Optimize 8x8 idct function

Wrote sse2 functions of vp9_short_idct8x8 and vp9_short_idct10_8x8.
Compared to c version, the sse2 version is 2X faster. The decoder
test didn't show noticeable gain since 8x8 idct doesn't take much
of decoding time (less than 1% in my test).

Change-Id: I56313e18cd481700b3b52c4eda5ca204ca6365f3

1 files changed, 2 insertions, 2 deletions

diff --git a/vp9/decoder/vp9_dequantize.c b/vp9/decoder/vp9_dequantize.c
index cb4601a15..2b8c61f1e 100644
--- a/vp9/decoder/vp9_dequantize.c
+++ b/vp9/decoder/vp9_dequantize.c
@@ -229,7 +229,7 @@ void vp9_dequant_idct_add_8x8_c(int16_t *input, const int16_t *dq,
     input[17] *= dq[1];
     input[24] *= dq[1];
 
-    vp9_short_idct10_8x8_c(input, output, 16);
+    vp9_short_idct10_8x8(input, output, 16);
 
     input[0] = input[1] = input[2] = input[3] = 0;
     input[8] = input[9] = input[10] = 0;
@@ -245,7 +245,7 @@ void vp9_dequant_idct_add_8x8_c(int16_t *input, const int16_t *dq,
       input[i] *= dq[1];
 
     // the idct halves ( >> 1) the pitch
-    vp9_short_idct8x8_c(input, output, 8 << 1);
+    vp9_short_idct8x8(input, output, 8 << 1);
     vpx_memset(input, 0, 128);
     vp9_add_residual_8x8(output, pred, pitch, dest, stride);
   }