shithub: libvpx

--- a/vp9/common/vp9_blockd.h

+++ b/vp9/common/vp9_blockd.h

@@ -390,14 +390,13 @@

                                                   void *arg);

 static INLINE void foreach_transformed_block_in_plane(

-    const MACROBLOCKD* const xd, BLOCK_SIZE_TYPE bsize, int plane,

+    const MACROBLOCKD *const xd, BLOCK_SIZE_TYPE bsize, int plane,

     foreach_transformed_block_visitor visit, void *arg) {

   const struct macroblockd_plane *const pd = &xd->plane[plane];

+  const MB_MODE_INFO* mbmi = &xd->mode_info_context->mbmi;

   // block and transform sizes, in number of 4x4 blocks log 2 ("*_b")

   // 4x4=0, 8x8=2, 16x16=4, 32x32=6, 64x64=8

   // transform size varies per plane, look it up in a common way.

-  const MB_MODE_INFO* mbmi = &xd->mode_info_context->mbmi;

   const TX_SIZE tx_size = plane ? get_uv_tx_size(mbmi)

                                 : mbmi->txfm_size;

   const BLOCK_SIZE_TYPE plane_bsize = get_plane_block_size(bsize, pd);

--- a/vp9/encoder/vp9_encodeframe.c

+++ b/vp9/encoder/vp9_encodeframe.c

@@ -574,6 +574,7 @@

   MACROBLOCK *const x = &cpi->mb;

   MACROBLOCKD *const xd = &x->e_mbd;

+  // Use the lower precision, but faster, 32x32 fdct for mode selection.

   x->use_lp32x32fdct = 1;

   if (bsize < BLOCK_8X8) {

@@ -1757,8 +1758,6 @@

-  // Use 4 subblocks' motion estimation results to speed up current

-  // partition's checking.

   x->fast_ms = 0;

   x->pred_mv.as_int = 0;

   x->subblock_ref = 0;

@@ -2178,7 +2177,7 @@

     build_activity_map(cpi);

-  // re-initencode frame context.

+  // Re-initialize encode frame context.

   init_encode_frame_mb_context(cpi);

   vp9_zero(cpi->rd_comp_pred_diff);

@@ -2414,7 +2413,7 @@

   // decoder such that we allow compound where one of the 3 buffers has a

   // different sign bias and that buffer is then the fixed ref. However, this

   // requires further work in the rd loop. For now the only supported encoder

-  // side behaviour is where the ALT ref buffer has opposite sign bias to

+  // side behavior is where the ALT ref buffer has opposite sign bias to

   // the other two.

   if ((cm->ref_frame_sign_bias[ALTREF_FRAME]

        == cm->ref_frame_sign_bias[GOLDEN_FRAME])

@@ -2481,13 +2480,12 @@

       filter_type = SWITCHABLE;

-    /* transform size (4x4, 8x8, 16x16 or select-per-mb) selection */

     cpi->mb.e_mbd.lossless = 0;

     if (cpi->oxcf.lossless) {

       cpi->mb.e_mbd.lossless = 1;

+    /* transform size selection (4x4, 8x8, 16x16 or select-per-mb) */

     select_tx_mode(cpi);

     cpi->common.comp_pred_mode = pred_type;

     cpi->common.mcomp_filter_type = filter_type;

--- a/vp9/encoder/vp9_rdopt.c

+++ b/vp9/encoder/vp9_rdopt.c

@@ -1217,6 +1217,7 @@

   MB_PREDICTION_MODE mode;

   MB_PREDICTION_MODE mode_selected = DC_PRED;

   MACROBLOCKD *const xd = &x->e_mbd;

+  MODE_INFO *const mic = xd->mode_info_context;

   int this_rate, this_rate_tokenonly, s;

   int64_t this_distortion, this_rd;

   TX_SIZE best_tx = TX_4X4;

@@ -1230,7 +1231,6 @@

   /* Y Search for intra prediction mode */

   for (mode = DC_PRED; mode <= TM_PRED; mode++) {

     int64_t local_tx_cache[TX_MODES];

-    MODE_INFO *const mic = xd->mode_info_context;

     const int mis = xd->mode_info_stride;

     if (cpi->common.frame_type == KEY_FRAME) {

@@ -1240,7 +1240,7 @@

       bmode_costs = x->y_mode_costs[A][L];

-    x->e_mbd.mode_info_context->mbmi.mode = mode;

+    mic->mbmi.mode = mode;

     super_block_yrd(cpi, x, &this_rate_tokenonly, &this_distortion, &s, NULL,

                     bsize, local_tx_cache, best_rd);

@@ -1254,7 +1254,7 @@

     if (this_rd < best_rd) {

       mode_selected   = mode;

       best_rd         = this_rd;

-      best_tx         = x->e_mbd.mode_info_context->mbmi.txfm_size;

+      best_tx         = mic->mbmi.txfm_size;

       *rate           = this_rate;

       *rate_tokenonly = this_rate_tokenonly;

       *distortion     = this_distortion;

@@ -1272,8 +1272,8 @@

-  x->e_mbd.mode_info_context->mbmi.mode = mode_selected;

-  x->e_mbd.mode_info_context->mbmi.txfm_size = best_tx;

+  mic->mbmi.mode = mode_selected;

+  mic->mbmi.txfm_size = best_tx;

   return best_rd;

@@ -1437,7 +1437,7 @@

                        int *mvjcost, int *mvcost[2], VP9_COMP *cpi) {

   MACROBLOCKD *const xd = &x->e_mbd;

   MODE_INFO *const mic = xd->mode_info_context;

-  MB_MODE_INFO * mbmi = &mic->mbmi;

+  MB_MODE_INFO *mbmi = &mic->mbmi;

   int cost = 0, thismvcost = 0;

   int idx, idy;

   const int num_4x4_blocks_wide = num_4x4_blocks_wide_lookup[mbmi->sb_type];

--- a/vp9/encoder/vp9_variance_c.c

+++ b/vp9/encoder/vp9_variance_c.c

@@ -46,7 +46,7 @@

                                            const uint8_t *dst_ptr,

                                            int dst_pixels_per_line,

                                            unsigned int *sse) {

-  uint16_t fdata3[65 * 64];  // Temp data bufffer used in filtering

+  uint16_t fdata3[65 * 64];  // Temp data buffer used in filtering

   uint8_t temp2[68 * 64];

   const int16_t *hfilter, *vfilter;

@@ -68,7 +68,7 @@

                                                int dst_pixels_per_line,

                                                unsigned int *sse,

                                                const uint8_t *second_pred) {

-  uint16_t fdata3[65 * 64];  // Temp data bufffer used in filtering

+  uint16_t fdata3[65 * 64];  // Temp data buffer used in filtering

   uint8_t temp2[68 * 64];

   DECLARE_ALIGNED_ARRAY(16, uint8_t, temp3, 64 * 64);  // compound pred buffer

   const int16_t *hfilter, *vfilter;

@@ -103,7 +103,7 @@

                                            const uint8_t *dst_ptr,

                                            int dst_pixels_per_line,

                                            unsigned int *sse) {

-  uint16_t fdata3[65 * 64];  // Temp data bufffer used in filtering

+  uint16_t fdata3[65 * 64];  // Temp data buffer used in filtering

   uint8_t temp2[68 * 64];

   const int16_t *hfilter, *vfilter;

@@ -125,7 +125,7 @@

                                                int dst_pixels_per_line,

                                                unsigned int *sse,

                                                const uint8_t *second_pred) {

-  uint16_t fdata3[65 * 64];  // Temp data bufffer used in filtering

+  uint16_t fdata3[65 * 64];  // Temp data buffer used in filtering

   uint8_t temp2[68 * 64];

   DECLARE_ALIGNED_ARRAY(16, uint8_t, temp3, 32 * 64);  // compound pred buffer

   const int16_t *hfilter, *vfilter;

@@ -160,7 +160,7 @@

                                            const uint8_t *dst_ptr,

                                            int dst_pixels_per_line,

                                            unsigned int *sse) {

-  uint16_t fdata3[33 * 32];  // Temp data bufffer used in filtering

+  uint16_t fdata3[33 * 32];  // Temp data buffer used in filtering

   uint8_t temp2[36 * 32];

   const int16_t *hfilter, *vfilter;

@@ -182,7 +182,7 @@

                                                int dst_pixels_per_line,

                                                unsigned int *sse,

                                                const uint8_t *second_pred) {

-  uint16_t fdata3[33 * 32];  // Temp data bufffer used in filtering

+  uint16_t fdata3[33 * 32];  // Temp data buffer used in filtering

   uint8_t temp2[36 * 32];

   DECLARE_ALIGNED_ARRAY(16, uint8_t, temp3, 32 * 16);  // compound pred buffer

   const int16_t *hfilter, *vfilter;

@@ -217,7 +217,7 @@

                                            const uint8_t *dst_ptr,

                                            int dst_pixels_per_line,

                                            unsigned int *sse) {

-  uint16_t fdata3[33 * 32];  // Temp data bufffer used in filtering

+  uint16_t fdata3[33 * 32];  // Temp data buffer used in filtering

   uint8_t temp2[36 * 32];

   const int16_t *hfilter, *vfilter;

@@ -239,7 +239,7 @@

                                                int dst_pixels_per_line,

                                                unsigned int *sse,

                                                const uint8_t *second_pred) {

-  uint16_t fdata3[33 * 32];  // Temp data bufffer used in filtering

+  uint16_t fdata3[33 * 32];  // Temp data buffer used in filtering

   uint8_t temp2[36 * 32];

   DECLARE_ALIGNED_ARRAY(16, uint8_t, temp3, 16 * 32);  // compound pred buffer

   const int16_t *hfilter, *vfilter;

@@ -440,7 +440,7 @@

                                          unsigned int *sse) {

   uint8_t temp2[20 * 16];

   const int16_t *hfilter, *vfilter;

-  uint16_t fdata3[5 * 4];  // Temp data bufffer used in filtering

+  uint16_t fdata3[5 * 4];  // Temp data buffer used in filtering

   hfilter = VP9_BILINEAR_FILTERS_2TAP(xoffset);

   vfilter = VP9_BILINEAR_FILTERS_2TAP(yoffset);

@@ -466,7 +466,7 @@

   uint8_t temp2[20 * 16];

   const int16_t *hfilter, *vfilter;

   DECLARE_ALIGNED_ARRAY(16, uint8_t, temp3, 4 * 4);  // compound pred buffer

-  uint16_t fdata3[5 * 4];  // Temp data bufffer used in filtering

+  uint16_t fdata3[5 * 4];  // Temp data buffer used in filtering

   hfilter = VP9_BILINEAR_FILTERS_2TAP(xoffset);

   vfilter = VP9_BILINEAR_FILTERS_2TAP(yoffset);

@@ -488,7 +488,7 @@

                                          const uint8_t *dst_ptr,

                                          int dst_pixels_per_line,

                                          unsigned int *sse) {

-  uint16_t fdata3[9 * 8];  // Temp data bufffer used in filtering

+  uint16_t fdata3[9 * 8];  // Temp data buffer used in filtering

   uint8_t temp2[20 * 16];

   const int16_t *hfilter, *vfilter;

@@ -510,7 +510,7 @@

                                              int dst_pixels_per_line,

                                              unsigned int *sse,

                                              const uint8_t *second_pred) {

-  uint16_t fdata3[9 * 8];  // Temp data bufffer used in filtering

+  uint16_t fdata3[9 * 8];  // Temp data buffer used in filtering

   uint8_t temp2[20 * 16];

   DECLARE_ALIGNED_ARRAY(16, uint8_t, temp3, 8 * 8);  // compound pred buffer

   const int16_t *hfilter, *vfilter;

@@ -532,7 +532,7 @@

                                            const uint8_t *dst_ptr,

                                            int dst_pixels_per_line,

                                            unsigned int *sse) {

-  uint16_t fdata3[17 * 16];  // Temp data bufffer used in filtering

+  uint16_t fdata3[17 * 16];  // Temp data buffer used in filtering

   uint8_t temp2[20 * 16];

   const int16_t *hfilter, *vfilter;

@@ -577,7 +577,7 @@

                                            const uint8_t *dst_ptr,

                                            int dst_pixels_per_line,

                                            unsigned int *sse) {

-  uint16_t fdata3[65 * 64];  // Temp data bufffer used in filtering

+  uint16_t fdata3[65 * 64];  // Temp data buffer used in filtering

   uint8_t temp2[68 * 64];

   const int16_t *hfilter, *vfilter;

@@ -599,7 +599,7 @@

                                                int dst_pixels_per_line,

                                                unsigned int *sse,

                                                const uint8_t *second_pred) {

-  uint16_t fdata3[65 * 64];  // Temp data bufffer used in filtering

+  uint16_t fdata3[65 * 64];  // Temp data buffer used in filtering

   uint8_t temp2[68 * 64];

   DECLARE_ALIGNED_ARRAY(16, uint8_t, temp3, 64 * 64);  // compound pred buffer

   const int16_t *hfilter, *vfilter;

@@ -621,7 +621,7 @@

                                            const uint8_t *dst_ptr,

                                            int dst_pixels_per_line,

                                            unsigned int *sse) {

-  uint16_t fdata3[33 * 32];  // Temp data bufffer used in filtering

+  uint16_t fdata3[33 * 32];  // Temp data buffer used in filtering

   uint8_t temp2[36 * 32];

   const int16_t *hfilter, *vfilter;

@@ -643,7 +643,7 @@

                                                int dst_pixels_per_line,

                                                unsigned int *sse,

                                                const uint8_t *second_pred) {

-  uint16_t fdata3[33 * 32];  // Temp data bufffer used in filtering

+  uint16_t fdata3[33 * 32];  // Temp data buffer used in filtering

   uint8_t temp2[36 * 32];

   DECLARE_ALIGNED_ARRAY(16, uint8_t, temp3, 32 * 32);  // compound pred buffer

   const int16_t *hfilter, *vfilter;

@@ -785,7 +785,7 @@

                                           const uint8_t *dst_ptr,

                                           int dst_pixels_per_line,

                                           unsigned int *sse) {

-  uint16_t fdata3[16 * 9];  // Temp data bufffer used in filtering

+  uint16_t fdata3[16 * 9];  // Temp data buffer used in filtering

   uint8_t temp2[20 * 16];

   const int16_t *hfilter, *vfilter;

@@ -807,7 +807,7 @@

                                               int dst_pixels_per_line,

                                               unsigned int *sse,

                                               const uint8_t *second_pred) {

-  uint16_t fdata3[16 * 9];  // Temp data bufffer used in filtering

+  uint16_t fdata3[16 * 9];  // Temp data buffer used in filtering

   uint8_t temp2[20 * 16];

   DECLARE_ALIGNED_ARRAY(16, uint8_t, temp3, 16 * 8);  // compound pred buffer

   const int16_t *hfilter, *vfilter;

@@ -829,7 +829,7 @@

                                           const uint8_t *dst_ptr,

                                           int dst_pixels_per_line,

                                           unsigned int *sse) {

-  uint16_t fdata3[9 * 16];  // Temp data bufffer used in filtering

+  uint16_t fdata3[9 * 16];  // Temp data buffer used in filtering

   uint8_t temp2[20 * 16];

   const int16_t *hfilter, *vfilter;

@@ -851,7 +851,7 @@

                                               int dst_pixels_per_line,

                                               unsigned int *sse,

                                               const uint8_t *second_pred) {

-  uint16_t fdata3[9 * 16];  // Temp data bufffer used in filtering

+  uint16_t fdata3[9 * 16];  // Temp data buffer used in filtering

   uint8_t temp2[20 * 16];

   DECLARE_ALIGNED_ARRAY(16, uint8_t, temp3, 8 * 16);  // compound pred buffer

   const int16_t *hfilter, *vfilter;

@@ -873,7 +873,7 @@

                                          const uint8_t *dst_ptr,

                                          int dst_pixels_per_line,

                                          unsigned int *sse) {

-  uint16_t fdata3[8 * 5];  // Temp data bufffer used in filtering

+  uint16_t fdata3[8 * 5];  // Temp data buffer used in filtering

   uint8_t temp2[20 * 16];

   const int16_t *hfilter, *vfilter;

@@ -895,7 +895,7 @@

                                              int dst_pixels_per_line,

                                              unsigned int *sse,

                                              const uint8_t *second_pred) {

-  uint16_t fdata3[8 * 5];  // Temp data bufffer used in filtering

+  uint16_t fdata3[8 * 5];  // Temp data buffer used in filtering

   uint8_t temp2[20 * 16];

   DECLARE_ALIGNED_ARRAY(16, uint8_t, temp3, 8 * 4);  // compound pred buffer

   const int16_t *hfilter, *vfilter;

@@ -917,7 +917,7 @@

                                          const uint8_t *dst_ptr,

                                          int dst_pixels_per_line,

                                          unsigned int *sse) {

-  uint16_t fdata3[5 * 8];  // Temp data bufffer used in filtering

+  uint16_t fdata3[5 * 8];  // Temp data buffer used in filtering

   // FIXME(jingning,rbultje): this temp2 buffer probably doesn't need to be

   // of this big? same issue appears in all other block size settings.

   uint8_t temp2[20 * 16];

@@ -941,7 +941,7 @@

                                              int dst_pixels_per_line,

                                              unsigned int *sse,

                                              const uint8_t *second_pred) {

-  uint16_t fdata3[5 * 8];  // Temp data bufffer used in filtering

+  uint16_t fdata3[5 * 8];  // Temp data buffer used in filtering

   uint8_t temp2[20 * 16];

   DECLARE_ALIGNED_ARRAY(16, uint8_t, temp3, 4 * 8);  // compound pred buffer

   const int16_t *hfilter, *vfilter;

--

⑨