shithub: libvpx

Info  •  Files  •  Log  •  Branches

Two optimizations:

1. Reduced the size memset based on eob for 32x32 transform. The reset
of non-zero coefficient should probably go into where they are read in
inverse transform functions. (TODO)
2. Removed a redundant level of indirection.
vp9_iht4x4_add() checks transform type and call vp9_iht4x4_16_add()
for tranforms other than DCT_DCT. In this case, the DCT_DCT case
has been already handled here.

Change-Id: Iacbc77da761f0b308df5acea0f20c9add9f33d20

--- a/vp9/decoder/vp9_decodframe.c

+++ b/vp9/decoder/vp9_decodframe.c

@@ -262,7 +262,7 @@

         if (tx_type == DCT_DCT)

           xd->itxm_add(qcoeff, dst, stride, eob);

         else

-          vp9_iht4x4_add(tx_type, qcoeff, dst, stride, eob);

+          vp9_iht4x4_16_add(qcoeff, dst, stride, tx_type);

         break;

       case TX_8X8:

         tx_type = get_tx_type_8x8(pd->plane_type, xd);

@@ -285,6 +285,8 @@

     } else {

       if (tx_type == DCT_DCT && tx_size <= TX_16X16 && eob <= 10)

         vpx_memset(qcoeff, 0, 4 * (4 << tx_size) * sizeof(qcoeff[0]));

+      else if (tx_size == TX_32X32 && eob <= 34)

+        vpx_memset(qcoeff, 0, 256 * sizeof(qcoeff[0]));

       else

         vpx_memset(qcoeff, 0, (16 << (tx_size << 1)) * sizeof(qcoeff[0]));

     }

-- 

⑨