diff options
| author | gxw <guxiwei-hf@loongson.cn> | 2019-08-07 17:52:00 +0800 |
|---|---|---|
| committer | Michael Niedermayer <michael@niedermayer.cc> | 2019-08-13 16:48:38 +0200 |
| commit | a3e572d96fd1dd6291f6b28e173db858c08ff8d8 (patch) | |
| tree | 85807f6ec1442cc362cf8946e67f564c92267e07 /libavcodec/mips/hevc_idct_msa.c | |
| parent | 8f92eb05e063e6c4d6e36521020620d4e6e1c21d (diff) | |
| download | ffmpeg-a3e572d96fd1dd6291f6b28e173db858c08ff8d8.tar.gz | |
avutil/mips: refine msa macros CLIP_*.
Changing details as following:
1. Remove the local variable 'out_m' in 'CLIP_SH' and store the result in
source vector.
2. Refine the implementation of macro 'CLIP_SH_0_255' and 'CLIP_SW_0_255'.
Performance of VP8 decoding has speed up about 1.1%(from 7.03x to 7.11x).
Performance of H264 decoding has speed up about 0.5%(from 4.35x to 4.37x).
Performance of Theora decoding has speed up about 0.7%(from 5.79x to 5.83x).
3. Remove redundant macro 'CLIP_SH/Wn_0_255_MAX_SATU' and use 'CLIP_SH/Wn_0_255'
instead, because there are no difference in the effect of this two macros.
Reviewed-by: Shiyou Yin <yinshiyou-hf@loongson.cn>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Diffstat (limited to 'libavcodec/mips/hevc_idct_msa.c')
| -rw-r--r-- | libavcodec/mips/hevc_idct_msa.c | 21 |
1 files changed, 11 insertions, 10 deletions
diff --git a/libavcodec/mips/hevc_idct_msa.c b/libavcodec/mips/hevc_idct_msa.c index b14aec95eb..5ab6acd1df 100644 --- a/libavcodec/mips/hevc_idct_msa.c +++ b/libavcodec/mips/hevc_idct_msa.c @@ -803,8 +803,9 @@ static void hevc_addblk_16x16_msa(int16_t *coeffs, uint8_t *dst, int32_t stride) LD_SH4((coeffs + 8), 16, in1, in3, in5, in7); coeffs += 64; - CLIP_SH4_0_255(dst_r0, dst_l0, dst_r1, dst_l1); - CLIP_SH4_0_255(dst_r2, dst_l2, dst_r3, dst_l3); + CLIP_SH8_0_255(dst_r0, dst_l0, dst_r1, dst_l1, + dst_r2, dst_l2, dst_r3, dst_l3); + PCKEV_B4_UB(dst_l0, dst_r0, dst_l1, dst_r1, dst_l2, dst_r2, dst_l3, dst_r3, dst0, dst1, dst2, dst3); ST_UB4(dst0, dst1, dst2, dst3, dst, stride); @@ -825,8 +826,8 @@ static void hevc_addblk_16x16_msa(int16_t *coeffs, uint8_t *dst, int32_t stride) dst_r3 += in6; dst_l3 += in7; - CLIP_SH4_0_255(dst_r0, dst_l0, dst_r1, dst_l1); - CLIP_SH4_0_255(dst_r2, dst_l2, dst_r3, dst_l3); + CLIP_SH8_0_255(dst_r0, dst_l0, dst_r1, dst_l1, + dst_r2, dst_l2, dst_r3, dst_l3); PCKEV_B4_UB(dst_l0, dst_r0, dst_l1, dst_r1, dst_l2, dst_r2, dst_l3, dst_r3, dst0, dst1, dst2, dst3); ST_UB4(dst0, dst1, dst2, dst3, dst, stride); @@ -873,8 +874,8 @@ static void hevc_addblk_32x32_msa(int16_t *coeffs, uint8_t *dst, int32_t stride) LD_SH4((coeffs + 8), 16, in1, in3, in5, in7); coeffs += 64; - CLIP_SH4_0_255(dst_r0, dst_l0, dst_r1, dst_l1); - CLIP_SH4_0_255(dst_r2, dst_l2, dst_r3, dst_l3); + CLIP_SH8_0_255(dst_r0, dst_l0, dst_r1, dst_l1, + dst_r2, dst_l2, dst_r3, dst_l3); PCKEV_B4_UB(dst_l0, dst_r0, dst_l1, dst_r1, dst_l2, dst_r2, dst_l3, dst_r3, dst0, dst1, dst2, dst3); ST_UB2(dst0, dst1, dst, 16); @@ -905,8 +906,8 @@ static void hevc_addblk_32x32_msa(int16_t *coeffs, uint8_t *dst, int32_t stride) LD_SH4(coeffs, 16, in0, in2, in4, in6); LD_SH4((coeffs + 8), 16, in1, in3, in5, in7); - CLIP_SH4_0_255(dst_r0, dst_l0, dst_r1, dst_l1); - CLIP_SH4_0_255(dst_r2, dst_l2, dst_r3, dst_l3); + CLIP_SH8_0_255(dst_r0, dst_l0, dst_r1, dst_l1, + dst_r2, dst_l2, dst_r3, dst_l3); PCKEV_B4_UB(dst_l0, dst_r0, dst_l1, dst_r1, dst_l2, dst_r2, dst_l3, dst_r3, dst0, dst1, dst2, dst3); ST_UB2(dst0, dst1, dst, 16); @@ -928,8 +929,8 @@ static void hevc_addblk_32x32_msa(int16_t *coeffs, uint8_t *dst, int32_t stride) dst_r3 += in6; dst_l3 += in7; - CLIP_SH4_0_255(dst_r0, dst_l0, dst_r1, dst_l1); - CLIP_SH4_0_255(dst_r2, dst_l2, dst_r3, dst_l3); + CLIP_SH8_0_255(dst_r0, dst_l0, dst_r1, dst_l1, + dst_r2, dst_l2, dst_r3, dst_l3); PCKEV_B4_UB(dst_l0, dst_r0, dst_l1, dst_r1, dst_l2, dst_r2, dst_l3, dst_r3, dst0, dst1, dst2, dst3); ST_UB2(dst0, dst1, dst, 16); |