H264: change weight/biweight functions to take a height argument.
[ffmpeg.git] / libavcodec / arm / h264dsp_init_arm.c
index 20f5ac2..1c331a4 100644 (file)
@@ -32,47 +32,22 @@ void ff_h264_v_loop_filter_chroma_neon(uint8_t *pix, int stride, int alpha,
 void ff_h264_h_loop_filter_chroma_neon(uint8_t *pix, int stride, int alpha,
                                        int beta, int8_t *tc0);
 
-void ff_weight_h264_pixels_16x16_neon(uint8_t *ds, int stride, int log2_den,
-                                      int weight, int offset);
-void ff_weight_h264_pixels_16x8_neon(uint8_t *ds, int stride, int log2_den,
-                                     int weight, int offset);
-void ff_weight_h264_pixels_8x16_neon(uint8_t *ds, int stride, int log2_den,
-                                     int weight, int offset);
-void ff_weight_h264_pixels_8x8_neon(uint8_t *ds, int stride, int log2_den,
-                                    int weight, int offset);
-void ff_weight_h264_pixels_8x4_neon(uint8_t *ds, int stride, int log2_den,
-                                    int weight, int offset);
-void ff_weight_h264_pixels_4x8_neon(uint8_t *ds, int stride, int log2_den,
-                                    int weight, int offset);
-void ff_weight_h264_pixels_4x4_neon(uint8_t *ds, int stride, int log2_den,
-                                    int weight, int offset);
-void ff_weight_h264_pixels_4x2_neon(uint8_t *ds, int stride, int log2_den,
-                                    int weight, int offset);
+void ff_weight_h264_pixels_16_neon(uint8_t *dst, int stride, int height,
+                                   int log2_den, int weight, int offset);
+void ff_weight_h264_pixels_8_neon(uint8_t *dst, int stride, int height,
+                                  int log2_den, int weight, int offset);
+void ff_weight_h264_pixels_4_neon(uint8_t *dst, int stride, int height,
+                                  int log2_den, int weight, int offset);
 
-void ff_biweight_h264_pixels_16x16_neon(uint8_t *dst, uint8_t *src, int stride,
-                                        int log2_den, int weightd, int weights,
-                                        int offset);
-void ff_biweight_h264_pixels_16x8_neon(uint8_t *dst, uint8_t *src, int stride,
-                                       int log2_den, int weightd, int weights,
-                                       int offset);
-void ff_biweight_h264_pixels_8x16_neon(uint8_t *dst, uint8_t *src, int stride,
-                                       int log2_den, int weightd, int weights,
-                                       int offset);
-void ff_biweight_h264_pixels_8x8_neon(uint8_t *dst, uint8_t *src, int stride,
-                                      int log2_den, int weightd, int weights,
-                                      int offset);
-void ff_biweight_h264_pixels_8x4_neon(uint8_t *dst, uint8_t *src, int stride,
-                                      int log2_den, int weightd, int weights,
-                                      int offset);
-void ff_biweight_h264_pixels_4x8_neon(uint8_t *dst, uint8_t *src, int stride,
-                                      int log2_den, int weightd, int weights,
-                                      int offset);
-void ff_biweight_h264_pixels_4x4_neon(uint8_t *dst, uint8_t *src, int stride,
-                                      int log2_den, int weightd, int weights,
-                                      int offset);
-void ff_biweight_h264_pixels_4x2_neon(uint8_t *dst, uint8_t *src, int stride,
-                                      int log2_den, int weightd, int weights,
-                                      int offset);
+void ff_biweight_h264_pixels_16_neon(uint8_t *dst, uint8_t *src, int stride,
+                                     int height, int log2_den, int weightd,
+                                     int weights, int offset);
+void ff_biweight_h264_pixels_8_neon(uint8_t *dst, uint8_t *src, int stride,
+                                    int height, int log2_den, int weightd,
+                                    int weights, int offset);
+void ff_biweight_h264_pixels_4_neon(uint8_t *dst, uint8_t *src, int stride,
+                                    int height, int log2_den, int weightd,
+                                    int weights, int offset);
 
 void ff_h264_idct_add_neon(uint8_t *dst, DCTELEM *block, int stride);
 void ff_h264_idct_dc_add_neon(uint8_t *dst, DCTELEM *block, int stride);
@@ -92,42 +67,35 @@ void ff_h264_idct8_add4_neon(uint8_t *dst, const int *block_offset,
                              DCTELEM *block, int stride,
                              const uint8_t nnzc[6*8]);
 
-static void ff_h264dsp_init_neon(H264DSPContext *c)
+static void ff_h264dsp_init_neon(H264DSPContext *c, const int bit_depth, const int chroma_format_idc)
 {
+    if (bit_depth == 8) {
     c->h264_v_loop_filter_luma   = ff_h264_v_loop_filter_luma_neon;
     c->h264_h_loop_filter_luma   = ff_h264_h_loop_filter_luma_neon;
     c->h264_v_loop_filter_chroma = ff_h264_v_loop_filter_chroma_neon;
     c->h264_h_loop_filter_chroma = ff_h264_h_loop_filter_chroma_neon;
 
-    c->weight_h264_pixels_tab[0] = ff_weight_h264_pixels_16x16_neon;
-    c->weight_h264_pixels_tab[1] = ff_weight_h264_pixels_16x8_neon;
-    c->weight_h264_pixels_tab[2] = ff_weight_h264_pixels_8x16_neon;
-    c->weight_h264_pixels_tab[3] = ff_weight_h264_pixels_8x8_neon;
-    c->weight_h264_pixels_tab[4] = ff_weight_h264_pixels_8x4_neon;
-    c->weight_h264_pixels_tab[5] = ff_weight_h264_pixels_4x8_neon;
-    c->weight_h264_pixels_tab[6] = ff_weight_h264_pixels_4x4_neon;
-    c->weight_h264_pixels_tab[7] = ff_weight_h264_pixels_4x2_neon;
+    c->weight_h264_pixels_tab[0] = ff_weight_h264_pixels_16_neon;
+    c->weight_h264_pixels_tab[1] = ff_weight_h264_pixels_8_neon;
+    c->weight_h264_pixels_tab[2] = ff_weight_h264_pixels_4_neon;
 
-    c->biweight_h264_pixels_tab[0] = ff_biweight_h264_pixels_16x16_neon;
-    c->biweight_h264_pixels_tab[1] = ff_biweight_h264_pixels_16x8_neon;
-    c->biweight_h264_pixels_tab[2] = ff_biweight_h264_pixels_8x16_neon;
-    c->biweight_h264_pixels_tab[3] = ff_biweight_h264_pixels_8x8_neon;
-    c->biweight_h264_pixels_tab[4] = ff_biweight_h264_pixels_8x4_neon;
-    c->biweight_h264_pixels_tab[5] = ff_biweight_h264_pixels_4x8_neon;
-    c->biweight_h264_pixels_tab[6] = ff_biweight_h264_pixels_4x4_neon;
-    c->biweight_h264_pixels_tab[7] = ff_biweight_h264_pixels_4x2_neon;
+    c->biweight_h264_pixels_tab[0] = ff_biweight_h264_pixels_16_neon;
+    c->biweight_h264_pixels_tab[1] = ff_biweight_h264_pixels_8_neon;
+    c->biweight_h264_pixels_tab[2] = ff_biweight_h264_pixels_4_neon;
 
     c->h264_idct_add        = ff_h264_idct_add_neon;
     c->h264_idct_dc_add     = ff_h264_idct_dc_add_neon;
     c->h264_idct_add16      = ff_h264_idct_add16_neon;
     c->h264_idct_add16intra = ff_h264_idct_add16intra_neon;
-    c->h264_idct_add8       = ff_h264_idct_add8_neon;
+    if (chroma_format_idc == 1)
+        c->h264_idct_add8   = ff_h264_idct_add8_neon;
     c->h264_idct8_add       = ff_h264_idct8_add_neon;
     c->h264_idct8_dc_add    = ff_h264_idct8_dc_add_neon;
     c->h264_idct8_add4      = ff_h264_idct8_add4_neon;
+    }
 }
 
-void ff_h264dsp_init_arm(H264DSPContext *c)
+void ff_h264dsp_init_arm(H264DSPContext *c, const int bit_depth, const int chroma_format_idc)
 {
-    if (HAVE_NEON) ff_h264dsp_init_neon(c);
+    if (HAVE_NEON) ff_h264dsp_init_neon(c, bit_depth, chroma_format_idc);
 }