vp7dsp_rvv.S | 29 |
2 files changed, 31 insertions, 0 deletions
diff --git a/libavcodec/riscv/vp7dsp_init.c b/libavcodec/riscv/vp7dsp_init.c index 1406eb1c6a..ae7f2d4277 100644 --- a/libavcodec/riscv/vp7dsp_init.c +++ b/libavcodec/riscv/vp7dsp_init.c @@ -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 } diff --git a/libavcodec/riscv/vp7dsp_rvv.S b/libavcodec/riscv/vp7dsp_rvv.S index 06e251f5ce..2a4c404bbb 100644 --- a/libavcodec/riscv/vp7dsp_rvv.S +++ b/libavcodec/riscv/vp7dsp_rvv.S @@ -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 |