avcodec: loongson optimize h264dsp idct and loop filter with mmi

Change-Id: Ic87fb8f5cd22a502ff9dbbc5a5a8ea97cfc8a1dd
Signed-off-by: ZhouXiaoyong <zhouxiaoyong@loongson.cn>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

libavcodec/mips/h264dsp_init_mips.c

@@ -82,11 +82,33 @@ static av_cold void h264dsp_init_msa(H264DSPContext *c,
 #endif  // #if HAVE_MSA
 
 #if HAVE_MMI
-static av_cold void h264dsp_init_mmi(H264DSPContext * c,
-                                     const int bit_depth,
-                                     const int chroma_format_idc)
+static av_cold void h264dsp_init_mmi(H264DSPContext * c, const int bit_depth,
+        const int chroma_format_idc)
 {
     if (bit_depth == 8) {
+        c->h264_add_pixels4_clear = ff_h264_add_pixels4_8_mmi;
+        c->h264_idct_add = ff_h264_idct_add_8_mmi;
+        c->h264_idct8_add = ff_h264_idct8_add_8_mmi;
+        c->h264_idct_dc_add = ff_h264_idct_dc_add_8_mmi;
+        c->h264_idct8_dc_add = ff_h264_idct8_dc_add_8_mmi;
+        c->h264_idct_add16 = ff_h264_idct_add16_8_mmi;
+        c->h264_idct_add16intra = ff_h264_idct_add16intra_8_mmi;
+        c->h264_idct8_add4 = ff_h264_idct8_add4_8_mmi;
+
+        if (chroma_format_idc <= 1)
+            c->h264_idct_add8 = ff_h264_idct_add8_8_mmi;
+        else
+            c->h264_idct_add8 = ff_h264_idct_add8_422_8_mmi;
+
+        c->h264_luma_dc_dequant_idct = ff_h264_luma_dc_dequant_idct_8_mmi;
+
+        if (chroma_format_idc <= 1)
+            c->h264_chroma_dc_dequant_idct =
+                ff_h264_chroma_dc_dequant_idct_8_mmi;
+        else
+            c->h264_chroma_dc_dequant_idct =
+                ff_h264_chroma422_dc_dequant_idct_8_mmi;
+
         c->weight_h264_pixels_tab[0] = ff_h264_weight_pixels16_8_mmi;
         c->weight_h264_pixels_tab[1] = ff_h264_weight_pixels8_8_mmi;
         c->weight_h264_pixels_tab[2] = ff_h264_weight_pixels4_8_mmi;
@@ -94,6 +116,21 @@ static av_cold void h264dsp_init_mmi(H264DSPContext * c,
         c->biweight_h264_pixels_tab[0] = ff_h264_biweight_pixels16_8_mmi;
         c->biweight_h264_pixels_tab[1] = ff_h264_biweight_pixels8_8_mmi;
         c->biweight_h264_pixels_tab[2] = ff_h264_biweight_pixels4_8_mmi;
+
+        c->h264_v_loop_filter_chroma       = ff_deblock_v_chroma_8_mmi;
+        c->h264_v_loop_filter_chroma_intra = ff_deblock_v_chroma_intra_8_mmi;
+
+        if (chroma_format_idc <= 1) {
+            c->h264_h_loop_filter_chroma =
+                ff_deblock_h_chroma_8_mmi;
+            c->h264_h_loop_filter_chroma_intra =
+                ff_deblock_h_chroma_intra_8_mmi;
+        }
+
+        c->h264_v_loop_filter_luma = ff_deblock_v_luma_8_mmi;
+        c->h264_v_loop_filter_luma_intra = ff_deblock_v_luma_intra_8_mmi;
+        c->h264_h_loop_filter_luma = ff_deblock_h_luma_8_mmi;
+        c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_8_mmi;
     }
 }
 #endif /* HAVE_MMI */

libavcodec/mips/h264dsp_mips.h

@@ -319,6 +319,26 @@ void ff_vp8_pred8x8_129_dc_8_msa(uint8_t *src, ptrdiff_t stride);
 void ff_vp8_pred16x16_127_dc_8_msa(uint8_t *src, ptrdiff_t stride);
 void ff_vp8_pred16x16_129_dc_8_msa(uint8_t *src, ptrdiff_t stride);
 
+void ff_h264_add_pixels4_8_mmi(uint8_t *_dst, int16_t *_src, int stride);
+void ff_h264_idct_add_8_mmi(uint8_t *dst, int16_t *block, int stride);
+void ff_h264_idct8_add_8_mmi(uint8_t *dst, int16_t *block, int stride);
+void ff_h264_idct_dc_add_8_mmi(uint8_t *dst, int16_t *block, int stride);
+void ff_h264_idct8_dc_add_8_mmi(uint8_t *dst, int16_t *block, int stride);
+void ff_h264_idct_add16_8_mmi(uint8_t *dst, const int *block_offset,
+        int16_t *block, int stride, const uint8_t nnzc[15*8]);
+void ff_h264_idct_add16intra_8_mmi(uint8_t *dst, const int *block_offset,
+        int16_t *block, int stride, const uint8_t nnzc[15*8]);
+void ff_h264_idct8_add4_8_mmi(uint8_t *dst, const int *block_offset,
+        int16_t *block, int stride, const uint8_t nnzc[15*8]);
+void ff_h264_idct_add8_8_mmi(uint8_t **dest, const int *block_offset,
+        int16_t *block, int stride, const uint8_t nnzc[15*8]);
+void ff_h264_idct_add8_422_8_mmi(uint8_t **dest, const int *block_offset,
+        int16_t *block, int stride, const uint8_t nnzc[15*8]);
+void ff_h264_luma_dc_dequant_idct_8_mmi(int16_t *output, int16_t *input,
+        int qmul);
+void ff_h264_chroma_dc_dequant_idct_8_mmi(int16_t *block, int qmul);
+void ff_h264_chroma422_dc_dequant_idct_8_mmi(int16_t *block, int qmul);
+
 void ff_h264_weight_pixels16_8_mmi(uint8_t *block, int stride, int height,
         int log2_denom, int weight, int offset);
 void ff_h264_biweight_pixels16_8_mmi(uint8_t *dst, uint8_t *src,
@@ -335,6 +355,27 @@ void ff_h264_biweight_pixels4_8_mmi(uint8_t *dst, uint8_t *src,
         int stride, int height, int log2_denom, int weightd, int weights,
         int offset);
 
+void ff_deblock_v_chroma_8_mmi(uint8_t *pix, int stride, int alpha, int beta,
+        int8_t *tc0);
+void ff_deblock_v_chroma_intra_8_mmi(uint8_t *pix, int stride, int alpha,
+        int beta);
+void ff_deblock_h_chroma_8_mmi(uint8_t *pix, int stride, int alpha, int beta,
+        int8_t *tc0);
+void ff_deblock_h_chroma_intra_8_mmi(uint8_t *pix, int stride, int alpha,
+        int beta);
+void ff_deblock_v_luma_8_mmi(uint8_t *pix, int stride, int alpha, int beta,
+        int8_t *tc0);
+void ff_deblock_v_luma_intra_8_mmi(uint8_t *pix, int stride, int alpha,
+        int beta);
+void ff_deblock_h_luma_8_mmi(uint8_t *pix, int stride, int alpha, int beta,
+        int8_t *tc0);
+void ff_deblock_h_luma_intra_8_mmi(uint8_t *pix, int stride, int alpha,
+        int beta);
+void ff_deblock_v8_luma_8_mmi(uint8_t *pix, int stride, int alpha, int beta,
+        int8_t *tc0);
+void ff_deblock_v8_luma_intra_8_mmi(uint8_t *pix, int stride, int alpha,
+        int beta);
+
 void ff_put_h264_qpel16_mc00_mmi(uint8_t *dst, const uint8_t *src,
         ptrdiff_t dst_stride);
 void ff_put_h264_qpel16_mc10_mmi(uint8_t *dst, const uint8_t *src,